Professional Documents
Culture Documents
ALEXANDER J. DE VOOGT
Editorial board: M. Forrer; K. Jongeling; R. Kruk; G. J. van Loon; W. van der Molen; F. E.
Tjon Sie Fat (chief-editor); W. J. Vogelsang; W. van Zanten.
All correspondence should be addressed to: Dr. F. E. Tjon Sie Fat, chief-editor CNWS
Publications, c/o Research School CNWS, Leiden University, PO Box 9515, 2300 RA
Leiden, The Netherlands.
Copyright reserved. Subject to the exceptions provided for by law, no part of this publication may be
reproduced and/or published in print, by photocopying, on microfilm or in any other way without the
written consent of the copyright-holder(s); the same applies to whole or partial adaptations. The
publisher retains the sole right to collect from third parties fees in respect of copying and/or take legal
or other action for this purpose.
To the memory of my grandmother
ACKNOWLEDGEMENTS
In 1990 during a Swahili language course at the Institute of Kiswahili and Foreign
Languages, first contact was made with a group of Bao players of the Kidongo
Chekundu Bao club in Ja'ngombe, Zanzibar. During subsequent travels, this contact
was extended by meetings with other clubs, the local Bao Society, and in particular
by meetings with the masters of Bao. A first preliminary study was conducted in
1992, followed by two intensive research periods in the months of May and October
1994. Only the research of May 1994 received official financial support, for which I
am very grateful to the Society for the Advancement of Research in the Tropics
(Treub-Mij.).
Gratitude is also extended to Mr. H.M. Mshindo, Mr. Ali H. Haji, and their
Ministry of Culture and Tourism for their cooperation in obtaining a research permit,
Mr. Mgeni Mwalim Ali and the Sports and Games Department for their kind
permission to organize Bao championships, Mr. Ali Khamis Ali, Mr. Abdallah A.
Muombwa, and the other teachers and researchers of the Institute of Kiswahili and
Foreign Languages who have supported and assisted me during each visit in the past
few years.
I am especially grateful to Rosalyn Walker, Laurence Russ, Philip Townshend,
and Vernon Eagle for their insights in manqala games, their support during this
research and for making their publications available to me. Also, I would like to
thank the members of the Dutch Bao Society for their enthusiasm and for keeping up
practice in Bao.
Without the help and friendship of the various masters of Bao this work could not
have been written. Firstly, my principal informant and friend Abdulrahim Muhiddin
Foum, who was not only an extraordinary informant, but who also performed the
task of guide through the outskirts of Zanzibar Town and beyond. Secondly, the
masters: Majaliwa J. Seif, Ibrahim S. Mkiwa, Masoud H. Ali (Kijumbe), Shamte M.
Shamte, Nasoro 0. Ali, Rajab 0. Rajab, Masoud Kh. Masoud, and Juma A.
Njowine, who all made great efforts to perform well on experiments and who made
this research successful. Together with Ali M. Hussein, Nuhu Ali, Khamis A.
Khamis, Mbaya 0. Juma, Mwinjuma H. Mabruki, Mgeni A. Omar, and Ameir Sh.
Kumba (Kasanga Turnbo), I thank them for their insights and support.
I would also like to thank Gogola S. Gogola for his friendship and together with
Ali J. Ali (Kipara) for his guidance in DarEs Salaam. Also, Mr. Abbas A. Mdungi
for his guidance in the direction of Tumbatu, and Mr. Kondo Pandu who generously
made the documents of Shaame Kondo Khamis (Nahodha) available to me.
For their correction of the text, I gratefully acknowledge among others my father,
Jean Marc Blankert, Lucien Reurich, Constance Dickmeyer, and Abdulrahim Foum.
Finally, I would like to thank my grandmother and my father who each made
valuable travel companions in Zanzibar, and my mother whose advice and support
made this research such a pleasant one.
TABLE OF CONTENTS:
Introduction 1
I
r B Strategy in Bao 110
1 Methods of research 110
2 Structure of Bao strategy 110
3 Fani 112
c Generalising strategic theory 115
1 Opening play 115
2 The masters tournament 1994 118
D Conclusion 118
IX Conclusions 121
1. Practice 121
2. Limits of the mind 121
3. Dynamic constraint 122
4. Summary 122
Glossary 124
Literature 127
Appendix I: Experiments on Bao 131
Appendix II: Tournament games 135
Appendix III: Time management possible moves 158
Subject index 159
Author index 163
Summary 164
Samenvatting 167
Curriculum Vitae 170
LIST OF TABLES:
LIST OF FIGURES:
1
II
J
I
l
-I
l
I: INTRODUCTION
Masters of an East African board game called Baa appear to be able to calculate
extraordinarily complex moves, easy for a computer to emulate but very difficult for
humans. Instead of finding a limitation to the present state of calculating abilities of
computers, Baa shows a limitation to the human mind never incorporated in
computer simulations of human thinking.
Board games are invented only by humans. The study of such games should tell
us about those humans. Board games have been studied in many disciplines.
Psychologists have studied players, archaeologists have studied game boards and
game pieces, and computer scientists have written programs to play games.
Huizinga (1938) made the study of games the study of Man, with Homo Ludens. The
masters are the best players. Mastership should allow us to identify those
characteristics which define the limitations of human playing skills.
The aim of this study is to determine what is particular to mastership in Baa.
What makes a Baa master? Can such mastership be defined in terms of limitations
to the mind? The answer will lie in the mental skills used to calculate Baa moves. In
order to understand these characteristics and their relevance to Baa it will be
necessary:
i. to understand the historical and geographical context within which the game is
played;
ii. to document and structure the rules of the game;
m. to collect sufficient material to document Baa masters at play, illustrating where
human mental limitations constrain the way masters play;
iv. to assess empirically the limitations of human calculating and problem-solving
skills appropriate to Baa.
We find games research in anthropology (e.g., Townshend, 1986), studying the roles
that games play in various cultures. Psychologists use games to study people,
studying players from many perspectives. Such different approaches can be divided
into the ethnological, the pedagogical and the use of Artificial Intelligence (see
Reysset, 1995). In the first, scholars use games as a symbol in ethnological theory.
Games are seen as a fundamental biological necessity to accelerate the organic
development of infants, or a mechanism supporting symbols in the fantasy of the
child or the adult (for a summary see Reysset, 1995:98). The pedagogical and
psycho-therapeutic interests result from the assumption that games reflect the
personality and conscience of people. This leads to social developmental studies of
children through using games.
Retschitzky (1990) studied African children's development as measured by their
skill in Awe!e, a West-African game also known as Awari. This game is also used
in Artificial Intelligence (Allis, 1994). In one approach to Artificial Intelligence,
efforts are made, often based upon games-playing, to simulate human thinking by
way of computers. Within this approach Chess has dominated the literature and
revived scholarly interest in board games (e.g., van den Herik, 1983; Levy & Beal,
1991). Every board game has different characteristics for computer scientists. The
study of games other than Chess has provided this field with problems of computing
a winning performance on games like Go (Allis, van den Herik & Herschberg,
2 LIMITS OF THE MIND
1991), with moves that appear almost impossible for computers to calculate but not
for humans.
Bao is a board game, related to Awari, that does not pose considerable problems
for computer scientists. Although Bao rules are complex, Chess has many more
possibilities. Bao itself has provided an example for anthropological studies of
symbolism in East-African culture (Townshend, 1977; 1979; 1986). Bao in
Zanzibar, however, is interesting for its masters who demonstrate extraordinary
abilities in playing the game and, thereby, provide insights into problem-solving and
memory techniques of experts.
The study of masters first became prominent with de Groot's thesis (1946) on
the thinking of masters. Since Simon and Newell (1972), the study of masters has
received much attention. Their studies became interesting for cognitive scientists
who could simulate problem solving techniques in computers. However, studies of
masters in other board games, especially those not influenced by Chess thinking,
remained absent.
A study of Bao masters will have to deal with many perspectives and consists of
intensive interdisciplinary research. The study of the Zanzibar culture and the
Swahili language are obvious prerequisites. However, experimental psychology and
elements of computer science are also central to this thesis. Research on experts in
general, on strategy in games, on African culture, on Chess masters, on blind play,
and even on warfare are all involved for a documented study of this kind.
The following assertions, covered in different Chapters, represents the body of the
argument to be developed in this thesis. The two major underlying assumptions, to
be tested, are that:
2. Although masters have several skills beyond a certain minimum, I assert that a
characterisation of a Bao master is an individualised array of limited skills. Bao
mastership is then, in the first place, a combination of special abilities in playing
Bao, different for each master. This is contrary to de Groot's belief (1946:10) that
differences in mastership should be based entirely on the ability to acquire
experience, thereby dismissing special, possibly inborn, abilities. Although I
consider the special abilities of Bao masters not to be innate, the necessity of special
abilities is considered to be central in characterising Bao mastership.
Firstly, in Chapter II, it is argued that the backgrounds of masters do not explain
their performances. The playing circumstances in Zanzibar .will show that neither
differences in experience and knowledge (de Groot, 1946), nor in practice (Simon,
INTRODUCTION 3
1979:426) suffice to explain the differences in Bao skills between master and novice
or even between master and master.
Chapter III describes the classification and the rules that determine the thinking
space within which the limits of the master's mind matter. De Groot, Simon, and
others have based their findings on studies of Chess masters. This Chapter argues
that the key difference between Chess and Bao is in the nature of the moves. The
nature of the move also illustrates the key difference between Chess skill and Bao
skill, which is calculating a move - in the case of Bao this is calculating duru.
Chapter IV documents the specific rules of Zanzibar Bao, identifying where the
limits of the mind can be established. The chapter concludes with a quantitative
characterisation of Zanzibar Bao and introducing the concept of mutational
complexity.
Chapter V illustrates combinations in and outside the master's ability to
calculate.
Chapter VI shows that calculating duru appears to provide evidence for the key
difference between Chess memory skills and Bao memory skills. Bao memory skills
introduce an additional limitation to human memory, a dynamic constraint, which
has not been identified in skills for Chess or other games.
Chapter VII continues the experiments, now applied to understanding the skill of
playing Bao blind.
Chapter VIII introduces a number of skills which are essential to masters, in
particular to masters of Bao. Subsequently, the absence of skills in certain masters
identifies qualitative limitations to those master's skills in Bao. It discusses Bao
mastership broadly, taking into account the competitive aspects and reporting the
distinctions in mastership made by the masters themselves.
Chapter IX concludes, summarising certain limits of the mind and, thereby,
characterising Bao mastership. Such a characterisation in terms of limitations
appears to break new ground in the study of masters. Memory and problem solving
skills are more varied among Bao masters than literature leads us to expect for Chess
masters. A description of mastership in Bao, therefore, makes a contribution to
psychological studies on problem-solving and memory.
II: BAO: THE GAME AND ITS MASTERS
The aim of this chapter is to provide the game of Bao and its masters with a historical
and geographical context. It provides the state of affairs in championship Bao in
Zanzibar and discusses the history and some of the backgrounds of Bao
(grand)masters over the last 100 years. Three research questions will be investigated:
Firstly, although research into the backgrounds of Chess masters (de Groot,
1946) and various other experts (e.g., Ochse, 1990; Chi, Glaser & Parr, 1988;
Simonton, 1984) have produced little evidence to the contrary, I will have to confirm
that expertise in Bao, and the possession of other (memory) skills, have little to do
with conventional education, social life, general intelligence, or success. To some
people, Bao appears as a purely mathematical game at which only mathematically
skilled people should excel. In Zanzibar itself, there is a widespread view that Baa-
players are idlers and that the game disturbs your social life, as if it was a dangerous
addiction. The question can be posed whether excellence in Bao makes you an
intelligent person or whether one needs to be intelligent in order to become a Bao
expert. All these biases and related questions are briefly discussed with the help of
the data collected.
Secondly, the age at which Bao is learnt and the age at which players reach
mastership can give us insight into the development of mastership and might confirm
the early findings of de Groot (1946), that Chess masters usually achieve mastership
in their twenties.
Thirdly, what is the relative significance of the number of informants used in this
study? This chapter will relate the number of informants used in psychological
experiments to the total number of present masters and to the total number of
(grand)masters in the last 100 years. This Chapter gives an inventory of the masters
over the last 100 years, together with details of all extant Bao masters. The
limitations of the data are discussed at length, so that the historical value of the
material can be determined. The general background of the participants in the
championship, my informants, and my assistants, can be given comprehensively at
the same time.
This Chapter presents a historical discussion of Zanzibar, the Bao Society, the
1994 Masters Tournament, and Bao champions, followed by a detailed discussion of
Table 1, that lists all the current masters. From this table, presented at the end of this
Chapter, the names listed in the second column will be used when referring to a
particular master.
A. The game
Bao is played between two people. In Zanzibar, most players use a board which is
placed on the floor, one player on each long side. On the board there are four rows
of eight holes, two holes enlarged, usually square, in the middle of the board.
Together with the board there are 64 playing seeds. In the field of computer science
a similar game, Awari (Allis, 1994), has been characterised as a zero-sum, two-
person game with perfect information (Von Neumann & Morgenstern, 1944).
BAO: THE GAME AND ITS MASTERS 5
Each player owns two rows of eight holes, i.e., the ones on his side, and 32 seeds.
The object of the game is to capture all the seeds of the opponent's front row, or to
disable him so that he cannot play any further. Moves are played by spreading seeds,
i.e. placing seeds one by one in adjacent consecutive holes as is common to manqala
games.
The square hole starts with six seeds right from the beginning. Sometimes this
'house' fills up quickly and is constantly attacked and defended. The player who
owns such a full house will try to empty its contents most advantageously and make
many captures in one move by spreading all the seeds in the house.
stock A
22
2 2 6
6 2 2 stock a
22
TYtagram 1...
......
In the first stage of the game, namua, only a few seeds are in play and with each turn
the player adds one seed from his stock to a hole on the board. He then starts
spreading, i.e. placing one seed at a time in consecutive holes, seeds according to the
rules of the game (see Chapter IV). He either captures, which is obligatory if
possible, taking seeds from the opponent's front row and putting them into his own
front row, or he spreads seeds in his own rows starting from the hole of his choice.
Imagine the following situation:
1 1 1 1 1 1 stock A
13
1 1 2 3
1 9 3 5 stock a
1 1 1 1 2 14
Diagram 2
One has to capture if two neighbouring holes on the inner rows contain seeds. If (a)
has the turn, he could either capture two seeds, by entering a seed on his row (into his
hole of three seeds), or three seeds by entering a seed into his hole of five seeds. The
captured seeds are spread one by one, starting from the far left or the far right, into
one's own front row. If the last seed spread in this way reaches a hole that contains
seeds, its opposite hole will also be captured and spread into the front row, this time
in the same direction. If one captures from the two far-end holes on either side (in
this case the three seeds of (A) opposite the five seeds of (a)), it is obligatory to enter
the captured seeds from that end into one's front row. With these rules complicated
captures can be made. Note that the back row, although containing seeds, cannot be
BAO: THE GAME AND ITS MASTERS 7
captured. (For the explanation of complicated moves see Chapter V, and the section
on complex duru in Chapter VII.)
Once all the seeds are entered into the game, the second stage or mtaji stage has
begun. Instead of entering a seed, the player now lifts the contents of a particular
hole, either capturing or spreading round if he cannot capture, adhering to practically
the same rules as in the first stage. Consider the following example:
1 4 6 5 1 2 2 1 stock A
0
1 1 2 1
3 5 2 14 1 1 stock a
3 2 1 1 3 1 0
Diagram 3
If it was (A)'s tum, there would be two ways of capturing, i.e. lifting the five seeds in
the back row (which can capture three seeds ), or lifting the two seeds on the front
row. Singletons cannot be moved to capture. If (a) had the tum, the fourteen seeds,
if spread anti-clockwise, would make a capture. Also, the hole with two seeds in the
front row could capture, in either direction. There are no holes that (a) can lift from
the back row. Note that the occupied hole has to be reached directly, i.e. in one
lifting.
If all the seeds in your front row are taken, the game is lost. The game is also lost
when in the mtaji stage you only have singletons left. Play singletons is not allowed,
your inability to play means instant defeat.
A game between masters usually takes less than sixty turns, played in an average
of twenty minutes. The players play relatively rapidly, thinking times rarely exceed
five minutes. Often a combination of captures is made in one turn, and seeds can
make complete laps in the two rows of one player. A skilled player will play such a
complicated move with shortcuts. Instead of placing each seed in the appropriate
hole, adhering to the rules, players pick up a number of seeds and throw them in
adjacent holes, three, four, sometimes five at a time. The end situation is thrown
down without showing how it came about in consecutive spreadings of seeds. Such
shortcuts make the game complicated to the layman, who has difficulty deriving the
rules from what he sees.
After one game of Bao, the players usually insist on completing a set of three.
Three wins to nothing makes a serious beating, this is sometimes topped with sets of
six, nine, or even more. The loser is called tobwe, the winner bingwa. This thesis is
concerned with the bingwa.
B. History of Bao
Bao is one of a group of manqala games (see Chapter III) which are played in large
parts of Africa, Asia, the Caribbean, South-America, and some parts of Europe
(Herskovits, 1932; Murray, 1952; Bell, 1960; Russ, 1962; Deledicq & Popova,
1977). Four-row games, like Bao, are found in East-Africa, Sri Lanka, and have
recently been discovered in China (Eagle, 1995). Bao, as it is described in this
thesis, is only played in East-Africa, in the Swahili speaking areas. Since manqala
8 LIMITS OF THE MIND
games in large parts of Africa and Asia are still remain to be described I will refrain
from speculating on the origin of four row manqala games, or manqala games in
general. Systematic research on manqala games played in China and neighbouring
countries is now starting (Eagle, 1995) but there is still insufficient data for a general
understanding of the historical development and spread of manqala games.
1. History of Zanzibar
The research on the game of Bao has been conducted in Zanzibar. This island is part
of a group of islands off the coast of Tanzania. Together with Pemba it provides the
basis for the continuing Zanzibar Bao championships. Zanzibar has more than
200.000 inhabitants of which more than 70.000 live in the capital Zanzibar (or Stone
Town) itself. Zanzibar has 1658 km 2 of land and is made of coral like Pemba.
Pemba is a smaller island of 984 km2 with less than 200.000 inhabitants.
Zanzibar has a long colonial history. During the larger part of the Middle Ages it
was a small Arab empire, becoming Portuguese in 1503. Then, in 1784, it was
conquered by the Imam of Muscat. This period brought many Arabs to the island and
gave the architecture in the capital a very Arab character. In 1880, Zanzibar became
a German protectorate which was given to the British in 1890. Its long history of
slave trade, being the centre of slave distribution in East Africa, ended with the
British abolition of slavery in 1897 and again, and more completely, in 1907. The
commercial success of this thriving business has been only partly replaced by
plantations exporting cloves all over the world. (See also Ingrams, 1931)
In 1963, Zanzibar became an independent Sultanate. However, the Arab
supremacy was countered by a black revolution in 1964, which put the Afro-Shirazi
Party (ASP) in power. The party changed the Sultanate into a Republic and
nationalised the land. Shortly afterward, the islands of Zanzibar and Pemba together
with Tanganyika became part of an independent Tanzania. The leader of the ASP,
Abeid Karume, became vice-president of Tanzania. After his assassination in 1972,
he was replaced by Aboud Jumbe.
At the time of this research, the political situation in Zanzibar and Tanzania
mainland was stable. The 'socialist' government has produced a large and dominant
public service in which Zanzibar has played a somewhat independent role. Research
visas are different for Zanzibar, and there are separate customs and health officials
checking the incoming passengers from the Tanzanian mainland or, nowadays,
directly from abroad. ·
The scheduled election of 1995, the first free elections in Tanzania, continued the
strife between the old Arab and more recent Shirazi factions, even though racial
differences are blended into one Zanzibar people.
Kiswahili , or Swahili, is the official language of Tanzania. The Zanzibar dialect
was chosen as the official Swahili dialect for all Tanzania. Islam is the dominant
religion in Zanzibar and Arabic is, therefore, a second language to many Zanzibari.
Fortunately, Arab and Indian migrants have only enriched the vivid East African
culture in Zanzibar. Today, Zanzibar's architecture, climate, and history form the
basis for increasing tourism on this African tropical island.
Kikwajuni, which is one of the main clubs today. Ibrahim Mkiwa presided over the
first years and took up the chairman's position again in 1991.
During all these years, Mwalim Amur was secretary of the Chama. He collected
various details about players and clubs and still has a document of about 150 pages
with all his research written down in Swahili. In 1990, possibly for political reasons,
he was replaced by Fadhil Mwinyi Mohammed, who is not very experienced in the
Bao circuit.
In the last years, the society has lost the participation of the mainland. Amur was
secretary of the club while it still covered the whole of East Africa. Then the players
travelled to exchange players and hold championships. Today, the club organises
very few trips to Dar Es Salaam and its members are only located on Zanzibar
(Unguja) and Pemba.
There are about sixteen official playing grounds. These playing grounds pay a
yearly fee and also delegate one referee each to the championships and/or matches.
Apart from the official playing grounds, there are unofficial groups of players who
own a few boards but who do not send referees to the championships.
1. Methods of research
All the data presented in Table l.b were taken from personal interviews with the
masters in question. There are two exceptions: Nahodha and Abdi (I will refer to the
masters using their short names that can be found in the second column of Table l.b).
The latter was working on Pemba at all times during my research. Subsequently, his
data were given to me by his half-brother, Masoud. Information about the famous
Nahodha, who died in the early 1980s, was also collected with help from his family.
He is the only player on the list who is no longer alive.
Interviews were conducted in two ways. About eight players were interviewed at
length on video. All other players were interviewed in short sessions and were
specifically asked about the details presented in the list. Some facts, especially about
the older masters, were cross-checked to eliminate accidental mix-ups in people's
memory.
All interviews on video were conducted in Swahili by Abdu. On occasion,
Shamte or Mwinyi replaced him for short sessions, and in Dar Es Salaam both
Kipara, Shamte, and Abdu were present. I can state that almost every interview was
conducted in the presence of at least one Zanzibari. This eliminated any possible
misunderstanding because of language.
10 LIMITS OF THE MIND
Data about masters no longer alive were collected from their fellow masters.
Mkiwa was the only informant who recalled a large number of masters playing
before the war. Kijumbe and Mkiwa provided information about the post-war
masters. Nuhu was too old to remember anything, but he clearly responded when
given a famous name. All the masters mentioned by these people deserve the kind of
attention I gave to Nahodha's story. However, this special and difficult task was
beyond the scope of my research and was not pursued any further.
Nahodha was remembered by almost all the masters on the list, except for the
two youngest ones. However, his full name and his date of birth could not be
collected from his fellow players and only partly from his family. Only when we had
reached his uncle- after nearly reaching his place of birth, Tumbatu- who presented
us something similar to a passport, were we able to complete the details about his
life.
Some older players, who were not able to tell us their year of birth, were also not
able to provide us with any records. In most cases these records would have been
incorrect anyway since their year of birth was simply not registered.
competition became equally strong. From this time we have at least two sources,
Mkiwa and Kijumbe, Nuhu being too old by the time of this research.
First, Sengeru became well-known. Originally from Tanga, but raised and trained
in Zanzibar Town at a club that is no longer existent. Sengeru died in the late 1980s. I
Then the players from Kwahajitumbo: Kijumbe, Mohammed Issa, Twaha Saidi,
Kasim Mwinyi, and Khamis Bakari all played behind the fish market. The last three
I
have died, Twaha became ill, Kasim, the fisherman, was lost at sea, and Khamis, I
about the age of Kijumbe, died as well. In their time, they were joined by Nahodha,
who became the most famous of them all. Together with Mkiwa, he was one of the
few who travelled to the mainland regularly and received some national fame. As far I'
as we know, Kijumbe, Nuhu and the others travelled to play in Pemba, and perhaps
once in Tanga, but their names are not well known in Dar Es Salaam.
After Tanzanian independence in 1961, the Bao players of Tanzania were
organised in a Society. Mkiwa became chairman, and Mwalimu Amur (a minor
player), secretary. The subsidies now available made it possible to build nice
clubhouses. Such a kikao usually had a roof with a cement floor, and sometimes, like
in Kikwajuni, a storage room. Other chairmen followed, like Mzee Jai and Mzee
Masqat, but Mkiwa returned recently to his former position. Only four years ago, the
secretary was replaced and his job went to Fadhil Mwinyi Mohammed (also a below
average player).
It was after a dispute about travel funds that the Kikwajuni Bao club, where
Mkiwa held office as chairman of the Bao society, and the Kwahajitumbo Bao club
refused to play each other. More accurately, the players from Kwahajitumbo refused
to play Mkiwa. Only Nahodha and later Abdi played both clubs, not too much
disturbed by the fight. Travel after independence did not increase and government
funded trips to Dar Es Salaam and back were hindered by incompetent leadership
and money handling on the part of the Bao Society. In recent years, the funds for
championships have also disappeared.
The high concentration of grandmasters at the Kwahajitumbo club resulted in a
new generation of masters trained at this location. Mwinyi, Dr Mgeni, Abdi, and to a
lesser extent Masoud became well-known players of the 1970s.
In this account we miss the names of Nasoro and Zerkani. They became well-
regarded grandmasters in the 1970s and before. Zerkani blended in with the others,
he was just not from such a well-known club. Nasoro, who was born on the
mainland, travelled to play in various championships. After independence, he
attended an international Bao championship where players from Ethiopia, Zambia,
Kenya, and other countries attended. They were all beaten by the Tanzanian players.
Nasoro taught both Shamte and later Abdu (not be confused with Abdi), when
they played at Kidongo Chekundu. This former Bao club, destroyed by rain in 1993,
was close to their parental homes. Shortly afterwards, Shamte changed to the
Mikunguni club. At that time, the Kwahajitumbo players were regular visitors in
Mikunguni. There was Shamte, Majaliwa, who visited from another nearby club and
also moved to Mikunguni, and Rajab, who originally came from Wengiwape A,
close to the Mikunguni Trade School, also very close by. By the end of the 1970's,
Kijumbe, Mkiwa, Nuhu, and the other older players started to drop out. Nahodha
became the leading player with much competition from Dr Mgeni, Rajab, and Abdi.
But in his mid-forties, Nahodha took ill, returned to Tumbatu and died around 1984.
Kijumbe had left the game altogether by then, and Rajab would soon quit too in
order to study at the University of Dar Es Salaam. Nuhu ~became too old, and Dr
Mgeni left because of the leadership in the Society that would still not produce good
12 LIMITS OF THE MIND
for Oman at the time of my arrival. His replacement had no news of the tournament.
Mzee Muombwa had left for India, having suddenly received a scholarship for three
years. The director of his Institute had no news of the tournament and was fully
booked for the coming weeks. Only Shamte and Abdu had been able to complete
their tasks for the preparation of the tournament.
On Friday (7/1 0), the acting director of Sports and Games promised to talk to the
ministry. The letters sent to the department had been lost, the Bao file had no new
letters since 1989. Unfortunately, the minister of Culture had left for Pemba,
eventually the acting director himself left for Dar Es Salaam. In the end, no funds
were made available to the organisation of the tournament. Players from Dar Es
Salaam, Pemba, and Matemwe had to remain absent. The director himself regretted
the unfortunate situation that had arisen and generously put another promise in
writing that should assist the organisation in a new tournament on 1/7/1995.
Monday (10/10), the Institute, after first requesting some payment, agreed to
allow us to use a class room for the first meeting of players. Two players from
Matemwe showed up, but had to return home without entering the league since there
were not enough funds for travel, food, and housing. Ten players from Zanzibar
Town were present who all agreed to enter the league of the tournament and play for
a full two or possibly three weeks.
Tuesday (11/10), we agreed to play at Shamte's Bao club, Mikunguni. However,
the secretary of that Bao club refused to assist. He objected that decisions had been
made in his absence. After some confusion he gave in and the championship could
start. Each player was to play two games with every other player, in total about 55
games.
Few other problems arose. One player left the league prematurely, and in his
stead Rajab agreed to play despite his many years of absence. Half-way through the
tournament Majaliwa left the championship because he lost interest - he had just
become an elected party member for the new Civic United Front -party.
On 22/10, the games of the league had all been played. Since there were only 10
players, we left out the quarter finals. The next Sunday, we started the semi-finals
with the four best scoring players of the league. It now became a knock-out system,
but they had to win two games in a row in order to enter the finals. Shamte played
Nasoro and won two games in a row the very first day. On 26/10, after many draws,
Maulidi and Kijumbe finally ended their match. Kijumbe was exhausted, Maulidi
won two games in a row. Shamte and Maulidi agreed to continue with the finals the
next day. They also played several draws. On 30/10, we decided to change the rules
and instead of winning three games in a row, the first to win three games would win
the finals. This change was not necessary, since on that day Maulidi won three
games in a row and was pronounced the winner.
The rules for the championship were taken from previous championships.
Winning three games in a row is the accepted way of defeating your opponent and
considered to be a difficult task. The idea of a league was also put forth by the
players. Together with the knock-out (semi-)finals they made for an interesting
tournament.
As in other sports, the finals in the tournament do not necessarily present the best
games. Nerves, exhaustion, etc. played an important role in both the league and the
(semi-)finals. Most players were very tired after two weeks of solid play.
14 LIMITS OF THE MIND
They started the tournament with a large percentage of draws which gradually
changed into series of 2-0. All were agreed that only this kind of championship
would bring the standard of play back to a higher level. Indeed, some games were
already showing signs of excellent play.
The results and transcription of the games played in this tournament are listed
towards the end of this manuscript.
D. Players' background
2. Year of birth
All dates, with the exception of Nahodha and Abdi, were taken from personal
interviews. This does not guarantee the accuracy of the data. If people did not
hesitate to tell me either their year of birth or their present age, I judged the
information to be accurate enough.
Two people were not able to answer the question: Nuhu and Kijumbe. They also
did not have any papers or records to show me. Even with reference to certain
historical events, conversations with family, and interviewing other masters about
being younger or older, things did not become much clearer. Answers appeared very
contradictory. The experience with Nahodha's history had learnt that estimating age
or year of birth is very difficult for friends or even relatives. Therefore, I placed them
in relative order to the best of my knowledge. Nuhu being certainly older than
Mkiwa, with Kijumbe certainly younger.
One of the youngest masters on the list appeared to be one year younger when his
answer was checked with his records at the hospital. Apparently, old age has very
little to do with inaccuracy. Birthdays are usually not celebrated, and passports are
16 LIMITS OF THE MIND
expensive and hard to obtain, which makes remembering or checking dates of birth
not an easy task for both player and researcher.
behind the fish market. Ulaya Jamhuri is said to be an official club, but I was not
able to verify this nor to locate its position on the map.
Donge died as a club as well and is not registered anymore. The clubs up north
now being Chaani (former Gamba) and Matemwe. Other clubs can be found inland
but many are not registered with the Society.
Tandika is not a club. The research on Bao clubs was restricted to Zanzibar.
Tandika is just a neighbourhood where Kasanga Turnbo and Njowine happened to be
interviewed near a club called 'Mikogo'. Moreover, there is no official organisation
left in DarEs Salaam that would justify the label 'official'.
6. Educational levels
A master's place and date of birth are very much linked to his educational
background. Those who were raised outside Zanzibar Town usually lived in fishing
or farming communities where school education was either absent or not considered
very important. Secondary schools were present on Zanzibar from the early 1940's
onwards. However, not many people from inland Zanzibar are likely to have
attended such schools. The oldest players listed received a number of years of Koran
teaching in their home towns. As far as I know, none of them know how to write the
modern Swahili alphabet. They do know how to write their names in Arabic.
The people from inland, including Njowine and Kasanga Turnbo who were also
born in farming communities, all have a below average education. That is if we
believe the Department of Education of the Government of Zanzibar (1994/1995)
that states that form 2 or 3 (like that of Maulidi) is average, or at least compulsory.
One generation of players has exceptional high educational backgrounds, those
playing at Kwahajitumbo in the early 1970's. Form 4 (Majaliwa) is reached by
hardly 15% of the children finishing form 3, and form 6 (Shamte, Abdi) by about
15% of those reaching form 4. There is no university on Zanzibar, therefore a
Bachelor's degree (Rajab) is not only a very difficult but also a very expensive
undertaking.
Njowine, who is a player of the same generation, makes a useful exception to this
group. Although I adhere to the belief that Bao masters have a higher level of general
intelligence, this certainly cannot be proven with such a small, even though almost
exhaustive, list of players. Intelligence tests will fail if people cannot write, have had
hardly any formal education, and if intelligence tests are not made by people who
understand Zanzibar society. And since intelligence itself is for the most part
defined by the tests that were developed to test it, the question about Bao masters'
intelligence remains unanswered. All that can be stated here is that Bao masters are
not idlers, who fail school, and who have no impressive intelligence. They are
players with many different kinds of educational backgrounds, largely related to their
place and date of birth.
A few remarks should be made about the data presented here. 'Prof.dipl.' means
professional diploma and in the cases presented in the table the content of that
diploma appears as their profession. 'Adult educ.' is adult education and this assisted
Nasoro and Kasanga Turnbo in learning how to read and write the Roman alphabet.
Koran or koranschool is usually a few years and consists of memorisation and often
some basic understanding of Arabic. Some masters attended many years of
koranschool, this is, however, not shown in the table.
18 LIMITS OF THE MIND
7. Profession
Statistics about educational background measure the relative success of people in
school. If Baa-masters are successful people in general, i.e. if they show a relatively
high intelligence in their lives, then some generalisation might still be possible.
A profession appeared to be one of the most commonly known aspects of
masters. Even professions of masters whose full names were forgotten were readily
identified. Twaha Saidi, Mohammed Issa, Kasim Mwinyi, Khamis Bakari, and
Sengeru had the following professions: porter, tailor, fisherman, fruit salesman, and
rent collector, respectively. None of their jobs were highly profitable.
There is one complication, which is that people's professions change in the
course of years. Their profession of the past might show whether jobs influence the
amount of time spent on the game, and their profession at present might show up in
their relative successes. After careful investigation it is concluded that no
generalisations are possible, i.e. professions neither influence nor indicate the
possibilities of becoming a master. Some are successful, and some never were.
There are a few possible exceptions. Nuhu was a clove-gatherer and this
profession was apparently profitable enough for him to spend large amounts of time
in the city in order to play Bao without attending to his work. Abdi has to spend
most of his time on Pemba as a customs official, which sometimes interferes with his
playing. He has not attended championships since 1989. However in 1994 he started
practising again, but was stuck in Pemba at the time of the October championship.
Many jobs are not profitable at all, like cleaning the market place, masonry, or
fishing. A few masters are well to do or have good jobs because of their education,
but they usually become very busy and have difficulty finding time to play. Since
most players reach mastership during or just after school, jobs have· very little
influence at the developmental stages. At a later age, very few claimed to have left
the game because of their job, they simply lost interest because there were no
incentives, i.e. prizes, or good players to compete with.
Some masters changed jobs. Mkiwa sold fish, but became a koranschool-teacher
in 1975. Zerkani held various jobs in governmental offices but is now a fruitseller (at
the Old Slave Market). Nasoro had a coffee-stall, but sold it and accepted a job at the
municipal works, where he now cleans the marketplace. At the time of my research
Majaliwa quit his job as a medical assistant and was elected as a Civic United Front
-party member. Dr Mgeni !s a medical doctor for the Malindi Sports Club now,
usually working late afternoons and, therefore, unable to attend Bao clubs or
championships anymore.
Organising championships brought Rajab and Kijumbe back to the game. Only
Dr Mgeni seems to have lost interest all together.
There is nothing unusual, from a Zanzibar perspective about the marital stages of
Bao masters. Considering the findings in the other columns of Table l.b, this is not
surprising.
E. Literature
Bao players learn Bao without the help of written literature. There is only one
manuscript in Swahili that deals with Bao, but it has never been published and is
more an incomplete history with a list of playing rules than a guide to players.
English literature also limits itself to history (Murray, 1952), and rules (National
Museums, 1971), with an occasional work on its anthropological (Townshend, 1986)
aspects.
Since Philidor's "L' Analyze des Echecs" (Hooper & Whyld, 1992), it is difficult
to play modern Chess at a master's level without studying the literature. It appears
that some results from Chess research have to be reconsidered if we compare them
with the results from Bao research.
Hence, the overridh1g factor in chess skill is practice. The organization of the
master's elaborate repertoire of information takes thousands of hours to build
up, and the same is true of any skilled task (e.g., football, music). That is why
practice is the major independent variable in the acquisition of skill.
Simon (1979:426)
contrary, Chess has so much literature, and therefore access to masters' play, that
mastership can be reached with much less playing experience. Since even the rules
of Bao are learnt through pure playing practice, Bao should be considered a game
where people learn almost everything almost exclusively through playing experience.
This makes the argument that players cannot play more frequently than other players
even stronger, since there is no tradition of playing on your own, or after playing
hours.
Therefore, we have to conclude that practice, while important, is not the major
independent variable in the acquisition of skill.
F. Summary
Full name Known as Year Place Student Master Latest club Education Profession Marital stage,
of Birth of Birth• ofBaa of Baa Number of children
NuhuAli Mz.Nuhu 1900's Donge Donge koranschool clove-gatherer 2nd wife, 3 children
Ibrahim Said Mkiwa Mkiwa 1916 Bara 14 18 Muungano koranschool fishman, teacher 5th wife, 2 children
Masoud Hassan Ali Kijumbe 1920's Shamba +7 Kwahaiitumbo koranschool fishman, farmer 5th wife, ? children
Khamis Amir Khamis Zerkani 1933 Stone Town +9 20's Mkamasini grade 7 + koran civil servant 3rd wife, 3 children
Nasoro Othman Ali Nasoro 1935 Bara 7 18 Boraimani grade 3 + koran municipal works 1x divorced
Mbaya Othman Juma Mbaya 1935 Maternwe 18 +56 Matemwe grade 5 + koran , farmer/fisherman 2 wives, 7 children
Shaame Kondo Khamis Nahodha 1938 Tumbatu ±20 Kwahajitumbo koran+ adult educ. customs official 1x divorced
Ameir Shabaan Kumba Kasanga 1942 Bara 13 23 Tandika grade 2 + adult ed. business, farming 1x divorced, 1 child
Turnbo
Majaliwa Juma Seif Majaliwa 1949 Shamba 12 15 Mikunguni form 1 + prof. dip!. 1 medical assistant 3rd wife, 5 children
Mgeni Abdallah Omar DrMgeni 1950 Pemba 10 24 Kwahajitumbo form 4 + prof. dip!. dentist 1st wife, 5 children
Shamte Mussa Shamte Shamte 1951 Stone Town 13 28 Mikunguni form 6 + prof.dipl. English teacher single
Juma Ali Njowine Njowine 1956 Bara 7 18 Tandika grade 4 farmer 1st wife, 4 children
Mwinjuma Haji Mabruki Mwinyi 1957 Stone Town 12 25 Kwahajitumbo form 4 + prof.dipl. , medical assistant single
Abdul Hussein Moha=ed Abdi 1959 Stone Town ±13 20 Kwahaiitumbo form6 customs official single
Masoud Khamis Masoud Masudi 1960 Stone Town 12 24 Kwahajitumbo grade 7 , businessman single
Rajab Omar Rajab R.Mtoto 1963 Pemba 12 13/14 Mikunguni form 6 + univ.BA accountant 1st wife, 1 child
Ali Maulid Hussein Maulidi 1968 Stone Town 15 26 Kikwajuni form 3 + koran ass. businessman single
Abdulrahim Muhiddin Foum Abdu 1971 Stone Town 16 18 Boraimani grade 8 + koran mason 3rd degree single
Other Players:
Ali Ramadhan Mrangi 1939 Bara 25 - Mikunguni koran mason 1st wife, 6 children
Ali Juma Ali Kipara 1951 Stone Town 10 18 Ulaya Jarnhuri grade 8 soldier , 3 wives, 6 children
Othman Omar Othman Othman 1953 Donge 15 - Mwembe grade 7 fisherman ' 2x divorced,
Makumbi 2 children
Ali Abdallah Hassan Abdallah 1954 Makunduchi 15 - Muungano form 3 +prof. dip!. welder 1st wife, 3 children
Omar Moha=ed Ali Omar 1958 Pemba 7 - Kikwajuni koran + adult ed. shopkeeper ! 2nd wife, 3 children
* Bara refers to the mainland of Tanzania. Sbamba refers to the interior of Zanzibar.
A board game differs from other games in that the board should play a crucial
role. It is possible to design a board for dominoes that puts pieces on fields instead
of randomly on the table. However, in this case the board does not necessarily limit
the possibilities in the game. If a game has borders, like a football-field, with a
limited number of positions within these borders, like the suggested domino board,
which reduces the number of possible positions of the limited number of playing
pieces, then we would speak of a board game. The board might be in the shape of a
table, or made with street tiles, or as large as a football-field, but it should have a
'specifically arranged surface', i.e. discreet positions, that defines the game.
Previous ideas
Board games can be classified in different ways (Murray, 1952; Bell, 1960). The
principles on which such a classification is based determine the outcome. However,
the classes distinguished by Murray (1952) and Bell (1960) seem poorly defined.
Hunt games in Murray are considered to be war games by Bell. They both have a
separate class of 'mancala' games, which I refer to as 'manqala', which remains
unexplained. Murray (1952:4) based his classification on the view that games are
typical of the early activities and occupations of man. Although Murray's
classification has been widely accepted, there seems to be no theoretical base for the
classification of board games. The reasons for distinguishing games of skill or pure
reflexion (Reysset, 1995) from games of chance, or addressing manqala games as
games of strategy continue to be arbitrary.
Formal classification
Race and war games imply a certain purpose to the game, while hunt and positional
games imply certain techniques. Classes of board games should be distinguished
according to well-defined theoretical principles. Only in this way can we progress in
determining which game belongs to which class. There are five principles for
classification that need to be recognised:
The five principles above determine my choice of classes. The fifth principle
deserves special attention. In the classification of board games many interests of
scholars have converged. Historians, art historians, anthropologists, etc. have all
expressed their views. However, the nature of a game is not well explained by its
shape or by the age or gender of its players. These elements explain the relation of
the game towards the society or towards the historical developments of variants with
a similar shape. Historical data cannot explain, for example, why certain manqala
games are played in championships and others are dismissed as children's games, or
why war and race games can be played on the same board.
victory, and a certain purpose for the player. The race game is won by reaching a
position sooner than the opponent, the war game is won by destroying the opponent.
The major difference is that the war game tries to disable the opponent, while the
race game just tries to slow him down. Such differences I will call characteristics of
the game.
Board games
involved. Hunting games (and games of alinement - referring to the lines that are
formed when playing such games) are usually games where positioning in itself can
lead to victory. In most other war games capturing is the only means of reaching a
victory. These distinctions can also be made in the division of race games.
In the above classification, we differentiate between two classes that identify the
nature of a game; war games and race games. The nature of a game implies a certain
purpose that these games have in common. The purpose of war games is defined as
destruction. In the subclass of war games we then find two purposes of this kind:
war games of pure destruction and those of implied destruction. These members
find their distinction confirmed in a further division. This division is based on the
different means that can be used for victory. The nature of the game should agree
with the purpose and the means by which a victory is achieved. Consider the
following two examples:
1. In a war game where pure destruction is the purpose, we expect that only
capturing leads to victory, since capturing is a direct form of destruction.
Subsequently, implied destruction can lead to victory through positioning, but only if
this positioning implies destruction, such as in the game of Go. 'Enclosement' is the
only victory that adheres to this requirement. Escape is the necessary counterpart in
games where there is only one 'encloser'.
2. In a race game one expects that only positioning leads to victory, in which case
capturing is used for obstruction of a position but cannot lead to a victory in itself.
All race, war, positional, alinement, manqala and other board games can be
classified in this way. There seems, however, a certain prestige involved if a game is
grouped with Chess rather than with noughts and crosses. This prestige, that is
usually communicated with the words skill or strategy, is best exemplified by
championship games.
3. Championship games
Games with dice or other randomizing devices can be played in championships,
where the structure of the championship eliminates part of the element of chance.
By averaging over a significant number of results, games such as backgammon can
earn their reputation as championship games. Most games, however, change when
they are played in championships. The time limit in play and special impasses in the
game need to be determined and solved. Most importantly, the game needs to have a
substantial unknown element, that cannot be analysed beforehand. All war
presupposes human weakness and seeks to exploit it (Von Clausewitz, 1817:304).
The human element makes war war, and a war game a war game. It is only for this
reason that some games are considered as games of skill and others are not.
However, if the unknown element is too large, then the game turns into a children's
game, since the moves are played at random too often. This leads to the following
conclusion: A game cannot become a championship game if the choice between
moves cannot be determined by the human mind, being either imposed too much or
beyond human comprehension. If a player cannot calculate further ahead, then the
player's move will be a random one as soon as deeper calculations are required for a
choice. Therefore, as long as a game has an unknown element beyond the scope of
the human mind it also has an element of chance. This element of chance has also
been identified in war but as in war games it has often been neglected.
Bell (1960:199) presented a list of strong contenders for the title of the best
game: Backgammon, Wari, Wei-ch'i (Go), Pachisi, Nine Men's Morris, Continental
Draughts, International Chess, and others. Apart from Backgammon and Pachisi,
CLASSIFICATION OF BAO 27
which are race games, these games are all war games (including Nine Men's
Morris), and, with the exception of Draughts, are war games with implied
destruction. This is perhaps based upon coincidence and, indeed, interesting race
games have already been mentioned that do not fit this classification. However, there
is only a limited number of games in the world in which major championships are
held. Most games mentioned by Bell are championship games in that respect. The
games proposed by Allis et al. (1991) are also problematic, as is shown in Chapter
IV. Complexity for computers does not define complexity for humans.
Not all games of implied destruction should be considered as possible
championship games; on the contrary, most hunt games are considered very simple
in their nature. It is already argued that championship games depend on an exact
degree of uncertainty that must be attained in order to challenge the human mind just
enough. War games with implied destruction seem to have more ways of controlling
this uncertainty than other classes.
It should be noted that championship games still do not form a separate class. In
theory, each individual game can become a championship game, however, some
games have intrinsically greater likelihood of reaching such a status.
4. M anqala games
Manqala games have long been considered a separate class within the classfication
of board games (Bell, 1960; Murray, 1952). Not only are there so many kinds of
manqala games, but they also have a distinctive look. The outlook of board games
does not allow for a separate class of games, as was pointed out before. The name
'manqala' is, therefore, useless from a theoretical perspective. However, from a
historical perspective this might be quite different, since manqala games underwent a
separate historical development. For this reason, it might be useful to discuss
manqala games separately and discuss the classes of games that can be found among
its variants.
All manqala games appear to be war games. Their primary purpose is capturing
which leads to destruction. The only feature that distinguishes manqala games from
other board games is at a much lower level, namely the ways of moving. The ways
of capturing in manqala games can be found in other race or war games as well. But
the way of moving by spreading seeds (or similarly shaped pieces) is unique to
manqala. Therefore, from a theoretical point of view, manqala games should be
defined as games with a particular move, that of spreading pieces. Manqala games
should not have the status of a separate class.
5. The classification
In the above classification, hunt games are a type of war game with implied
destruction in which positioning leads to victory. Manqala games seem to be war
games but have games of pure and of implied destruction. Certain games of
alinement (three-in-a-row) are considered race games, since they do not destroy the
opponent in play. They do not capture pieces of the opponent and are, therefore,
purely positional. Others (nine-men's-morris) do destroy and are classified under
war games. The following picture shows the level of each group of board games as
defined by Murray (1952) and Bell (1960) in the classification proposed in this
study. This picture shows, for instance, that some alinement games are war games,
in the above classifcation and some alinement games are race games.
28 LIMITS OF TilE MIND
Board games
Race games are different from war games in that there divisions are based on the
concept of getting ahead of rather than destroying the opponent. They can be divided
in the same way as war games.
Murray (1952:9-11) suggested a subdivision into ways of capturing and ways of
moving. Beyond this subdivision I do not think it is necessary to classify any further
since most games will already have a separate classification attached to them.
If race games are observed in depth then it appears that capturing leads to
obstruction instead of destruction. The opponent is slowed down by captures. The
nature of capturing becomes different in war games, where captured pieces almost
always change owners. They either go to the opponent or they are taken out of the
game.
7. Conclusion
Most classes of games that were proposed by either Murray (1952) or Bell (1960)
can be defined individually with the classification below. This classification not only
shows the relationships between the games, but also the level on which such games
should be distinguished.
In summary, the traditional groups of games can be classified as follows:
1. War games: class of board games
2. Race games: class of board games
3. Hunt games: class of war games of implied destruction where
positioning in itself can lead to victory.
4. Alinement games: class of race games where only positioning can lead
to victory (Three-in-a-Row)
class of war games of implied destruction (Nine
Men's Morris)
5. Manqala games: class of war games of implied or pure destruction
where capturing in itself will lead to victory, with a
particular move of spreading pieces
For reasons of convenience one can still refer to the traditional names for these
groups of games as long as it is understood which theoretical relevance they have in
the overall classification of board games.
Board games
B. Classification of manqala
continues to be a war game, but it does change the nature of the manqala game, since
the objective of the capturing move is different.
Diagram 7. Manqala
AAA~AA
one direction direction options
~
continuing not continuing
~
continuing not continuing
/\/\/\(\
re-enter not re-enter not re-enter not re-enter not
1. directional option: some games have one direction of play only (usually anti-
clockwise), some games have two directions to play to.
2. continuing option: some games always end a move after the first spread, other
games continue a certain spreading with a consecutive spreading.
3. re-entering option: some games re-enter the seeds that are captured in the same
move, other games take them out of the game.
These divisions overlap, however, within the tier of the means of the move we can
make consecutive distinctions (see Diagram 7).
We can only include all these options because they are uniform in their effect on
the means of moving. In the course of a move, contrary to the beginning of a move,
a direction can change, the move may continue or the seeds can be re-entered, all
these options are ways of continuing a move. The start of the move and the direct
effect of reaching a certain hole, is dealt with in the discussion of 'purpose'. Each
time we continue a move we continue with the same purpose of capturing. The
options of continuing, therefore, can distinguish the manqala games only in the tier
of means. After the nature of the move, spreading follows the purpose of the move,
which is capturing, and both evolve by means of continuing the move. Spreading
(and capturing) continues:
1. in another or the same direction
2. with or without another spreading
3. with or without the captured seeds
If we take the Bao game, we should speak of a four-row manqala game, where
only one hole at a time can be captured, where the move can continue in various
directions, with a consecutive spreading, and with captured seeds. A Bao expert
might add that direction is not always optional, that continued spreading allows for
more captures and also has exceptions and that capturing is sometimes possible but
not allowed. All these comments should, nevertheless, not matter for the initial
classification of the game.
4. Position in manqala
In order to complete our classification we should distinguish position. Position
distinguishes games in a strategic way. Some positions allow for interesting opening
play, some positions have to be constructed with different rules, other positions
require more seeds. The distinctions imply that the game has certain strategic
limitations added.
The positions should in general be distinguished by the number of holes and
seeds. This allows for unlimited distinctions, however, 5, 6, 7, and 8 holes are most
dominant, as are 2, 3, and 4 seeds per hole. Since the numbers are interdependent, it
is possible to combine them as one distinction. Also, some games have holes empty
and seeds outside the game, and these same games often have specific numbers of
seeds in specific holes. These variations determine the opening strategies of play.
During the game, most games return to the more general features of their fellow
games without the exceptions of the position making much of a difference. These
positions are most important for distinguishing strategies and determining the rules
of strategy. The variations of position should, therefore, be applied conservatively
in the classification.
CLASSIFICATION OF BAO 33
5. Conclusion
Manqala can be successfully classified following the same principles of
classification as board games in general (see III.A). However, we have to apply these
principles using nature, purpose, and means of board games to the moves of play.
The literature (Murray, 1952; Bell, 1960) has classified manqala using many similar
distinctions, but has not created a hierarchical order, nor did it construe an unified
approach towards classification. Most importantly, the definitions of the classes
should give insight into the classification and the role of games in this classification.
It is only for this reason that a new classification is presented here, in order to
structure the lists of games and give a point of departure for further discussion on
classification.
C. Variants
In order to distinguish between two variations of the same game and two instances of
different games we need a set of definitions that help us determine the boundaries of
a single game. Such a set of definitions also depends on the approach that is taken.
From a historical perspective, it is perfectly legitimate to include predecessors of a
game in the definition of a modern game. It is also possible to limit a game to a
certain social setting or geographical area. However, all these definitions either
include too many games or exclude too many games from a general theoretical
perspective. It seems also, that, intuitively, players have a general idea of what
constitutes a variant and what a different game. For example, in the time of Ruy
Lopez, approximately 1560, Spain, France, and Italy had differences in their way of
playing Chess (Murray, 1913:812). These differences did not make them different
games, considering the interchange of players and theories.
As soon as the nature of a game is changed by a change in rules we speak of
different games. This is not only a logical conclusion after our classification but is
also understandable in that different strategies apply to games where the object of the
game is essentially different.
The above provision still includes too many games as variants of the same game.
Classification in itself does not suffice.
In the history of modern Chess we can identify variations in rules that posed few
problems in 1560. These rules concerned, for instance, castling and the en passant
move. It not only concerned changes in the circumstances where castling was
allowed but it also referred to how this should be executed and which pieces were
involved (Murray, 1912:830-833). This eventually led to the first internationally
accepted set- The London Rules. Later this was complicated with time restrictions,
the 50-move rule, then with exceptions to 75 moves for certain end games, etc.,
without a perfect solution for all situations (Hooper & Whyld, 1992:134). A similar
set of variations can be made for Bao, for instance, takasia, and opening moves
(toboa), and also what to do if the nyumba has only 5 pieces left. The moves have
the following characteristics with which we can define all other rules of variation.
The moves are all 'circumstantial' in the game, which can be defined with the
following three characteristics
34 LIMITS OF THE MIND
Variants of a game can now be defined as games with the same nature where the
variation only applies to unnecessary, rare, and restricted rules. The variants should
not be classified in the formal way applied before, but should be seen as in a family
relationship (Wittgenstein, 1953:§66).
IV: ZANZIBAR BAO RULES
A. Zanzibar Bao
1. Previous research
Four-row manqala games are only found on the East coast of Africa (and in China,
see Eagle, 1995). They can become extremely complicated, Bao being the ultimate
example. Flacourt (1658) was the first writer to describe a move of Bao (Townshend,
1986). This French text of the modes of play of Fifangha is, unfortunately, rather
obscure to the layman who is unfamiliar with the game. In 1913, no other earlier
accounts have been discovered, a much better attempt was made to describe bau (=
Bao) as it was introduced to the Yao people in Nyasaland (Sanderson, 1913). A well
structured description, and an example of a game, explained in a reasonably
adequate manner the characteristics of the game.
Ingrams (1931) was probably the first to describe the game in Zanzibar. His
scanty description captured a few interesting rules, although some moves were
described incorrectly. Murray (1952), a collector of many manqala rules, combined
Sanderson and Ingrams into a new description which is a poor representation of both
of their works.
It is said that in the 1970s the Tanzania Bao Society started collecting material
about their game and also wrote their rules as they were to be used in championships
(especially those between DarEs Salaam and Zanzibar). A copy of these rules shows
that all intricate rules are mentioned, but not explained, and therefore, need an expert
(bingwa) to teach a layman. In this same period the National Museums of Tanzania
published a pamphlet containing the rules of East Africa. Their source is still
unknown. However, their description is one of the most accurate and useful to date
and widely used in tourist shops to accompany tourist boards.
In addition, Townshend's work (1986) concentrates on the Lamu players (Lamu
is part of an archipelago off the coast of Kenya). His account is lengthy but very
accurate and mentions many of the complex rules in detail.
Surprisingly, none of the above descriptions is complete or without question as to
certain situations in the game. After years of playing, it has taken me much effort to
find the key to the problem of describing Bao rules. This problem appeared to be the
hierarchy of Bao rules. Only a presentation that shows the hierarchy of rules in a
structured way is likely to be complete and insightful as to all Bao game situations.
The hierarchy is problematic in all other descriptions and never fully explained. My
analysis is meant to clarify and solve these problems. The structure of the rules has
gained more importance than the listing of the rules. All further amendments to the
rules I describe should fit the basic structure I have create_d.
Because of the complete and detailed explanation of the Bao rules, my
presentation of rules cannot teach the game effectively. If a player is taught the
game, he should first master the most frequent rules before looking at exceptions.
Reading the earlier researchers might well give better insights into the most frequent
rules and which rules should be the first in the learning stage of the game.
2. The analysis
In the analysis of the Zanzibar Bao rules, I will concentrate on the question of how
the rules of a championship game work. The deeper the understanding of playing,
36· LIMITS OF TIIE MIND
the closer we come to the understanding of the reasons why the Bao game has
developed with this structure of rules at its core. Rules are not decided upon by
commissions. Groups of players decide upon rules and championship referees
mediate when groups of players meet. This interactive process gives us the only
reason for suspecting an underlying psychology of the Zanzibar Bao rules.
Since all players in Zanzibar are male, although the women sometimes play an
easier variant, I will refer to a player as 'he'. The rules specifically refer to the
Zanzibar variety. While there are few variations between mainland players and
Zanzibari, specific rules do vary, so a choice for 'Zanzibar only' was made.
Various scholars have tried to capture all the rules in one document. Even the
Tanzania Bao Society has made an attempt in this direction. The Swahili document
that contains this description is, however, written for a players' audience and
requires considerable prior knowledge of Bao terminology and play. It is the only
document that lists most of the very specialised rules. Therefore, with the help of this
document and my own fieldwork, the following analysis will present a more
complete description of the Bao game together with a display of the dynamics of the
rules. The analysis of Bao will determine the hierarchy within the rules and will help
us define more accurately the basic concepts of the game.
B. The board
In order to understand the rules we have first to define certain physical elements of
the board. The definitions of these aspects are, sometimes, more specific than this
first physical characterisation suggests.
The board
The Bao board consists of 4 rows of 8 holes. Each player owns two rows.
The seeds
Each player has six seeds in the house (see nyumba) and two seeds each in two holes
directly right of the house. Both players have a remaining i:mmber of 22 seeds in
stock.
Nyumba
There are two bigger, usually square, holes in the middle. These holes are called
nyumba (house) or kubwa ya mpango (big one of the plan, i.e. the big hole which
was planned as a big hole from the start).
Kichwa
The far end holes on each side of the inner rows are called kichwa (head).
(Sometimes also used for the far end holes of the outer rows, but not in the following
account.)
Kimbi
The penultimate holes on each side of the inner rows are called kimbi.
ZANZIBAR BAO RULES 37
~
Kkhwa
•Kimbi
Diagram 8.
Reverse holes
I define kimbi and kichwa, the two far end holes at each side of the front row, as the
reverse holes.
Singleton
I define a singleton as a hole containing only one seed. Although it also refers to a
single seed, it is sometimes more convenient if 'singleton' refers to a hole with one
seed as well.
The above terms are the basic terms of the Bao game, it is impossible to explain any
of the following rules without referring to the above definitions.
Such definitions are necessary for any board game. I defined a board game as a
game in which the board and the pieces are well-defined and limit the number of
positions in the game. This characteristic of a board game is shown by the limited
number of playing fields and the restricted number of seeds in Bao. The definition of
rules limit these -positions even more and are shown in the following sections.
The basic move, spreading, is the most characteristic element of the manqala games.
The move distinguishes manqala games from all other board games. A basic move in
Bao is a consecutive spreading of seeds (or other pieces), i.e. placing seeds in holes
one by one, in adjacent holes of the board in a (anti-)clockwise direction. In Bao, this
spreading of seeds also occurs when capturing or emptying holes in general.
Moving occurs in a certain direction, in Bao this is either clockwise or anti-
clockwise within the rows of one player.
1. Lala
If holes are not allowed to be emptied the move stops. This is called !ala (to sleep)
in Swahili. The end of a move only occurs when the last adjacent hole in a
consecutive spreading of seeds is an empty hole, i.e. when the last seed of a move
falls into an empty hole.
38 LIMITS OF THE MIND
It is, therefore, impossible to end a move by simply placing one seed in a hole, since
a hole has to be emptied before that move can stop. Only spreading, which is the
characteristic feature of all manqala games, is considered a move of the game.
2. Endelea
When you continue (endelea), the last seed falls in an already occupied hole. You
continue by taking up all and you continue spreading these seeds in the same
direction.
It is possible to design a move which continues forever, i.e. never reaches an empty
hole. Such moves rarely occur and, surprisingly, have no rules that decide on the
problem. Masters usually do not play a move of which they do not know the
outcome. Their opinions differed widely when they were asked what the best
solution to a never-ending move would be.
1. Namua
In namua, the stage of the game where there are still seeds in stock, the definition of
the house becomes more complicated, since its existence is also bound by rules:
2. Mtaji
Mtaji is making a capture by taking up all seeds of one hole when the stock is
already finished:
1. from the front or the back row and
2. from a hole with more than one seed and less than 16
A player plays:
1. namua and captures
2. namua and takasa (see below)
a. unless he can play namua and capture
3. mtaji (see below)
a. unless he can play namua (=unless there is still stock)
ZANZIBAR BAO RULES 39
3. Kula
Kula is capturing the adjacent occupied hole of the opponent. The circumstances, as
described above, vary between mtaji and namua. Kula itself proceeds as follows:
After taking up all seeds from the opponent's hole, the captured seeds from the
opponent are spread into one's own front row:
1. from the far end nearest to the reverse hole they were captured from
2. from the far end in the direction where the capturing seeds were spreading to.
a. unless they were captured from a reverse hole
3. from any far end the capturing player chooses.
a. unless they were captured from a reverse hole
b. unless the capturing seeds came spreading from a certain direction
When capturing seeds, situations occur where players can choose. In one case, one
can choose the side from which to enter the captured seeds. In another case, there is
a choice between continuing with the house after a capture or stopping at the house
(lala). Continuing with seeds after a capture is not takasa, but is usually called
endelea (continuing). Takasa refers specifically to spreading seeds without capturing
(see below).
In the above description, there are two hierarchies. The first shows that mtaji is a
less complicated version of capturing. There is always a direction of play, and
therefore, the ability to choose the direction is lost. In the second hierarchy, the
choice is with the house, which also disappears when playing mtaji.
The rules of spreading after the first capture are not hierarchical. However, they
show a certain order. Capturing always needs an adjacent occupied hole, a front row
and an occupied hole. The house is a special case when there is no adjacent occupied
hole. Takasa is independent of the front row, while sleeping does not even need an
occupied hole. These rules can be presented in various ways; however, the order
shows their relative complexity most clearly.
As long as there are seeds in stock, the players are obliged to make a namua
move. Therefore, namua is higher in the hierarchy than mtaji.
40 LIMITS OF TilE MIND .
Takasa is a move that is inferior to any move which captures. This explains the
variation between the rules of takasa in namua and takasa in mtaji situations. These
two varieties of capturing pose different problems in takasa and are solved
separately. The players speak of takasa in both instances.
In namua takasa, one seed from the player's stock is placed in his row with the
following restrictions:
1. Only occupied holes in the front row can be played
This rule defines a lost game if the front row is empty.
1 '. A kichwa cannot be emptied towards the back row if it is
the only occupied hole in the front row.
This rule defines a lost game if the front row is temporarily empty.
2. The house cannot be emptied
a. unless it is the only occupied hole in the front row when it will be
emptied as if it was a singleton.
3. Singletons cannot be emptied
a. unless it is the only occupied hole in the front row
b. unless the house is still present
The house cannot be emptied. However, various moves that do not directly empty
the house may indirectly reach the house. Instead of emptying the house or
disallowing the move, a solution is found by making the move stop at the house if it
reaches the house with its last seed. This is analogous to the optional departure of the
house in namua rules (and the takasia moves discussed below).
If the move does not reach but only passes over the house, there is no change of
rules necessary, the seeds of the house are simply added to.
Takas a shows a strong hierarchy of rules and exceptions. In namua situations the
front row is at the top of the hierarchy. No move starts outside a front row. The
object of the game is to empty all the holes of the opponent's front row. Even a
temporarily empty front row is, therefore, a lost game.
The house is the focus of all play in the namua stage, but it cannot overrule the
front row rule. Therefore, if the players want to prevent the house from being
emptied, rules have to be constructed that bypass the front row rule. In order to
understand 2.a., rule 3 has to be understood first. A singleton cannot be emptied but
must obey the front row rule and the rule of the house. Therefore, singletons are
lower in the hierarchy. Although singletons are lower in hierarchy than the house, to
play a singleton remains an exception. This exception is used to solve the problem
of the house in 2.a. Instead of playing the house, the house is played as if there was
only a singleton, i.e. only one seed is taken from the house. This causes minimal
changes in the number of seeds in the house and adheres to the hierarchy.
Furthermore, it refers to the other exception to singletons that had already been
installed.
ZANZIBAR BAO RULES 41
The rule of the kichwa is part of the front row rule, since it only further
determines a lost game. The rule is not linked to the house or to singletons and is,
therefore, only a subrule to rule 1.
In the mtaji stage (more precisely: when there is no stock), takasa means that all
the seeds from one hole are spread in a certain direction with the following
restrictions:
1. Singletons cannot be played (be emptied to takasa or make captures).
This rule defines a lost game if there are only singletons in both rows of a
player.
2. Only occupied holes in the front row are emptied.
a. unless they are singletons and the back row can be emptied instead.
This rule defines a lost game if the front row is empty.
2'. A kichwa cannot be emptied towards the back row if it is the only
occupied hole in the front row.
This rule defines a lost game if the front row is temporarily empty.
In mtaji situations the back row can be played as mtaji as well. Therefore, the
high ranking of the front row rule is lost. This confirms my earlier statement that the
capturing moves determine the hierarchy of the takasa rules. The rules for the house
,, disappear once we reach the mtaji stage. Therefore, singletons automatically become
1 the top of the hierarchy.
Before we continue with the discussion of takasia rules, it should be noted that
the hierarchy shown in the above paragraphs assumes a certain history of the game.
Rules high up in the hierarchy are likely to have been there first. It is also likely that
the first stage of entering seeds in the game is a later invention, since many manqala
games are known without this first stage, but no game is known without the second
stage, i.e. capturing without entering seeds. Similarly, one can argue that other rules
are of later or earlier origin in a similar way. This is certainly the case with the rules
of takasia, which, in my opinion, are an invention postdating the rise of mastership in
Bao.
2. Takasia
The lowest in the hierarchy of rules are the rules for takasia (not to be confused with
takasa).
A player who takatas in such a way as to directly menace only one of his
adversary's occupied holes without exposing any of his own seeds cannot be
deprived of the capture he has set up. It must be reserved for him in the next
move.
Townshend (1986:118)
The takasia move is the lowest in the hierarchy, it rarely occurs and only when
the namua stage has ended.
When a player empties in such a way that the opponent cannot capture and that
he himself will be able to capture in his following move, then the takasia rule obliges
the opponent to leave his takasia-ed hole to be captured in the next move. The
opponent cannot empty his takasia-ed hole:
42 LIMITS OF THE MIND
There is no further hierarchy within these exceptions, except that the takasia rule
is below any other rule in the game. It also confirms that takasa is below capturing in '
the hierarchy. This is not very surprising because capturing is obligatory throughout J
the game.
II
see paragraph C.
2. Continues
a. unless it falls in a takasia-ed hole
If a hole is indeed takasia-ed it may not be emptied directly, however, as with the
house in the takasa rules, it may be reached indirectly. Without disallowing such
moves, the move will end at the takasia-ed hole if the last seed of the move ends
there. This is analogous with the moves for the house in namua capture and takasa
rules. If the move does not reach, but only passes over the hole in question, there is
again no change of rules necessary, the seeds of the hole are simply added to.
4. Thehouse
The house does only exist if it has not, at any time, been emptied. In that case, the
house would become an ordinary hole. If the house has only 5 seeds the front row is
higher in hierarchy than the house and, therefore, the house should be takasa-ed if
necessary, unless it reaches 6 seeds again. (In DarEs Salaam the house is takasa-ed
if it has only 5 seeds, even if there are singletons left to play.)
The house still exists when the namua stage is over, circumstances providing, and
when there has. been no mtaji yet. In this period, all the namua rules for the house do
not apply. Moreover, all the takasia rules that may occur in the mtaji stage do not
apply in any way to the house either. Therefore, the house can be takasa-ed, but not
takasia-ed.
When considering both the takasa rules for namua and for mtaji, it becomes clear
that the rules for namua are a more complicated version of those of mtaji. The
introduction of the house at the beginning of the game complicates the hierarchy of
rules. Only the takasia rules which, by definition, cannot occur in a namua situation
(the opponent can always capture when he is takasia-ed in the stage of namua), adds
some complexity to the mtaji situation. The house continues to be a special case in
both situations, as long as it is there.
At the time of this study, another rule appeared, which said that a hole that was
takasia-ed could be emptied in the move that takasa-ed the house. The house never
sleeps, stated the rule. However, in a discussion about the rules Kijumbe strongly
opposed this rule, he stated that such a rule would become a swindle, since it was
unlikely to be calculated beforehand by the takasia-ed party. The group present at
that time, Shamte, Abdu, Rajab, Kijumbe, and myself eventually agreed that it was a
bad rule. Later, Mkiwa also denied this rule as being correct. Since the situation in
which the rule occurs is very rare, the opposition of Kijumbe had no serious effect
ZANZIBAR BAO RULES 43
on the Tournament. It is likely, however, that the consensus will be for having the
rule changed (back), which means that even the house will respect a takasia-ed hole,
when the house is takasa-ed.
1. Rejesha mistakes
If a player makes a mistake during the play of a move, the opponent is allowed to
object. The move is returned (rejesha) after which the play continues. If a player,
usually the one who made the mistake, returns a move and fails to reconstruct the
situation he will have lost the game.
!.
44 LIMITS OF THE MIND
In my opinion, each championship game has rules necessary for the advanced
players to increase competitiveness in championships. These rules facilitate a chance
of recovery in play without completely eliminating the advantage of the winning
party. I will call this counteractive process: the increase of skill, i.e., the rules
demand more skill from a player in order to win.
These rules can give us insight into the definition of skill in general, by studying
and analysing the influence of the rules on play. It should "be noted that the process
mainly involves rules that we consider variations in a game. We consider only the
ZANZIBAR BAO RULES 45
1. Recovery
In championship Bao there are a number of rules that increase the chance of
recovery. For instance:
1. takasia, increases the chance for a mtaji
2. the restrictions to the movements of the house facilitates the defence against a
powerful house
3. the move of a house with less than 6 kete gives an advantage in a weakened
position, where the house could be easily lost
4. the rules for the house in the takasa stage facilitate the spreading of the house at
the point where it already failed to be dangerous
The counteractive process mentioned above also constricts the extent to which the
chances of the defence are increased. Takasia is the most dramatically amended rule
with more than five subtle exceptions. The rules for the house are also highly
complicated and delay the point of capture and/or departure in many ways. This
confirms that in order to balance chances most evenly, rules in Bao tend to become
more complicated.
2. Delay ofvictory
Takasia involves three moves at once. One player cannot capture, and plays in such
a way that the opponent cannot capture either, and that he himself does capture in the
next move. If the amendments to the takasia rule were not enforced then the move
would gain too much importance. The move could make important gains for both
players without increasing, in any particular way, the chances of the losing party.
The considerations I put forward concerning the takasia rule appear to be rather
marginal. In fact, the adjustment that this rule makes is of such minor importance to
lesser players that this rule is hardly mastered by this group. The specifics of this
rule are only known to championship players. For these players the rule can make an
important difference. The rule is at the bottom of the hierarchy of rules in Bao.
Therefore, the rule changes very little but offers a chance for a player to recover.
This subtlety identifies our first border of master's excellence. The definition of this
border is in the definition of this rule, a variation at the bottom of the hierarchy of
rules.
The takasia rule can increase the possibilities a player has in an almost lost
position. It delays the opponent's success. The fact that this rule has so many
exceptions implies that the losing party needs very little in order to gain enough for
recovery and also that the rule itself might, without such restrictions, tip the balance
too much in one particular or in other similar situations. The takasia rule defines a
border in the Bao master's excellence in the following way: A game is not decided
until a forcing move cannot hurt the winning party in any significant way. A forcing
move can indeed complicate matters so much that there is delay of victory.
Kijumbe translated the rules of takasia as a reward for a good takasa move. It is
not necessary to be in a losing position. Kijumbe and others agreed, however, that
the rule made it more difficult to win a game. It remains an important weapon if you
have a lack of capture moves.
46 LIMITS OF TilE MIND
3. Delay of climax
We can consider the rules for the house in Bao in a way similar to the rules for
takasia. The house has many restrictions and privileges. The house is often described
as the bomb or the treasure that is defended and captured. Indeed, the departure of a
large house can result in a deciding blow to the opponent, as can its capture. In Dar
Es Salaam, I watched a form of handicap-play, where the lesser player is allowed to
capture the house in his second move (he starts with A6L, his opponent playing a6,
thereby leaving his house undefended). They add to the seeds in the house if they
want to increase the handicap. In Bao, both the departure and the capture of the
house are complicated by rules. In championships, these rules are applied with the
following complications:
1. a house cannot be takasia-ed
2. a house of five (alone in its row) has to depart
3. takasa with one seed from the house is pennitted
4. singletons can be takasa-ed when the house has not departed ·
5. except for a house of five, the house is never obliged to depart
These specific rules have little meaning for a layman, the description of the rules
might even confuse regular players. However, these rules illustrate an aspect of the
Bao game which is grasped only by masters. Although the exceptions are less
complicated, the rules are still low in the hierarchy of rules and should be called
variations. As with the rules for takasia, they significantly constrict the playing
possibilities: In this case the departure possibilities of a house for Bao masters (the
capture of a house is unusual in their games and usually coincides with the departure
of a house).
If a house has not departed before reaching the game of mtaji then this house is in
an unfortunate situation, since it is more and more likely to be captured. The only
chance of saving it is to takas a the house favourably. It is essential that it cannot be
takasia-ed (see 1) since that would force the player to give up his house. There is
another reason for such an exception. Since takasia can only occur in a mtaji
situation, the house will have lost most of its defending privileges it had in namua
and is, therefore, more vulnerable.
Another defence rule can keep the house from being captured when it reaches
five (see 2). There is a forte play possible that slowly empties the house:
This is when all holes of the opponent are filled, except for a4, while end holes al
and a8 have only one seed. The other player has only the house left in his front row.
He takasa-s to the right or to the left, in each case all his singletons can be captured
by playing a2 or a5R, respectively. This would empty the biggest house without a
chance for defence. The house of five can, however, be departed and sometimes
gains a small advantage because of often increasing attacking possibilities.
For a Bao champion the above rules increase the chances of retaining the house
and also increase its danger: the bigger the house the bigger the danger. If a house is
not departed then it has to be takasa-ed in the mtaji game, where another rule
protects it. Again, the many rules that involve the house also define a boundary to a
master's excellence. The continuous presence of a house might complicate matters
so much that the victory of a player when a house is still in play is beyond the
powers of calculation of the players.
4. Boundaries of excellence
The two discussions above refer to rules that complicate matters for the master and,
therefore, define a boundary to his excellence if he cannot calculate further. It is as
ZANZillAR BAO RULES 47
if not all the cards are played and the master has to wait until all information is
present. In a similar way we can consider castling in Chess, which as long as it is
possible can change a situation with two pieces at the same time and might,
therefore, change the situation. A Chess master often intends to keep the possibility
of castling. Castling rules have indeed become more detailed in Chess in order to
define the exact circumstances in which castling may proceed.
A game which offers both regular players and masters certain borders within
which the game is played most competitively is, in my opinion, a championship
game.
Even in comparison with other manqala games, Bao has striking characteristics
because of its special rules. Townshend (1986) concluded that Bao, in comparison
with other manqala games, features by far the lowest quantity of captured seeds per
take and the lowest turn-over per move of seeds between players. In addition, Bao
achieves by far the highest number of takes per move, the greatest preponderance of
seeds in the inner row, and the most seeds per hole in the inner row (see also IV.J).
5. Increase of boundaries
I stated that the rules described as characteristic for championship games are
necessary to increase competitiveness. This development can be described in the
following way:
There is a competitive board game and there are two reasons for changing it:
1. The organisation of competitors seeks to delay victory in the game in order to
make it more exciting for the public. Therefore, the game has to become more
competitive; the decisive blow or the climax of the game has to be delayed. The
decision is delayed if the chances of recovery are greater. Subsequently, the climax
is amended by new rules that increase and regulate the odds in the game more
accurately.
2. The organisation only wishes to make the game more competitive for the players
themselves. Again, the same procedure applies.
The system implies that an official organ constructs a championship game. This is
only partly true. It is often a group of people who decide which variant is preferred
and this group may also suggest new time limits and other constraints. However, the
masters themselves develop the game by playing. The process of developing a game
is what has given the modem championship games their reputation. The hierarchy of
rules usually evolves in time This hierarchy decides which rules prevail and come
first and also allow for interpretation. The newest rules, therefore, undergo much
more scrutiny than the earlier rules. As soon as a consensus about a certain rule has
been reached the rule has actually entered the game. The masters who play across
variant boundaries eventually decide which variant, and, therefore, which
development they prefer to play in championships. The development in itself will
always refer to the situation in which the boundaries of a master's excellence serve
to level the fairness of the game and to eliminate random outcomes or a
(dis)advantage of starting the game. Finally, championships ask for maximal
competitiveness since masters play masters. Important situations that dictate the
course of the game are amended and become as subtle as the rules of the games
described.
48 LIMITS OF THE MIND
6. Championship games
At this point, we should briefly return to the discussion of championship games. It
appears that each game can be played at a low level of skill if players choose their
moves randomly. The extent to which a player can influence a game by his choice
of moves determines the general status of a game. An expert plays with skill to a
limited extent, i.e. the number of significant moves he can think ahead. If all moves
can be envisaged by a regular player, as in noughts and crosses, we do not have any
competition and we cannot speak of a championship game.
I maintain that a random choice of a move is an unavoidable element in each
board game. The randomness is reduced according to the abilities of the individual's
mind, most especially when that individual is a master. Reduction is partially
imposed by rules. In a championship game the higher the abilities of the players the
greater the reduction of random moves has been or should be taking place.
Therefore, a game needs a strong competition in order to prove that this ability really
makes a difference. The element of randomness is still present, but reduced by the
existence of rules, and most importantly, can be reduced still further by a player's
abilities.
Such a concept of championship game is different from 'perfect' or 'imperfect
information' games as used in computer science. All information to calculate the
outcome of the next move is available. There is no imposed element of chance, like
dice, that influences the outcome. However, random choices or imperfect
calculations also count as elements that influence the outcome. Expert skill reduces
random choices to a minimum.
Note that the definition of a game does not necessarily include all its rules. Often
we speak of variants of one game when specialized rules are introduced at a later
stage of a game's evolution.
7. Conclusion
The specialised rules in Bao define the thinking space within which Bao masters can
compete. They determine the setting in which the limits of their minds matter.
Bao is the most complicated manqala game. In order to substantiate this claim, we
should provide some numerical data that prove such complexity. The numerical data
presented here cannot, however, distinguish between complexities for humans and
for machines. For instance, Go is currently impossible for machines to conquer
because of the pattern recognition that is involved, while Awari is almost cracked by
computers, because of its calculability and the relatively few possibilities of play
(Allis, van den Herik & Herschberg, 1991).
It was explained that the rules of Bao are complex. Direction changes and
numerous laps to the move all complicate calculation for the human mind. For the
machine, these computational difficulties seem less problematic. This section
presents a brief quantitative characterisation of Bao in relation to other (manqala)
games.
1. State-space complexity
At the present state of computer science, games can be cracked if their total number
of possible positions does not exceed 1012 • Such a number represents a so-called
--. •,--,
ZANZIBAR BAO RULES 49
state-space complexity, defined as the number of legal positions reachable from the
initial position of the game. (Allis 1994: 158-159)
Awari (also Warri, Wari, Awele, see Murray, 1952) has 48 seeds and 12 holes
and is also played in championships. If we calculate the number of possible positions
of 48 seeds on a Awari board, we should have a fairly accurate idea of the state-
space complexity of Awari. In the same way we can calculate Bao and other
manqala games. Certain impossible positions on the board are too few to make a
significant difference on the total number.
We can apply the following formula (Tijdeman, personal communication):
(h-1+s)!
(h-1)! · s!
h =number of holes
s = number of seeds
~ 'I
!i
I Awari: 'I
1!
(12-1+48)! or ____22_L amounts to 2.8 x 1011
(12-1)! . 48! 11!. 48!
i
I
l
~
Bao:
(32-1+64)! or ___2iL amounts to 1.0 x 1025
(32-1)! . 64! 31!. 64!
f
{
Songka:
(14-1+98)! or ___l.1ll amounts to 3.0 X 1016
(14-1)! . 98! 13!. 98!
These are three examples which show that Bao has significantly more possibilities
than Awari, and less but still significantly more possibilities than Songka. Songka is
a manqala gameJrom the Philippines. It is only played by women and children and
has no status as a championship game. This example shows the limited importance
of the above calculations. Awari will be cracked by computers in the near future,
while Bao and Songka are unlikely to undergo the same fate. However, Bao and
Awari seem much more intricate for humans than Songka. This can only be
explained by the fact that outcomes are too random to calculate; the game becomes
'silly'.
The data prove that Bao is much more complicated than other manqala games,
and is very likely the most complicated manqala game.
Manqala is played with identical seeds, each of which represent exactly the same
value (with the possible exception of some Chinese variants, see Eagle, 1995).
Therefore, there is a wide gap between the numbers for Chess and for Bao. In Chess
the pieces have different values, although less numerous and only one per field.
Also, the board has twice as many positions. Chess has, therefore, a much higher
numerical complexity of 1050 (Allis, van den Herik & van der Meulen, 1991).
2. Game-tree complexity
We can give a reasonable estimate for the branching factor of Bao by calculating the
number of possible moves in each turn of a Bao match. If we average a number of
50 LIMITS OF TilE MIND
3. Mutational complexity
Mutational compJexity is not a recognized quantitative criterium for measuring
complexity in board games. However, the number of mutations, not to be confused
with the possible number of mutations, in each game contributes most importantly to
Bao complexity.
In the blind Bao match (Chapter VILA), there are on average 6 positions that
have changed after one move. This excludes the double changes of one position in a
particular move. This mutational complexity is the most crucial part of Bao
calculations. In Chess, such complexity only arises when several moves are thought
ahead. The eventual position has then to be calculated. The complexity of such a
calculation is significantly inferior to the calculations needed in Bao.
The mutational complexity within one move has formed the center of this
investigation into Bao thinking. This type of complexity is completely unknown in
Chess. Draughts and Checkers occasionally have combination captures where a
certain mutational complexity is present. However, in my view, no other board game
matches Bao in this respect.
4. Capture modes
Townshend (1986:134) provides us a table with 'comparative results of simulated
moves for various capture modes'. This list shows percentages and averages of
ZANZIBAR BAO RULES 51
manqala games with various capture modes. Townshend concluded that Bao features
by far the lowest quantity of captured seeds per take and the lowest turn-over per
move of seeds between players. In addition, Bao achieves by far the highest number
of captures per move, the greatest preponderance of seeds on the inner row, and the
most seeds per hole on the inner row.
Key:
A Sambi, resowing from behind
B Sambi, resowing forward
c Sambi, resowing captures plus the contents of one's own last hole forward
D as C, but capturing from inner row only
E Swahili Bao, using kimbi (reverse holes)
F as E, experimentally ignoring kimbi
Further discussion of this table can be found in Townshend (1986:133-134). For our
discussion only the rounded figures of the table are presented without paying much
attention to the statistical details.
The rounded figures of the table illustrate conclusively that the capture mode of
Bao is more 'extreme' and much more dynamic, i.e. more seeds are involved, more
changes are made, and more possibilities are created for the opponent. This
discussion largely supports the argument that was made in a previous Chapter on the
constriction of excellence (IV.H).
5. Conclusion
With the average state-space and game-tree complexity in Bao it is possible to
compare Bao to other board games. Its complexity appears not much greater than
that of Checkers (Allis, 1994:161). However, such a comparison is based on criteria
taken from computer science. Mutational complexity, the essence of the difficulty of
Bao for humans, is not 'complex' for a computer. On the contrary, a computer can
deal with considerable mutational complexity without any serious memory problems
while human beings are severely limited. In later Chapters, it is exactly this
maximum which is investigated in experiments with Bao masters.
V: ILLUSTRATIONS OFBAO PLAY
This Chapter shows the various levels of calculating Bao. With the help of elaborate
examples the different stages in a move, a capture, a combination of captures in one
move, and a combination of moves, are shown. The purpose of this Chapter is
threefold:
Firstly, the elements of a move are illustrated for the beginning player and give
an indication of the workings of the game (examples 1 to 8). Secondly, the
complicated move illustrates the difficulty of calculating Bao and often shows a
limitation of the layman's mind (examples 9 and 10). It is not uncommon for a
player to be unable to calculate his own move. Examples of this sort can also be
found in the tests of duro and complex duru in the following Chapters (VI & VII).
Thirdly, the combination of various moves illustrates the traps and tactical moves
that are involved in Bao (section C). More importantly, when these combinations go
far beyond three or four moves deep, they exemplify a first limitation to the human
mind (Table 2). Even the greatest of masters is not able to calculate beyond six or
seven moves deep. Such a limitation is immediately bypassed by memorising
specific combinations. It is obvious that a combination of considerable length can
surprise many opponents and is, therefore, useful in play. The combination is a tool
for gaining advantage in the game. The player then makes use of the limited
knowledge of the opponent.
Different types of move combinations are shown. Forced moves are common
and, as the expression implies, impossible to escape. Short examples of only two or
three ply deep are standard knowledge for a Bao player (examples 9 and 10). The
very long example which is shown in a two page table is not easily recognised as a
forced move situation. However, despite the wide range of possibilities the player
remains trapped. Finally, some combinations lead to early victory (examples 11 and
12). Such combinations are usually part of opening knowledge (Chapter VIII), and,
therefore, part of long opening combinations.
Only new, but complicated examples of forced move situations can surprise a
master. If the combination is beyond his calculation ability then only memorising
will bypass his mental limitations.
A Rules of play
Rules ofplay
Each player has two rows of eight holes with a square house (nyumba). The opening
set-up: there are six seeds in each house, and two times two seeds in the holes
adjacent to the houses. Each player, therefore, starts with ten seeds, and twenty-two
seeds in stock. The player whose inner (or front) row is empty, has lost the game.
Part I
Each turn a player throws one kete (seed) from his stock in an occupied hole with at
least one seed. If this hole borders on an occupied hole on the opponent's side then
the kete of the opponent are taken (see below).
ILLUSTRATIONS OF BAO PLAY 53
Spreading:
In case there are no seeds on the opposite side, the player empties his own hole and
spreads the seeds, i.e. leaving one seed at a time, in his own row in one or the other
direction. If the last seed drops into an empty hole the player's turn is over. If this
seed drops into a hole with (at least) one seed then this hole is emptied in an identical
way by speading all its seeds until eventually a last seed drops into an empty hole.
Spreading of the house is not obligatory and not even possible if a seed from the
stock is thrown into it. If the house (via another hole) is emptied and spread (or
captures the opposite hole), all rules for the square hole are lost for the rest of the
game.
Capturing:
Capturing is obligatory if possible. When capturing all seeds from the opponent's
hole these seeds are spread one by one over the player's own front (inner) row,
starting from the far left or the far right. If the last spread seed drops into an empty
hole, the turn is over. If it reaches an occupied hole and the opponent has seeds in
his adjacent hole then another capture is made. If not, the player continues spreading
his seeds in his own row.
Those seeds that are captured from the outer two holes are brought into the own
(front) row from the side where the capture was made. If the middle four holes are
captured, the player is allowed to choose from which far end he wishes to enter the
opponent's seeds into his own row. If a capture is made for the second (or more)
time in one turn, then the direction is set: the player plays into the same direction as
in his last capture, or if he captures one of the outer holes, he plays the direction
from the side of the outer holes.
Part II
When all twenty-two seeds have entered the game, each hole with more than one
seed (even from the back row) can be spread in such a way that it reaches an
occupied hole in the front row where it can capture. If this is not possible directly,
then a hole from the front row is spread in a random direction without capturing.
Ones (singletons) cannot be played in this manner. If the front row contains only
singletons, then the player has to play his back row. If the back row also contains
only singletons, the game is lost.
B. Tactical combinations
There are three kinds of tactical movements illustrated in the following paragraphs:
1. Tactical combinations: these moves illustrate the important situations in the
game that have to be recognized. I will limit this discussion to forced moves.
Forced moves are situations where the player is forced to accept the course of
play, i.e. the opponent is able to reach his strategic objective without possible
counterplay.
2. General tactical moves: generalisations about useful tactical moves.
3. Short wins: series of moves that end the game very quickly and need to be
recognized, since they are easily overlooked.
87654321
b 00000000 b
right a 02260000 a left
left A 00006220 A right
B 00000000 B
12345678
The system of indicating moves has certain redundancies that are sometimes
eliminated for reasons of brevity. For example:
1. When capturing the holes A1, A2, A7, and A8 (similar for the a-row) they do not
need the addition left and right, since the direction is already determined by the
rules.
2. The letters a/A do not have to be mentioned in the first 22 moves, since there is
no option of playing from the back row.
3. A forced move, i.e. a move where there is only one obligatory capture possible
(from only one direction) is not necessary to mention at all (or just the direction),
however, eliminating this redundancy adds little to the clarity or brevity of the
notation.
4. In the transcription of the Tournament games, there is a redundancy added,
namely takasa and takasia moves are indicated with (tak) and (takasia),
respectively. This helps the interested reader to re-play the game, and double-
check the correctness of the moves.
Each number indicates the number of example in the diagram. Na5 has 6
seeds in a hole. An x indicates that seeds and Na6 and Na7 have 2 seeds
the number of seeds is unspecified, each.
but more than 6 and will often be
used for the house. 1. Combinations to the house:
short flank
Only the relevant rows are shown:
Example 2: 'A' moves A8 The house is central in the first part of
i. the Bao game. Most combinations
a 000x0001 seek to develop and empty the house
A 0000x001 most profitably. In order to acquire
skill in the defence and development of
ii. He enters a seed: the house one is required to know the
a 000x0001 various combinations that empty a
A 0000x002 house, since it cannot be emptied
without a preceding capture.
iii He captures: The short flank is the right side of
a OOOxOOOO the house (three holes), the long flank
A 0000x003 is on its left (four holes).
The examples shown in the
iv He continues: combinations to the house (examples 1
a OOOxOOOO and 2) consist of one move only. The
A 0000x110 separate steps of the move are
illustrated (i, ii, iii, etc.) for greater
v He reaches the house: clarity.
a OOOxOOOO
A 000011110 Example 2.i to 2.vi
vii.
a. OOOxOOOO OOOxOOOO
A 031lx000 301lx000
!!!l'l!Z1kS:~--~
<C.. -~;.,:;,. .,,,,_.;:, '-. ~'~·Mr•"'"'\'1:fl~~_i~·:r.'ljl\"-'fi.~.A (•~;;.,:';,·.~~-~'.;\tlt!t..:...,._~:,~,. ....,~~o<#i,.,.;~-.,,..,._~/C:.~.i*~~~~.~-0 '();''~",':..1'•':'.'-"••',.--'""·'J!;,.,.'i-'J'"<""·""···~·~-: • ' •"r-to'-4;~ . .,,,_......, •·-
choice 1S 1 1 1
1
1
I
21
1
21 2I
I 21
I
OIOX0411 010X1S30 010XOS30 100XOS30 OOOXOS30 OOOXOS31 OOOXOS31 11IX1040 111X0040 I11X004I lmX0001
0020X011 0020XOOO 1020XOOO 1020XOOO 0131XOOO 0130XOOO OIOIX110 OI01X010 0101X011 0100X011 12110I22
choice 16 1 1 1 1 I 21 21 2I 21
010X0411 010X1S30 010XOS30 100XOS30 OOOXOS30 OOOXOS31 OOOXOS31 111X1040 I11X0040 111X0041 111X0040
0020X011 0020XOOO 1020XOOO 1020XOOO 0131XOOO 0130XOOO 0101X110 0101XOIO 0101X011 0100X011 12110120
choice 17 1 1 1 1 21 21 21 21
010X0411 010X1S30 010XOS30 100XOS30 OOOXOS30 OOOXOS31 OOOXOS31 111X1040 111X0040 020X1040 i 020XOOOO
0020X011 0020XOOO 1020XOOO 1020XOOO 0131XOOO 0130XOOO 0101X110 0101X010 0101X011 0100X011 i 12110230
choice 18 1 1 1 1 21 2I 21 21
OIOX0411 010X1S30 OIOXOS30 100XOS30 OOOXOS30 OOOXOS31 OOOXOS31 111X1040 I11X0040 020X1040 OOOX0040
0020X011 0020XOOO 1020XOOO 1020XOOO OI31XOOO 0130XOOO 0101X110 0101X010 OIOIX011 0100X011 OI020122
choice 19 1 1 1 I 2I 21 2I 21
010X0411 010X1S30 010XOS30 100XOS30 OOOXOS30 OOOXOS31 OOOXOS31 111X1040 111X0040 030XI040 030XOOOO
0020X011 0020XOOO 1020XOOO 1020XOOO 0131XOOO 0130XOOO 0101X110 0101X010 0101X011 OOOIX011 11120230
choice 20 010X0411 010X1S20 010XOS20 010XOS31 010XOS30
0020X011 0020X010 0020X011 0020X001 11310110
move2 move3 move4 moveS move6 move 7
choice 21 010X0411 010X1S30 010XOS30 121X1030 021X0030 130Xl130 130X0130 041X0130 001X0130 101X0130 OOOX0130
0020X011 0020XOOO 1020XOOO 1020XOOO 0131XOOO 0101XOOO 0101X001 0100X001 1311X001 1310X001 0431XOOI
moveS move9
100X0130 100X0130 011X0130 OllXOOOO
0430X001 040lX111~ ~00)~111 1Sl10340 ---- ----- -- ----- ----- ---- -
~·"'
62 LIMITS OF THE MIND
1. The bridge
The bridge is a choice of capture, where a captured seed is entered into the front row
in such a way that it reaches the initial hole of capture again. This hole I will call the
capturing hole. This procedure enables the capturing hole to continue while at least
two seeds have been added from the start of the move.
The bridge in Bao is an useful tool for reaching the other side of the board, in
order to launch an attack, continue an attack, empty the house, and/or develop the
back row.
The bridge has already been seen, in the previous examples' on the combinations
to the house. The capturing hole was the house, the captured seeds are entered in
such a way that they reach the house again and enable it to be emptied.
In a situation where many capturing possibilities exist, the bridge may function as
a tool with which both sides (flanks) of the front row can be captured. As soon as the
outer two holes are reached, the playing direction changes and continues the attack.
Since it is rare to capture seven seeds from one hole, which would enable the player
to reach the outer holes on the other side of the board at once, it is wise to look for a
bridge that continues your move and lengthens the attack.
long combinations. Many times the holes with many seeds end somewhere in the
back row and cause only limited damage compared with other combinations.
A short win is a game with less than 22 moves (44 turns). In that case, not all the
seeds have entered the game and a player is beaten mkononi which means 'in the
hand' (which refers to the seeds still held in the hand). I show the following wins
not only to illustrate other parts of tactics but also to make the reader recognize these
particular mistakes at the beginning of the game that lead to immediate defeat.
A Baa player needs a certain basic knowledge of openings in order to avoid short
wins. These traps (mtego) are considered general knowledge and can be compared
to the fool's mate or the scholar's mate in Chess play. They illustrate little tactical or
strategical knowledge and usually consist of one or more blunders, often on both
sides.
Opening moves in a strategic sense are illustrated in a later Chapter. The
following examples should be considered short term opening knowledge, or basic
knowledge for the average player. There are many examples known in Bao. There
are two examples given here. The first is the shortest game possible in Baa. The
second is also very short and ends with a short series of forced moves. Both
examples are instructive for the purposes of tactics. Each move is illustrated with
the situation after the move drawn in numbers. The reader who is interested in
learning how to play Baa will find these useful examples to understand the
notational system and the working of the moves.
Example 11
Short Baa 1 is extremely instructive. Many general principles are violated in only
six turns. Player 'a' abandons the house (c), a questionable move, usually not very
profitable. However, player 'A' allows an attack of only singletons by way of a
bridge. Player 'A' also develops his back row in a stage where he should neutralize
the attack. 'Greed' seems to be the human factor with which 'A' was lured and
tricked. 'A' captured the house of the opponent and wanted to secure his own house
by playing it into the back row. He also entered the captured seeds from the 'wrong'
end (d). Although fewer holes are attacked with this last move, it results in a
convenient bridge (a6) from which the opponent can launch the attack. 'A' captured
an almost maximum number of seeds in each move, this principle was, however,
completely invalid in his situation.
Short Baa 1:
a) 'A' has the move b) 'A' playsA6L
a 02260000 a 02260000
A 00006220 A 00117020
64 LIMITS OF TilE MIND
Example 12
The second example of a short win in Bao concerns a series of moves, where 'A'
follows a common opening until he falls into a trap.
Move (h) looks like a trap (mtego) by 'A', since it introduces a forced move.
This move leads, however, to a counter-trap (tegua) by 'a' in (i). From move (j)
onward, the series of moves are forced, i.e. only if 'A' does not want 'a' to empty his
house, and very similar to the examples given in the section on tactical
combinations.
Eventually, the opponent 'a' is allowed to empty his house, which, if he does,
leads to victory immediately. Instead of a short win, we should perhaps speak of a
extended opening analysis, with a note that a mistake leads to disaster in this
particular case. Note that 'a' has played aSR five times in only seven moves.
Only the end result is shown in (o) which complicates the deduction of the course
of the move. The move, in this case, ends in b8 just after the house of AS was taken.
ShortBao 2:
a) 'A' has the move b) 'A' plays A6L
a 02260000 a 02260000
A 00006220 A 00117020
I
.I
I
. '
VI: MASTERSHIP IN BAO IN TERMS OF EXPERIMENTS
De Groot (1946) designed a test that distinguished experts from novices in Chess.
Players were shown a genuine Chess position on a Chess board for about five
seconds. They then had to reconstruct the Chess position on another board
sometimes giving their opinion on the best possible move. On this five second test
grandmasters performed flawlessly, while masters made small mistakes, and lesser
players more serious mistakes, giving an accurate rating of the players.
When this test was performed with a random positioning of the Chess pieces,
creating an unreal Chess position, both expert and novice performed equally badly.
This showed that Chess experts have a specific memory for Chess, one that does not
necessarily transfer well to other situations.
This experiment was translated for Bao. Bao players were presented with a Bao
board and asked to remember the position and, if possible, think of a good move. ·{
They did not perform as well as expected. Even after exposing the Bao position for
about 30 seconds, the grandmaster reconstructed the position poorly; in fact, slightly
worse than my own performance on this test.
Bao consists of four rows of eight holes with various numbers of seeds in each
hole. This is similar to a four by eight matrix represented by.numbers. Such a
matrix was shown to two famous mnemonic experts (Luria, 1968; Hunt & Love,
1972). Hunt & Love (1972) showed their informant VP matrices consisting of eight
rows of six numbers. VP's study time was 246 seconds, however his recall was
perfect, even when rows were asked diagonally or backwards. Luria's (1968:17)
matrix-test was more complex and Hunt & Love repeated this experiment with VP.
Both Luria's subject and VP studied the matrix of digits (4 rows of 13 digits) for 3.0
and 6.5 minutes, respectively. Again, both mnemonic experts had perfect recall.
Apparently, it is perfectly possible to perform well on a matrix task. It is not known
how mnemonic experts would perform on a Bao matrix; however, it is very likely
that they will perform better than Bao experts. This gives reason to believe that
memory for matrices has very little to do with Bao expertise.
Since exposure to Bao positions provides little insight into expertise in Bao, a
duru experiment was designed. Before describing the exact nature of the duru
experiment, we should consider the various related experiments in the literature.
If we consider duru as a test where various holes are remembered in succession,
then Atkinson & Shiffrin (1968) should suggest the possible result. Their
experiment, slightly modified, was repeated for the mnemonic expert VP. In this
case, pairs of nonsense CVC syllables and two digit numbers· were presented. On a
test trial the CVC syllable was presented and VP was required to indicate the last
MASTERSHIP IN BAO IN TERMS OF EXPERIMENTS 67
number paired with this eve syllable. The number of pairs intervening between the
test trial and the CVC syllable in question was defined as lag. Hunt & Love also
studied the response time and found that VP was superior to the subjects of Atkinson
and Shiffrin, both in response time and in number of lags. When the pairs were
presented for no longer than 3 seconds, VP had a maximum lag of 18.
Atkinson and Shiffrin's subjects, who were much more practised than VP or
our own control subjects, were responding at chance level at this lag.
Hunt & Love (1972:242)
Other experiments have an even closer resemblance with the test of durn, especially
the various digit span experiments, that also use numbers instead of syllables.
According to Miller (1956), people have a limited short-term memory space for
random data. This constraint on memory is better known as the magical number
seven plus or minus two. For example, we remember about eight random digits if
these digits are presented at a rate of about one digit per second. However, this digit
span for people can be improved. Hunt and Love (1972) made some progress with
their subject, and Chase & Ericsson (1981) then shattered this constraint with a two-
year training of an average college student, who subsequently recalled 80 random
digits presented at a rapid rate.
The answer to this contradiction in the literature is chunking. Digits are grouped
or recoded and, therefore, chunks are created, the number of which does not exceed
the magical number seven that still constrains short-term memory. Newell & Simon
(1972) presented this argument convincingly for the memory of Chess positions as
well:
Finally, the number of chunks retained in short-term memory after brief
exposure to chess positions is about of the magnitude we would predict from
!
J
immediate recall of common words (Miller, 1956) and copying of visual
patterns (Ein-Dor, 1971).
Newell & Simon (1972:402)
This memory space for random data is an important constraint in the tasks described
t above, however, it can be by-passed by various chunking techniques.
I
Rehearsal is seen as an important way of keeping information in (short-term)
memory. When rehearsal is made impossible by interference tasks (Peterson &
Peterson, 1959), correct recall drops significantly to about 55% after an interval of
15 seconds. The interference from previous items in the test is said to be the main
cause for this rapid decline in recall. Again, Hunt & Love (1972) show that their
mnemonic expert is able to minimise this interferem;e by chunking the 3-consonant
trigrams that were used in the experiment.
Rehearsal seems to be important, but mainly to eliminate possible interference
with other items. This interference can then be by-passed with chunking-techniques
in the case of experts (see Simon, 1979). 'i
I I
In sum, in all the experiments described above, subjects store increasing amounts I
II
68 LIMITS OF THE MIND
masters are concerned (Allis et al., 1990). Modern psychological works on expertise
have not focussed on master memory skills, with the possible exception of Chi,
Glaser & Farr (1988). Rather, research has followed more traditional paths with
works on genius in general (Simonton, 1984; Ochse, 1990) - a topic which dates
back to Galton (1892) - and works with a focus on clinical (Howe, 1989) or
developmental (Retschitzki, 1990) aspects of mastership, as was outlined before in
reference to Reysset (1995). Their issues are less relevant to the argument made in
this Chapter.
Here, the question of what a master or genius actually is, is largely by-passed by
describing mastership in terms of mental limitations. This contrasts with common
views that mastership shows extraordinary capabilities by extraordinary people.
Instead, masters show the area in which the limitations of the human mind matter.
Masters are only extraordinary in reaching that area first. Philidor's mastership in
Chess is now inferior to that of a modern 'regular' master. The limitations of the
mind are not likely to have changed since Philidor. Problem solving techniques and
ready knowledge have changed dramatically, however. Since Philidor (Hooper &
Whyld, 1992) the amount of literature on modern Chess has increased rapidly.
Studying Chess literature has become obligatory in the development of Chess
masters. Apparently, some of the limitations of the mind can be overcome by
studying books or learning by heart, i.e. ready knowledge held in long-term memory
which is not subject to the constraints exercised by the limitations from short-term
memory (Kennedy, 1975; Eysenck & Keane, 1990).
Experiments on Chess and findings in computer science studies in general have
focussed on the body of data available to players. Literature on memory has focussed
on data storage. Even the more recent theoretical developments in the study of
human memory, i.e. the introduction (McClelland & Rumelhart and the PDP
research group, 1986) and subsequent developments of the theory of parallel
distributed processing are significantly concerned with data storage. Since there are
no books on Bao games apart from this study, the Bao master will be less concerned
about his data storage. .
In the following. experiments the allocation of information in the memory system
appears essential. When describing the problem solving model in the duru
experiment another limitation of the mind presented itself: the dynamic constraint.
2. Duru
Duru is a Swahili term meaning round(s). In a Bao game, a.move may consist of
many rounds, i.e. the seeds may spread through a player's own rows making several
laps. A player who has duru is a player who can calculate many rounds.
MASTIRSHIP IN BAO IN TIRMS OF EXPERIMENTS 69
There are different ways of knowing duru. Moves with many captures are
complicated to calculate and are referred to as complex duru in this study. A simple
move to calculate is one that goes round within one's own rows without capturing
(takasa). A master may calculate four or five duru in this way, a minor player two or
three. Still, if a player reaches only three duru in such a move, but can calculate this
in his own rows and subsequently in the rows of the opponent, i.e. calculating two
long duru in a row, then his minor ability to calculate duru may prove sufficient
against many opponents.
3. Takasa
For our experiment we chose the move of takasa, where fewer factors influence the
performance of the informants.
Takasa is a move where one picks up the seeds from a front-row hole and spreads
them one by one into consecutive holes in an (anti)clockwise direction. If the last
seed falls into an empty hole the move stops, but if it falls into an occupied hole all
the seeds in that hole are picked up again and spread in each consecutive hole in the
same direction until the last seed finally falls into an empty hole. It is possible that
such a move does not end at all. For example:
a. 01010131
10101010
Move the hole of three seeds in (anti)clockwise direction and continue spreading.
The last seed will never fall into an empty hole.
b. 01010151
10101010
Move the hole of five seeds in (anti)clockwise direction, etc.
c. 23232323
32323232
Move any hole in (anti)clockwise direction, etc.
The variable length of takasa moves forms the basis of the duru experiment.
4. Experimental design
The duru experiment consists of a calculation of a takasa move, i.e. a move without
capturing and played in one's own row(s). The position was set up in such a way that
the move ended beyond the calculation ability of the players. In a pilot test, it
appeared that players reached no more than four or five duru and rarely seven duru.
Therefore, the conditions provided them with moves continuing for eight or nine
rounds.
Apart from measuring the number of duru we also wanted insight in the kind of
duru. It is possible to design a condition where many holes are emptied (stations) in
relatively few duru, as in examples (a) and (b) above. In a similar way it is possible
to design moves that have few holes that are emptied but involve making many
rounds. This often happens when big holes (like the house) are emptied together with
other big holes. The conditions designed for this purpose have a significant
difference between the number of holes that were emptied, called 'stations', and the
number of holes that were passed, called 'length'.
In order to keep other variables from interfering,- the conditions were
standardised. The number of seeds involved in the move was set around 40. The hole
70 LIMITS OF THE MIND
from which the move was to be calculated was always AS (the house). And the
direction was always a clockwise direction. Also there were no patterns that repeated
themselves in a predictable way, as in examples (a) and (b) above.
It was not possible to have informants think out loud; the participants refused or
said it was impossible, there was neither a language nor a notational system to which
they could refer. Instead, the informants were asked to put their index finger at each
hole they reached in their calculations. This request was less problematic and also
reasonably accurate when a video camera recorded their finger movements. A few
informants said that they initially became confused when using their finger, but they
quickly adapted.
Using a video camera had several advantages. The length and the number of
stations could be recorded and also the exact place where informants made mistakes.
Most importantly, the factor time could be measured, i.e. the thinking time for each
hole, for each round, and for each complete condition.
5. Experimental situation
Each informant was invited to a relatively quiet place with no spectators (often
problematic in Zanzibar). Many times we started with a short interview recorded on
video in order to get some background information on their personal life, and to
make them accustomed to the presence of a camera.
The informant was presented with a Bao board and both my helper, Abdu, and I
would explain the experiments in the local language, Swahili. The question was to
tell us, by calculating and not by actually playing, the hole where the seeds would
end when leaving from AS, all the while indicating his progress with his index
finger.
We started with a neutral condition to help him get used to the obligatory finger
movements. This first condition was seven duru long (and was completed by one of
my talented informants in his first effort). Subsequently, he was presented with three
more conditions and, later, asked if he found them more or less difficult. The video
camera recorded the finger movements for each condition. After four conditions, the
informants usually stopped, before getting tired and confused by the intensive
calculations. Therefore, I limited the duru conditions to four only. Two informants
got tired after only two co~ditions, but they were too dissatisfied with their
performance. They also complained about having to use their finger during the
calculations. This supposedly influenced their results. All other informants claimed
this to be a insignificant adjustment.
After each condition the informants were allowed to play the move, i.e. spreading
the seeds on the board, and to explain where their mistakes were made. Surprisingly,
they did not always identify their mistakes at the right place, even after playing the
seeds.
Condition A:
87654321
b 00000000 b
right a 00000000 a left
left A 32414211 A right
B 22081433 B
12345678
Characteristics of condition A:
Task: calculate A5R
Set up: 41 seeds in two rows
Length of move: 89 holes
Number of stations: 18
Average number of holes between stations or holes per station: 5-6
Condition B:
12345678
Characteristics of condition B:
Task: calculate A5R
Set up: 41 seeds in two rows
Length of move: 135 holes
Number of stations: 28 Holes per station: 4-5
Condition C:
12345678
Characteristics of condition C: i
Task: calculate A5R I·
Set up: 48 seeds in two rows i,
Length of move: 125 holes '!I
It<'
Number of stations: 22 Holes per station: 5-6
'I,
Condition D: 'i' '
i '
12345678
72 LIMITS OF TilE MIND
Characteristics of condition D:
Task: calculate ASR
Set up: 30 seeds in two rows
Length of move: 109 holes
Number of stations: 37 Holes per station: 2-3
1. Time tables
The performance of the informants on the simple duru experiment, as opposed to
complex duru discussed in a later Chapter, are presented in four tables (fables 4, 5,
6, and 7). They are followed by a number of graphs that show the exponential
increase in thinking-time per station. In the following section, these data will be
discussed systematically together with the presentation of additional conditions that
strengthen the arguments presented.
X : this symbol indicates that the informant made a mistake in his calculation. If this mistake is not
corrected then the time at which the mistake was made, is taken as the end result of his performance.
- : this symbol indicates that the informant gave up. Usually, the informant becomes tired or 'mixed
up' with numbers and states that he cannot proceed any further.
: this symbol indicates that the informant skipped this station with his finger and immediately
proceeded to the next station. Therefore, the time for this station could not be recorded.
Note the differences in thinking time between AMF and MJS in the fifth round.
MJS forgot his finger at time 388s, when he found the next station after only 1
second. This resulted in a less fluent graph of his data.
74 LIMITS OF TIIE MIND
ISM had the habit of giving up after one minute of thinking. Despite his old age,
his calculation was very fast, but his stamina did not exceed one minute. It appears
that one minute is a reasonable thinking time in Bao matches (see Appendix II &
III). Seven minutes of calculating appeared to be the limit for both AMF and MJS.
If a player uses more than one minute for calculating a move, he will be in trouble if
he wants to calculate subsequent moves within, for instance, a three-minute time
limit in championships.
X : this symbol indicates that the informant made a mistake in his calculation. If this mistake is not
corrected then the time at which the mistake was made, is taken as the end result of his performance.
- : this symbol indicates that the informant gave up. Usually, the informant becomes tired or 'mixed
up' with numbers and states that he cannot proceed any further.
- : this symbol indicates that the informant skipped this station with his finger and immediately
proceeded to the next station. Therefore, the time for this station could not be recorded.
• See also condition A for comments on NOA.
MASTERSHIP IN BAO IN TERMS OF EXPERIMENTS 75
-'
t:
Table 7: Time table of condition D
Cnd.D ISM I ROR I AMF I MJS I MKM I NOA SMS JAN Rnds.
Holes Total time needed to calCulate stations.
per Best attempt:
station SECONDS
2 1 1 1 1 1 1 1 1 1
2 2 3 1.5 2 2 2 2 1.5 1
3 3 6 2 3 3 6 3 2 1
3 5 8 2.5 7 4 10 5 2.5 1
5 7 13 3 32 4.5 13 7 3 1
2 9 21 3.5 36 5 17 8 4 1-2
2 11 25 4 46 6 20 10 5 2
2 13 40 4.5 51 7 24 11 6 2
4 17 48 5 61 8 - 12 7 2
4 19 64 5.5 70 11 56 14 8 2
3 34 74 6 75 12 70 16 8.5 3
2 44 85 7 83 16 79 18 9 3
2 49 108 8 87 17 85 24 10 3
2 55 123 9 110 19 114 26 11 3
5 58X 139 14 115 22 125 36 12 3
6 160- 16 153 45X 162 40 13 3-4
2 17 202 176 50 14 4
2 18 218 193 56 15 4
2 21 255 204 66 16 4
3 22 290 236 71X 17 4
3 23 384 256 19 4
2 35 477 290X 20 4
3 41 534 21 4-5
2 57 666-- 22 5
2 61 23 5
2 83 100 5
6 110- 139 5
9 205- 5-6
2 6
4 6
2 6
2 -- f--- 6-7 --
3 7
2 7
2 7
2 7
3 7
76 LIMITS OF TilE MIND
2,5 + _111 I 3
I
~
g
ij"'
2,5 (/)
~
• 3 2
•
j 1~21 ,.,.,
predicted Y predicted Y
I ---o- 101og (t)
I>C
0
s 1,5 -o--- lOiog (t)
ztx1
_,.,. >
0
0.5 j
1
,.- I( z
g
0,5
0 0
• ~
(/)
0 4 8 12 16 20 0 4 8 12 16 20 0
":d
stations stations 1:!:1
~
1:!:1
2,5 2,5
~
.._,
;o
2 I • predicted Y .._,
0
I>C
2 ----predicted Y
0
§ 1,5 - - - D - lOJog (t) 8 1,5 ---o- 10log (t)
1 1
0,5 0,5
0 0
I -..,J
-..,J
0 4 8 12 16 20 24 0 4 8 12 16 20 24
stations I stations
l ' ~
Abdu A predicted Y Abdu B predicted Y
s~~
-...l
00
/
0,5
0+---+----t----+---t--~
4 8 12 0 5 10 15 20 25
stations stations
2,5
2
s
2
tl- predicted Y 3 1,5 t1- predicted Y
~
~ 1,5 ~ ~
8 \ - - o - 10log (t) ~ - - o - lOlog (t) 0
":!
1
0,5
~
@
0+------r------~----~
0 4 8 12 16 20 0 10 20 30
stations stations
I I
I
MASTERSHIP IN BAO IN 1ERMS OF EXPERIMENTS 79
The predicted Y and the 10log of the condition's time axis are plotted together (see
graphs). Only Abdu and Majaliwa are shown since they performed well and
provided accurate data for all four conditions. Condition D, where Majaliwa rested
his finger for 25 seconds at a hole in the first round, shows clearly that the influence
of such a discrepancy in the analysis of the graph is considerable and easily detected
as an error. The same experiment was repeated a few months later, and then this
discrepancy did not occur. I have included the data from his first session only. I
included the results of all the other informants on condition D for further
comparison. This condition presents a significant number of measurement points,
i.e. more than 15 stations, which I used as a minimum for the R2 calculations. The
R2 values are sometimes better when the 1°log of the time axis is not taken, as in the
case of Nasoro and Njowine. In the case of Nasoro, however, the standard error
would be 38.01, destroying the statistical significance of the number. In the case of
0
Njowine, the values without taking the 10log are near perfect, since only the last three
M time measurements in the experiment showed some exponential increase. Since, the
R 2 of Njowine's data is still 0.880, it is safe to conclude that there is a relationship
between the time increase with each station and an exponential increase. We should,
0
therefore, explain this element of the experiments when designing a thinking model.
01
~~
U>
c 3. Results
0
'.;:J
.'9 Only one master solved condition A. None of the masters was able to produce
'"
-
0 perfect results on all conditions. Apparently it is not possible, even after several
attempts, to calculate an unlimited number of rounds.
There is a difference between expert and novice in Bao, and there is a difference
between expert and novice in duro. An expert in Bao is usually able to calculate 4 or
0 5 rounds (between 65-80 holes), a novice can hardly calculate 3. An expert in Bao,
specialised in duru, is able to continue for 6 or 7 rounds.
There are three ways of measuring the performance on the experiment. One way
is used by the Bao players themselves and measures the number of rounds. There is
usually a margin of one or two rounds in different conditions. Another way consists
of calculating the number of stations or measuring points. However, this also varies
with each condition. A third way would be calculating the number of holes that
were passed, which is simply the number of rounds multiplied by 16.
There is no simple constant that indicates a master's performance on a test. Too
many conditions would be needed, and all on the same day, since a performance
varies slightly each day, in order to extract a constant. The indication used by players
seems most convenient and reasonably accurate.
i
I I
80 LIMITS OF TilE MIND
This model would be similar to giving a person sixteen numbers, one per second,
and chunking them all. If we know the allocation model procedure, we can start from
hole number one and proceed as follows:
Station 1: (we see) 4 seeds >next hole 4 holes further
remember: this place was emptied > one chunk
Station 2: (we see) 3 seeds, add one,
remember: this place was emptied > second chunk
Station 3: (we see) 8 seeds, add one,
remember: this place was emptied > third chunk
If the rounds did not overlap, it would not be necessary to
remember the place that was emptied and we would be able to
continue calculating for ever.
l Station 4: We enter the second round > add 1 extra to each following station
f > fourth chunk
(we see) 2 seeds, add one, add one extra, and remember this place
> fifth chunk
Station 5: 3 seeds, add one, one extra, and remember this place
> sixth chunk :f
Station 6: 2 seeds, add one, one extra, and remember this place
> seventh chunk
Station 7: 4 seeds, add one, one extra, and remember this place
> eighth chunk I.
This would be an average performance for a novice. ·I
'1.1
,I
1-,.
Station 8: We enter the third round > add 2 extra to each following
,j'!I
station 1:1
Station 15: no seeds, add one, four extra. Station 15 has not been emptied.
We used seven chunks.
Station 16: 4 seeds, add one, four extra > first chunk: fourth round
> second chunk: station 16
> third chunk: station 8
We chunk round one, two, three, and four in the process. Station 16
has been emptied before in round 2 as station 7. We have to
subtract 6 seeds, since there were 6 seeds in this station in the
second round. We keep station 16 chunked. We used seven chunks,
if we disregard the chunks needed for some simple mental
calculation 4 +1 + 4- 6 =3.
I Station 17: We enter a new round. 2 seeds, add one, add five,
'I: five extra > first chunk
iI station 8 > second chunk
I station 16 > third chunk
station 17 > fourth chuilk
We chunk round one, two, three, four, and five in the process.
Station 17 has been emptied in round 2. We have to subtract 4,
since there were 4 seeds in this station in the second round. We
keep station 17 also chunked.
We used nine chunks. .
:! Station 18: We will need more than nine chunks to calculate station 18. Only
two duru experts were able to calculate station 18.
MASTERSHIP IN BAO IN TERMS OF EXPERIMENTS 83
The model presented above raises a number of questions. Firstly, the chunking
process appears relatively simple. Apart from a limited number of chunks that is
needed, it also appears unnecessary to have a large body of knowledge about Baa
positions. In experiments on Chess masters and other experiments chunking consists
of creating larger pieces of information with the help of experience.
Secondly, it appears that chunking into larger pieces of information is not a
known procedure in calculating duru. There is the following evidence for this
assumption:
1. Informants never recognised a pattern, even if exactly the same pattern was
presented to them a little while later. Certain tests were rewritten, using the same
pattern without the first step. For example:
a. 31412010
34212222
b. 41012022
13342122
The above examples differ in only three holes, it is as if the first part of the move
ASR in (a) has already been played in (b) (the holes being moved three holes in anti-
clockwise direction). None of the masters recognised or performed better when the
test was presented again in this slightly different form, although the pattern was
exactly the same.
2. Informants repeated calculations, i.e. they started the condition from the
beginning, but they also started the calculation from the beginning. The calculation
as a whole would go faster, but each station still took more and more time when
proceeding. In condition D, Abdu's graph represents his last attempt after starting
over and over about five times. If the pattern were to be remembered we would
expect a more linear graph, since there would be no reason to calculate.
3. Patterns are unpredictable in Bao. A pattern may follow another pattern for many
moves and after several laps it may suddenly change its course. If players play an
opening, they do not remember the pattern of a certain move, but the changes after
the move, which is, of course, more useful information.
Thirdly, relating to the simplicity of the process, there seems no particular reason
why a novice should perform so badly on this test. Even the performance of a master
hardly seems impressive. The model only showed some complexity when entering
the sixth round. I will argue that there is an extra constraint that limits the
performances of the masters on this condition.
The situation, the set-up, and design of the experiment is similar to that of simple
duru. This time two rows of 4 holes were used, where short rounds were to be
calculated as in a takasa move.
The crucial element in this experiment is that the number of holes is reduced to
eight, now within the limits of Miller's constraint. So, we should find an almost
perfect performance on the short duru test. The allocation model introduced to
explain the results of the previous experiment predicts a worse performance, because
of the increasing number of changed stations. The short duru experiment should
distinguish these contradictory predictions.
Condition P: ConditionS:
4012 2321
0116 0232
Characteristics of condition P: Characteristics of conditionS:
Task: calculate A1R Task: calculate A1R
Set up: 15 seeds in 2 rows of 4 holes Set up: 15 seeds in 2 rows of 4 holes
Length of move: 87 holes Length of move: 71 holes
Number of stations: 25 Number of stations: 21
Condition Q: Condition T:
2123 2213
3310 0214
Characteristics of condition Q: Characteristics of condition T:
Task: calculate A1R Task: calculate A1R
Set up: 15 seeds in 2 rows of 4 holes Set up: 15 seeds in 2 rows of 4 holes
Length of move: 72 holes Length of move: 71 holes
Number of stations: 22 Number of stations: 22
Condition R:
4020
4014
Characteristics of condition R:
Task: calculate A1R
Set up: 15 seeds in 2 rows of 4 holes
Length of move: 65 holes
Number of stations: 19
The results of this experiment are shown in the tables below (see also Appendix I).
Again, duru experts, those who claim to specialize in calculating duru, performed
better than regular masters with a more basic knowledge. It appeared very difficult
to calculate short rounds. Not only were fewer stations_calculated, but there was also
a shorter length calculated. Only the number of rounds seemed to remain constant.
MJS and AMF, duru experts, reached 6-7 rounds. The others reached 4-5 rounds. I
) MASTERSHIP IN BAO IN TERMS OF EXPERIMENTS 85
e included some tests with early mistakes by NOA and SMS to illustrate the
e fluctuating performances of the informants. Therefore, as will be shown later, it is
difficult to collect comparable results.
a The results of this experiment were preceded by a simple duru test in order to
:t assess their form of the day. This made sure that their average performances make a
J reasonable comparison to their earlier simple duru performances, despite the
e ...'1· different days on which the conditions were conducted.
,-
j All masters agreed that this test was more difficult and that masters, especially
~-f
:::j,
duru experts, should perform better on this test than novices in Bao. One test, which
J· was briefly tried, appeared practically impossible for both master and duru expert:
e
J
'i Condition V:
i
r 27104
i 6453
Task: calculate A1R
I:
: i,
.,J
~[
,:fi,,
~·
I
Set-up: 41 seeds in 2 rows of 4 holes
This test was tried by Majaliwa who had to give up after only a few stations. Others
did not even consider a serious attempt. The unanimous commentary was that the
tr
:~j rapid changes in the pattern make condition V a much more difficult round to
calculate.
:i
,, Below the results of condition P are shown in Table 9, indicating the number of
'~ holes, the number of rounds, and the number of stations, that the informants
! jj{
rn
fi.
calculated in the condition. The tables of conditions Q, R, S, and T are presented in
Appendix I.
Table 9: Condition P
m~~
tf Cnd.P Round Station ROR MJS SMS NOA Exp.P* Round AMF
'.~ 4 1 1 7 1
1
~~
7 1-2 2 4 1-2
-1
\ 4 2 3 2 2
:·~ 2 2-3 4 3 3
');
,, 3 3 5 2 3
·f 2 3 6 4 3
' ~l 4 3-4 7 X 5 3-4
'.{ 4 4
I~ 5 8 2
j
'f 2 4-5 9 X 2 4
It 2 5 10 3 4-5
'if.
l 3
2
5
6
11
12
X 2
5
5
5-6 X
·J
-~
5 6 13 X 7 7
7 6-7 14 4 7
). ,<
:d ~ 4 8 15
.It ~
:o ll' • This condition was rewritten with comparable results. The pattern is identical apart from the first station
that was omitted.
tt. ;, '
_,.
I
'' i!
ll,-
·~
•l
tt
·~:f
86 LIMITS OF THE M1ND
\ .·''
_,, '
. ', _,l,-
D MASTERSJITP IN BAO IN TERMS OF EXPERIMENTS 87
d
te
'Y
t-
·e
:e
:o
!I
is
i
b
~
~
Ir
~';
~~
VII : THE BLIND PLAY EXPERIMENT
The blind play experiment shows that considerable changes in memory information
take place over a longer period of time without the influence of the dynamic
constraint. The changes in the position of such a blind game are so manifold that
they easily exceed the changes in a duru experiment. It is argued by our informant
and in previous research (Newell & Simon, 1972) that series of moves are chunked
into large pieces of information. The changes are, therefore, related to stored
information and reduce the dynamic constraint to such an extent that a perfect
memory performance is possible in both blind Chess and Bao games.
However, within one move the informant is still not able to calculate more than 5
duru. If very long takasa moves take place in a game, it is possible to beat any master
of Bao when they play blind.
A. Playing blind
2. Experimental situation
Abdu practiced blind Bao in three stages. The first attempt was played in July 1992.
Only one short game was played but not written down. In this first stage, there was
THE BLIND PLAY EXPERIMENT 89
no practice on the part of Abdu and a limited number of moves, less than twenty,
were played.
Two years later, on the 12th of June 1994, he made three consecutive attempts,
ation followed by another three times on the 24th, and one on the 25th, which was the
amic most successful game of them all. These games of blind Baa were played in a similar
I that manner to the game in 1992. There are three differences with blind chess that should
mant be noted here:
mked Firstly, the moves were communicated in a different way. The informant was
tared presented with a board without seeds. This board was necessary to point out the
:rfect different moves. I was his opponent and in the possession of a board with seeds that
he was not able to see. A move was pointed out by indicating the hole to be played
h.an5 and the direction to play in. All procedures afterwards are determined by the rules of
taster the game. If the house was emptied this was mentioned too. With this information a
game was played until the informant made a serious mistake or gave up. Sometimes
a small mistake was made, but if the informant quickly recapitulated and corrected
his error, the game was continued. The places where these mistakes occurred were
written down as well.
Secondly, the communication was difficult and slow. Since the moves and the
y and mistakes were written down -there would have been very little to see on a camera -
urs in it took some time for each move to be played. Also, some moves had to be indicated
~cord twice in order to avoid mistakes or confusion. The game was, therefore, much
slower than a regular game of Baa, both players taking extra time.
ieses, Thirdly, since the games were not completed very often and since the position in
es on the game changes rapidly, his mistakes and errors had to be analysed thoroughly.
:.:lling After a serious mistake, he was asked what he remembered of the board in order to
back determine the extent of his confusion. Sometimes, the informant's certainty in the
;vas a middle of the experiment was verified by asking the content of certain holes or the
tarted ending of complicated moves.
1 law The game was begun alternately by the informant and by myself. I tried to play
1w in his least favourite openings, which, coincidentally, gave the best results so far.
j it is In 1995, the third stage, various public performances have been held in Europe.
l.
Abdu did not have an empty board at his disposal and, after some further training,
aying only communicated by voice. None of these performances were written down.
ingin These performa'nces equalled the performance of June 25th, 1994. However, the
mr or quality of the games increased and the use of a board for pointing out moves was no
made longer necessary.
.. The
. The 3. Example of a blind Bao game
. Any The number of position changes in Abdu's blind game of Baa exceeds the number of
position changes he was able to calculate in the duru experiment: 112 (7 rounds =
aying 7 x 16 position changes) versus 277 changes in blind Bao.
with In my opinion, Abdu used his knowledge about Bao in order to calculate these
e first moves. His extensive knowledge about combination moves can explain his
performance. Such information is not useful in the duru experiment. Simple duru in
a blind game could not be calculated beyond five rounds. Apparently, calculating
duru is a separate skill in the game of Bao.
1992.
e was
Tf' '
b 00000000 b
right a 02260000 a left
left A 00006220 A right
B 00000000 B
12345678
1:~
Blind Bao game Alex -Abdu continued:
II
11111111 3 20202020 3 20202020 5
' '
il·
i
01111141 11240001 11322051 J
20202020 16 20202020 4 20202020 3 lj
1
I
09003911 012030t250 1011h10 I
1. Blind duru
In the previous Chapter, I proposed the allocation model as an accurate description
of duru problem solving procedures. At the same time, I introduced the dynamic
constraint as a necessary factor in the description of human memory limitations.
Only this factor can explain the ineffectiveness of the ideal model for duru problem
solving. In order to substantiate the claim that the allocation model is accurate,
another set of experiments was conducted with Bao masters that should give insight
in expert procedures for duru.
The more an experimental condition differs from a real Bao situation the less
likely a Bao expert is to co-operate or to produce results different from a Bao novice.
For this reason certain experiments were impossible to conduct in the Zanzibar
context. The short duru experiment was only conducted with masters who had
become good friends and who were interested in the experiments. Financial
compensation appeared an irrelevant factor at this stage. The experiments described
have, therefore, problematic aspects. The most important aspect is the increasing
number of 'bad' performances. Almost all informants would occasionally make
mistakes very early in the experiment. They were not familiar with the experiment
and were easily confused. The mistakes were usually corrected, and in the blind
experiments below, another experiment had to be set up in order to get reliable data.
The chances of getting comparable experiments became very slim, that is why some
tests were simply presented again in a slightly modified way which would still make
comparison possible. In the end, only the best performances were presented here,
since the object of the experiment was to assess the limits of the mind. The 'bad'
performances were not an object of study in this case.
were reached again in the course of the calculations. As is stated above, the
experiment has a certain disadvantage. If the informant makes a small mistake and
wants to start over, he cannot do so, since the experiment is intended to eliminate the
factor of repeating calculations. Therefore, if a small mistake was made, another
experiment had to be set up, usually a slightly modified condition, in order to
continue.
If we assume our allocation model to be true, then we should be able to predict
the results of the following experiment. The experiment makes it impossible for the
informant to recalculate duru without remembering every hole he emptied. This
recalculation is essential in the allocation model of solving duru problems. We
therefore expect problems to arise around station nine. If the player produces better
results than predicted - this was the reason for introducing the allocation model in
the regular duru experiments - then we should reconsider the allocation model for
experts in Bao.
The following table shows the results of condition A in the blind simple duru
experiment. The remaining results can be found in Appendix I.
Bao masters. All masters have some expertise in this field, as with simple duru, their
96 LIMITS OF THE MIND
skills only differ in degree. The experiment that was conducted on complex duru
illustrates the problems with de Groot's experiments on Chess. Therefore, de Groot's
experiments will be discussed first, followed by a description of the complex duru
experiment.
1. De Groot's experiments
De Groot (1946) studied 22 Chess players, of whom 10 were Chess (grand)masters.
His research involved psychological experiments performed in 1939 and 1943 on
different locations, whenever the opportunity of meeting Chess masters occurred.
De Groot designed two experiments for his informants. One experiment involved
a Chess board with a particular Chess configuration, taken from well-played but
rather obscure matches, which was to be memorised within 5 seconds. Then the
configuration was to be reproduced, sometimes after a short conversation about the
particulars of the Chess position. (Grand)masters and (almost) novices produced
different results in that masters performed almost flawlessly and novices produced
serious errors. In experiments with a random or impossible Chess configuration both
groups performed equally poorly.
Apart from the lack of material to design experiments, there also appeared a
strong cultural problem when performing such experiments as designed by de Groot.
It appeared fruitless, in the context in Zanzibar, to attempt an experiment with which
masters were not already familiar in the game itself. It became necessary to create
experiments with tasks almost identical to tasks found in a game of Bao. This is most
dramatically exemplified in the tasks of thinking out-loud, which informants simply
refused to do. They have no specific language in which they think, since they have
neither a writing system for the game nor literature on its strategy. Some
spontaneous utterances in the game mention numbers, but they can refer to many
different positions on the board in various stages of the moves. Tests that do not
relate to their experience will give unnecessarily poor results.
The complex duru experiment comes closest to de Groot's experiment in that a
certain Bao position has to be remembered in order to perform well. However, the
set-up is identical to a real game situation. The player is asked to undo (rejesha) a
very complicated move that empties the row of the opponent. It is impossible to
undo the move without knowing the starting position. It is also necessary to know
the system of undoing moves and the order of capture. The experiment is, therefore,
certainly more complicated than de Groot's, but it produces similar results. The task
of undoing a complicated move is often asked in play both for educational purposes
and analysis but also in championships when a player is thought to have made an
error. The skill of undoing moves is not essential for a good player, as such, but all
the masters seemed able to perform at a level well above average.
l In the game of Bao there can be many direction changes within one move. We
s assumed that the length of a move and the number of stations, but also the number of
ll direction changes, would influence the player's performance. The rejesha Bao
experiment can be designed in such a way that there is a standard length and a
standard number of stations, but with variable numbers of direction changes.
Rejesha example
11 Consider the following example where playing the move A6R (bold face) leads to
victory in the following way:
j
.t 87654321
e
e b 00000000 b
d right a 12323231 a left
d left A 01100210 A right
h. B 00000000 B
a 12345678
t.
h. One seed is added to A6, the two seeds from a3 opposite A6 are then captured and
e spread from the right into row A reaching A7, then three seeds from a2 opposite A7
;t
are captured and also spread from the right into row A reaching A6. There are no
y seeds opposite A6, so the fou! seeds of A6 continue to A2 and capturing the two
e seeds opposite A2, but this time (obligatorily) spreading into row A from the left
e side, reaching A2 again and since it has no seeds in the opposite position the three
y seeds in A2 continue and reach A5, capturing three seeds from a4, which are entered
>t again from the left into row A, reaching A3, capturing three seeds from a6, also
entered into row A from the left, then continuing with the accumulated five seeds in
a A3 reaching A8 and capturing one seed from a1, but this time (obligatorily) entering
e from the right into row A. This one seed only reaches A8 and, since a1 is already
a captured, the four seeds in A8 are continued to be spread and reach A4, where the
0
two seeds from a5 are captured and spread again from the right, reaching A7 which
N
has no seeds opposite and has accumulated 6 seeds that are continued to be spread
~,
towards Al. Opposite A1 one seed from a8 is captured and (obligatory) spread from
k the left, reaching A1 and now continued to be spread adding up to five seeds,
:s continuing with another four from A6 ending in B7. Meanwhile all the seeds of the
n opposite row have been captured.
II The directions in this move are changed three times, adding up to four directions
in total. There were fourteen holes/stations emptied. The end position is as follows:
87654321
b 00000000 b
1e right a 00000000 a left
a
left A 04266012 A right
le
B 00000011 B
10
es
12345678
98 LIMITS OF THE MIND
In order to undo this move one has to start from B7 and, subsequently, retrace each
step until the original position is recreated, building the opposite row partly from
memory, partly from reconstruction. This task is quite impossible for a beginning
player, but most masters succeeded in returning this complicated move.
Experimental design
The rejesha Bao experiment was designed to obtain insight into the influence of
direction changes upon the calculating abilities of Bao masters.
If we want to measure the influence of direction changes on performance we
have to eliminate all other factors and fix the variables. Therefore, the following
constraints were imposed upon the experiments:
Constant factors:
Number of stations 14 (+/-1)
Number of captures 8
Stage of the game namua (=first stage)
Rows involved a/A row only
Number of seeds 1 or 2 or 3 or 4 per hole
(preferably 1 or 2)
Variable:
Number of directions 2 or 4
The number of captures was set to 8, which meant that the complete row of the
opponent is captured. This forces the player to set up the original game without any
reference points left on the opposite row. The back (B/b) rows are not involved
because they are easy to reconstruct. The number of seeds is limited to small
numbers close to each other to increase the confusion; extremes in the number of
seeds makes them often easier to remember. Moreover, the total number of seeds
stays relatively constant this way.
The directions number either 2 or 4, which is a difference of 100%. There are
also moves with 1, 3, or 5 directions, but for the first and the last it is difficult to
create conditions meeting all the constraints mentioned above.
right or not. A second problem with the experiment is that masters do not necessarily
perform well. Lesser players may perform better sometimes, although the players
who excel are usually masters. The last problem is that the conditions were not
difficult enough to make all masters fail and therefore establish the limits of their
skill. A special variation of the rejesha experiment was designed to establish the
limits of one of the more talented informants.
I
Test techniques: procedural knowledge
As in the duru experiment, procedural knowledge is needed to perform well on the
experiment of rejesha Baa. The experiment is concerned with complicated
combinations in the game. Some familiarity with these combinations is important.
However, much more important is some experience with rejesha Baa. A Baa player
who has never undone a Baa move will immediately fail the experimental conditions
presented here. Rejesha Baa is a common act in a game of Baa. Therefore, all
experienced players are familiar with the skill.
Again, as in the duru experiment, the allocation of memory is all important.
Certain parts of a move are returned quickly but some crucial holes take extra
thinking time, quickly continuing when the number of seeds in the hole has been
decided upon. In one experiment, where most informants (masters and novices)
failed, they all got confused at exactly the same place. Sometimes, they failed even
after showing them this move several times. If the crucial information that was
needed to solve this part of the experiment was pointed out, the informants
performed flawlessly. This confirms once more, that procedural knowledge can be
taught and that practising the wrong track leads to continuing failure.
Results
It appeared that all masters performed well on the rejesha task. Very few mistakes
were made and they were minor ones. 'Since there seems to be no way of
complicating the experiment much more, the idea was not pursued much further.
There was, however, one informant who claimed not to make any mistakes at
anytime in any of these experiments. The thirty conditions that had been designed
were presented to test his claim. Eventually, the experiment was altered to
complicate matters. The informant, Abdu, was asked to look at the board, and then
look away when a move was played. Having seen the starting position, he was then
presented with only the end position. He subsequently, reversed the move I had
played from the end position slowly rebuilding the begin position. Much surprised, I
continued to complicate the test by waiting for up to 3 minutes before the end
position was shown and he was to return the move. The result was another perfect
performance. Now, I provided two positions consecutively, each time only showing
the begin position. Then a move concerning the first position was played while he
was looking away, after which he returned this move perfectly. The second position
was planned, played, and again he returned the move perfectly.
i
In sum, Abdu was able to undo moves blind, simultaneously, seeing only the
1
initial and final positions. Since all the seeds on one row were captured each time,
this performance became very impressive. Only when three games were shown
consecutively did he get into trouble. It took too much time to return the moves, and
1
j
the positions were confused. Each of the experiments was repeated with other sets
of positions in order to rule out coincidence or luck. There was no other player who
1
volunteered to repeat this performance.
s
100 LIMITS OF TilE MIND
The following sets are examples of the conditions that were used:
1. Move A3L was played (note that move A1 also clears the a-row):
87654321
b 00000000 b
right a 12121132 a left
left A 10111111 A right
B 00000011 B
12345678
I~" 12345678
,:lj
:Iii
!! !~
ill!
2. Move A1 was played:
I!Ji
!Iii
!i!: 87654321
•:I
Ill!. b 00000000 b
II::I right a 22121112 a left
,jl, left A 22000121 A right
:r" B 00000011 B
li:
r:,'
;·,1 12345678
!li
The experiment of rejesha Bao illustrates that it is a common skill, in which, as with
duru skills, some masters excel more than others. The complex duru that are
introduced in the experiment do not suffice to measure the limitations of complex
duru skills of Bao masters. Instead, only limitations of skills in rejesha Bao were
measured, which should be considered one of many fani, discussed in the following
Chapter.
Being a duru expert will never guarantee a win. Duru is just one tool in a design to
deceive the enemy. Those who have many tools have many ways to deceive. A
master of Bao has, therefore, many skills. At the basis of human competition in
championship games is the skill of deceiving the enemy. By definition a
championship game cannot be calculated completely and, especially when
calculations predict an unfavourable situation, the master has to turn to deceit. This
skill in all its forms, legal and illegal, is difficult to measure. However, an attempt is
made here to describe the skills of deceit in terms of psychological warfare and in
terms of Bao strategy.
This Chapter returns to the work of von Clausewitz (1832) and provides insight
in the workings of psychological warfare, i.e. warfare not related to the direct
destruction of the enemy. The elements of psychological warfare have examples in
both Chess and Bao. It appears, however, that only legal deceit has a place in
championship play.
Legal deceit borders on khatima, a specific strategic skill in Bao as it is known in
Swahili. Khatima, fani, and duru are the three skills that Bao mastership is made of,
according to the Zanzibar and Dar Es Salaam experts. Durn is part of fani, which is
part of khatima. They appear as classes that constitute a hierarchical order. Duru is
lowest in this hierarchy but it is also the most basic skill in Bao.
This Chapter, therefore, gives an insight into the place that duru skills have in
Bao playing. It also provides insight into the outcome of the championships. The
examples in Appendix II present a corpus of master games. From this corpus various
examples are explained and they illustrate the findings in this Chapter. As a result an
attempt can be made to present the first comprehensive description of Bao
mastership, by way of describing its games, the skills, and the players.
I. Literature on war
Most literature on the subject of conflict is accumulated and analyzed in the works
on war. The analytical work of von Clausewitz (1832) has not only proved its value
through two centuries of warfare but has also provided an analytical framework
within which such complicated conflicts as war can be systematically studied. Both ,':I
the strategic and general definitions of von Clausewitz and his line of reasoning have 1'1
,.,'JL'"'
I
~~~,_:
been used in various sections of this manuscript.
,, As a complement to von Clausewitz, Mahan (1890) elaborated on sea power,
c
since von Clausewitz dealt only with the military on land. Mahan also incorporates
t, the political setting of sea power and, therefore, goes beyond the considerations of
tJ. ~
,'~
von Clausewitz who recognized the importance of politics but did not discuss it.
e {? In order to understand the implications and workings of power politics,
K
e
i"
;~
Machiavelli's work has also some relevance in this respect. However, his work on
the art of war is not considered, instead The Prince serves as a reference for
g
~
psychological/political warfare. In this way, the spectrum from power, politics to
i"
.'J
war is satisfactorily covered for the context of this study.
The discussion of war in the literature mentioned above is culturally biased. In
our century significant non-Western wars have been fought, where the strategy of
guerrilla warfare, for instance, has become more and more prominent (see Friedrich,
102 LIMITS OF TilE MIND
2. Psychological warfare
In describing the thought-processes of masters it seems inevitable that these
processes are highly influenced by acts of psychological warfare when it comes to
match or championship play. The extent to which these acts influence the master and
the extent to which the master uses these acts of warfare himself determine to a large 1
extent his success in the championships and matches. n
There has been no analysis of psychological warfare useful for the purposes of n
analyzing a master's skill. We must understand, however, that disturbing the
opponent's play is mostly legal and, therefore, an integral part of playing a game. 0
In competition one can rank players, list them on a scale and compare their ac
performances. Human players will not perform at their best in each and every game.
Competition listings merely represent averages of players' performances. The higher ac
a ranking the closer a player comes to mastership ranking. I will demonstrate that m;
the element of confrontation, which determines the outcome of each individual op
game, also consists of discriminatory elements that belong to the master. These ex
elements do not measure his technical skills in the game, but his skills in the
successful human interaction or confrontation with the opponent.
MASTERSHIP IN BAO: CONFLICT AND STRATEGY 103
Human interaction with an opponent can consist of foul play, irritating moves,
trick moves, unpredictable moves, etc. Such play, which may even be accompanied
by voiced threats, bluff, insults, smoking or even garlic, is not well described
although well recognized by players and teachers. The effects on nerves and stress
are elements of championship play and influence performances.
Human interaction of the kind described above is considered psychological
warfare. It does not involve ranking or measuring different players. Instead, it
applies to players at crucial or troubled moments in championships or individual
games. I will distinguish the elements in psychological warfare, whether on the scale
of a competitive war game or on the scale of guerrilla warfare, all elements should
only vary in degree and little in importance. In the context of this thesis, I will
exemplify the elements of psychological warfare with playing practices instead of
war practices.
Deceit
There is a wide variety of tools that can be used for psychological warfare, but
unlike a deadly weapon, the results of using the tools vary with each opponent,
situation, and strength of the tool. Tools for psychological warfare are usually not
deadly, but deceptive In this context we should distinguish three ways of disturbing
the opponent's play, not based on motivations of the player but on the actual effect:
1. Illegal deceiving: foul play or other law or rule breaking moves that are
forbidden in a game.
2. Legal deceiving: (unusual) moves that have the effect of psychological warfare
by causing panic or other emotional disruptions on the opponent's side.
3. Setting change: acts that do not change the moves of play but only effect the
setting of play, for instance: killing the opponent's mother the day before the
finals, or just shouting at the opponent in the middle of the game.
Illegal and legal only refers to the playing rules and not to the etiquette. Etiquette
may add or even contradict the playing rules and many times etiquette is somewhat
modified for championship or match play.
Illegal deceiving is useful in disturbing the opponent as long as it is not too
obvious. If the deceit is noticed the player may be disqualified but the 'mistake' can
also be simply corrected. In the latter case, disruption has been successful, even
more so if the deceit has not been noticed at all and will bring further disadvantages
or bewilderment to the side of the opponent.
Legal deceiving is usually well prepared, for instance if the uncomfortable
openings for an opponent are known. Knowing the opponent's play or disguising
one's own play can result in confusion on the side of the opponent or create other
advantageous disturbances.
Setting change has multiple forms. It can range from violence, and other criminal
acts, to blowing one's nose too often. Perhaps, there are useful distinctions to be
made but they will hardly cover all outside influences that can be placed upon the
opponent. They can be done before, or during the game, they can be physical to the
extreme of violence or insulting to the extreme of life threatening etc.
104 LIMITS OF THE MIND
llegal deceiving
Illegal deceiving in the case of Lamu is described by Townshend (1986) as being
allowed as long as it not discovered and even admired if it is very much suspected
but not proven. Illegal deceiving also occurs in Zanzibar, usually by playing
kitakomwe, i.e. playing with one seed concealed in the hand while playing other
seeds (not to be confused with a strategical move of the same name). Playing
kitakomwe is not unusual and does not violate playing rules, but it makes cheating
easier since an extra seed can be dropped from the hand unnoticed.
Rusha, i.e. the throwing of seeds, and fast moves cause many mistakes in play
and may alter the arrangements. Such examples can also be found in the Masters
Tournament (see Appendix II).
Championship rules provide for both instances of illegal deceiving as described
above. The player who throws seeds incorrectly can be disqualified after doing so
twice. Playing kitakomwe is allowed but controlled by both referee and opponent
which makes cheating almost impossible.
Etiquette in Bao has rules for certain irregularities that may occur in the game
(see Chapter IV.F). Therefore, illegal moves do occur, but perhaps not on purpose.
Since it is not allowed to play illegal moves in championships, it is unlikely that
psychological warfare will be very successful in this particular way. There were only
two instances in the games of the Tournament where illegal deceit could be
suspected. In both cases, Kijumbe (Day four) and Nasoro (Day five), the effect was
too small to be decisive. The errors probably occurred by accident but were not
discovered.
Even in Chess, where regulations also prevent illegal deceit, we encounter
psychological warfare of this kind. Chernev (1948:12-13) tells the story of Harrwitz
who played a session of simultaneous Chess. One of his opponents attacked a Knight
with a Pawn. The obvious move would be to play the Knight, however, Harrwitz
found that if the Knight was taken it would be mate in four on the opponent's side.
If he did not move the Knight it would look suspicious, therefore, he turned to illegal
deceit. He makes an illegal move with the Knight. His opponent immediately claims
the usual punishment, i.e. the Knight has to stay where it is and the King has to be
moved instead. He moves the King, his opponent takes the Knight and,
subsequently, Harrwitz could announce mate in four.
Legal deceiving
Legal deceiving is allowed in every game, however, the intricacy of a game has to
allow for moves in order not to be seriously disadvantaged by such deceit.
Disadvantage, in this case, is a move towards losing the game.
Bao allows for many tricks and counter tricks that may blur the opponent's view
or make him wonder, be surprised, panic, afraid, become insecure, or become angry.
The best example in Bao is the shortest win, where the player first loses his house,
usually a big mistake, before clearing the opponent's row and winning the game. If
a player abandons his house it may be a mistake, it may be a good move. If it was
neither of the two, i.e. a bad move intentionally played, then we speak of the ultimate
MASTERSHIP IN BAO: CONFLICf AND STRATEGY 105
legal deceit which is successful if the opponent does not take the house and plays a
move that brings him into more trouble.
In the Tournament, legal moves have resulted in bewilderment or nervousness on
the side of the opponent. In one example, Abdu closed the net on Kijumbe, since
Kijumbe found himself trapped in a series of forced moves, he became increasingly
nervous. He barely escaped and Abdu gained sincere respect from the other players.
Setting change
The examples I presented for setting change were extreme for Zanzibar standards.
However, setting change is more common in Bao than it is in Western play. It is
best compared to street play, as for instance on Washington Square in New York.
Gambling, bluff, time limits, witchcraft, etc. all influence the unsuspected player. In
the same way, a Bao player may be urged to move, people may move seeds instead
of him, the opponent may constantly challenge and provoke him, and often the
opponent returns his move and shows him a 'better' move instead.
As in illegal deceit, championship rules limit the possibilities of setting change.
"Spectators are not allowed to speak or interfere, although they may praise a move
by saying 'Heko"', state the Bao Society rules. Respect is usually shown to masters
at play, since the spectators' comments will be of much lesser quality. Masters,
however, do speak with each other and comment on the game, but not to impress or
influence each other's emotional well-being. A lesser player may want to intimidate
a better player by throwing many seeds and speeding his play, but masters usually
rebut with good moves, the most effective answer.
The Tournament itself appeared the most important setting change to most
players. Abdu and Shamte became very nervous at the start of important matches.
Rajah and Abdu appeared to be much disturbed by losses, which resulted in more
bad games if they played again on that day. Although self-imposed, these influences
are considered setting change. In Chess, 'diseases' like the Kotov syndrome and
amaurosis scachistica have been blamed for blunders in similar circumstances
(Hooper & Whyld, 1992:47).
At a masters level, there are few stories known of extreme setting change. Ali
Kipara played Othman Omar in a friendly match. At a certain point, a complicated
combination would lead to the departure of a hole of 15 seeds on the part of Othman.
Kipara told Othman that this hole of 16 seeds would be a bad move, because he had
calculated it beforehand. Othman disputed the fact that they were 16. Kipara
answered something not very convincing, and Othman played the move, certain that
Kipara had made a mistake. However, the play would result in a complicated but
hardly profitable position. In fact, Kipara could clear the game the very next move.
He told Othman, that as soon as he started to dispute the correct number he knew
that he was going to play this move. This procedure is legal but not in championship
play. In another era, Mkiwa also had to use some deception to make masters play
him, since they were usually afraid of losing as soon as they knew who he was.
Such trickery is, of course, entirely legal but also essential to his becoming East
African champion.
1. Illegal deceiving: foul play or other law or rule breaking moves that are
forbidden in a game.
2. Legal deceiving: (unusual) moves that have the effect of psychological warfare
by causing panic or other emotional disruptions on the opponent's side.
3. Setting change: acts that do not change the moves of play but only effect the
setting of play.
between warfare and psychological warfare in their effect on the individual parties.
Clausewitz (1832) explained that honor is a key element in a soldier's motivation. In
psychological warfare, honor is at the mercy of higher stakes. Winning is more
important than nice play or even worse; winning is less important than the stakes
beyond. At such a stage we leave war-theory and enter into politics. Ugly play or
illegal deceiving resembles immoral ruling by a prince (Machiavelli, 1513), where
the end justifies the means. If, according to Clausewitz, war is the continuation of
politics with other means, then, in the opposite direction, psychological warfare
seeks the political motives in a war game. The game has become a reason to play
politics.
I believe that this depends upon whether cruel deeds are committed well or
badly. They may be called well committed (if one may use the word 'well' of
that which is evil) when they are all committed at once, because they are
necessary for establishing one's power, and are not afterwards persisted in,
but changed for measures as beneficial as possible to one's subjects. Badly
committed are those that at first are few in number, but increase with time
rather than diminish. Those who follow the first method can in some measure
remedy their standing both with God and with men, as Agathocles did. Those
who follow the second cannot possibly maintain their power.
Machiavelli (1513:33)
Extremely strong motivations to win are found in players like Bobby Fisher. Fisher
displayed some (in)famous acts of setting change as well. In general, championship
organizers and better players can counterbalance such power. In fact, better players
overrule 'powerful' players. A better player does not have to exceed the power but
will have to withstand its pressure and use other means of war. As players can be
immune to certain tricks in a game itself, they can also be relatively undisturbed by
setting change. Power play is only useful if technical play is also of high quality.
One cannot win a conflict if battle is always refused. In a similar context, Mahan
(1890:138) concludes that "such inquiries [i.e. commerce destroying cruises],
unaccompanied by others [i.e. sea battles], are more irritating than weakening."
Potter (1947) has been the only writer who has seriously addressed the issue of
setting change, or gamesmanship as he calls it. Since the art of winning games
without actually cheating is pure setting change, his work should not be neglected.
In Chess, several examples of gamesmanship are described by Hooper & Whyld
(1992:148). They vary from making a poker face when a mistake is made, to writing
down an extra move that confuses the opponent when he is in time tro_uble.
)_
r
108 LIMITS OF TilE MIND
llegal deceiving
Since a ruler, then, must know how to act like a beast, he should imitate both
the fox and the lion, for the lion is liable to be trapped, whereas the fox cannot
ward off wolves. One needs, then, to be a fox to recognize traps, and a lion to
frighten away wolves. Those who rely merely upon a lion's strength do not
understand matters.
Machiavelli (1513:61)
In the case of illegal deceiving we have to speak of the fox and not of the lion. We
still speak of ugly play or immoral behaviors. Setting change should be called brutal
force. Its messages are more obvious. Illegal deceiving occurs concealed from
onlookers. Machiavelli (1513:25) states, that the Duke of Milan very cleverly
concealed his intentions when he resorted to trickery and therefore became
successful and powerful.
Since concealment of the deceit usually is preferred to discovery, the attention of
the player is on his disguise. There is physical disguise by palming a card, putting
pieces up one's sleeve or playing kitakomwe, but, usually, illegal deceit occurs in
combination with other deceits. A setting change, like shouting, gesticulation or
other distracting actions is very common in this respect and much used in deceits by
magicians.
A more intricate deceit occurs when two deceits take place where only one is
discovered and corrected. Moreover, there are deceits discovered but not corrected
since it seemed favorable to the discoverer. The eventual advantage to the player
then becomes very intricate to calculate and there is a combination of illegal and
legal deceiving produced simultaneously.
The nature of illegal deceit is very different from setting change. Well played
illegal deceit is never fully discovered and such· masters remain under cover.
Legal deceiving
Legal deceit enters into the field of regular tactical or even strategic play. Many
moves are intended to mislead, surprise, or confuse the opponent. They are just an
introduction to a regular good moye that gives, in Chess terms, material or positional
advantage.
There are some aspects of legal deceit in Bao that have more general features.
The requirements for such deceit can then be defined in a general way:
Firstly, there is prepared knowledge, the memorization of a certain set of moves,
r: knowing the outcome without having to calculate too much. Such a set can have
implications for over 17 moves ahead (see example in Chapter V.B) while thinking
17 moves ahead appears an impossible task, especially in a limited time frame.
Secondly, there is the gambit, a common term in Chess for describing a sacrifice.
Pieces or positions are given up in order to gain other advantages that become
important at a later stage in the game. A greedy player in Bao is often and easily
punished, as is shown by the example of the shortest game in Bao where the house is
abandoned (Chapter V.D).
Thirdly, confusion or complexity can be added to the game that, especially with
constraining time limits, makes the opponent more vulnerable to the making of
mistakes. This effect can be reached by introducing many traps at the same time that
also combine into new traps. ·
,il-
MASTERSHIP IN BAO: CONFLICI' AND S1RATEGY 109
These three examples of legal deceit illustrate that legal deceit borders on regular
playing skills but differs when they cause an opponent to make mistakes. The
anticipation and induction of the opponent's mistakes is a crucial asset to a master
that is missing in computer generated play.
Conclusion
The discussion of psychological warfare does not offer tests to measure and quantify
the individual capabilities in the different aspects of deceit in play. However, a
structural analysis of psychological warfare is necessary for our understanding that a
master's excellence does not consist of just memory skills. Their thinking processes
in a particular game are guided by other intellectual processes as well. The manifold
possibilities of these extra skills demonstrate their importance.
Having shown the relative importance of psychological warfare in Bao, and stressing
the cultural differences as key elements for determining the individual factors, I now
turn to the factor of legal deceit which enters into every competition or conflict
where strategy can be determined. Legal deceit is not identical to strategy. Legal
deceit is concentrated on the human factor within the strategic plans. Legal deceit is,
therefore, part of but not equal to strategy.
By showing that there are such human factors in strategy, we also have to
recognize the inherent complications for computer scientists who intend to solve a
game. Moreover, we can hereby explain what element of a game remains unsolved
when it is 'solved' by computers and continues the challenge for human competition.
All providing that we speak of games of strategy as played in the present human
context.
Mzee Mkiwa, former grandmaster of Bao, played new opponents several times,
making draws of two games to one (three to zero being real victory), before entering
into his real combat play. Hereby, he was able to exploit the weak points he
discovered. Such search for understanding of the opponent is found in all strategic
competition. Mahan (1890:164-165) states: "The qualities and characteristics of the
enemy are among the principal factors which a man of genius considers ... "
Legal deceit in its basic elements consists of recognizing, anticipating and
exploiting the weak points of the opponent. Thus the player has to concentrate on the
opponent's strategic thinking process.
Elements of legal deceit can occur both in tactics and strategy, i.e.
instantaneously or within a general plan. Conceit, feint, and surprise occur most
often in this context. Surprise, as argued by Clausewitz (1832), is often limited to
one time and affects only a minor part of the operations. However, surprise is also an
outcome of successful conceit or feint entering into the climax of the match but
carefully developed beforehand. Such a development is part of strategy and applies
to both technical strategy in warfare and the elements of psychological warfare.
The ultimate strategical challenge should, therefore, be formed in the
concentration of human (legal deceit) and technical (game specific) strategy. When
recognizing the importance of both elements of strategic challenge, it appears that
quantification of one of these elements only will never suffice for the complete
understanding of the differences between masters and novic_es. Technical mastership,
in the above sense, does occur more frequently and may result, for instance in Chess,
from study practice by use of books without considerable playing experience.
110 LIMITS OF THE MIND
However, a champion or a master in the pure sense occurs when the opponent is
beaten and not necessarily when the technical skills of a game are best understood.
Conceit, feint, and surprise applied in combination or separately, cannot be
quantified with tests as easily as the technical skills. The elements occur in a
necessarily strong competition where an opponent triggers the use of skills of legal
deceit. However, as with technical skills, there should be some determinable
presence of skill. Situations which are sensitive to legal deceit should be easily
recognized. There is a problem, however, that there is no necessity in a masters'
game to have both technical skill and skill for legal deceit. Many mistakes, and
therefore technically unsound play, in a game may still result in a brilliant victory by
deceit. Compare the relative numerous mistakes in sea battles where brilliant draws,
or even victories can still be won through genius of conceit, feint and surprise as de
Ruyter did in the battle at Solebay (e.g., Mahan, 1890).
B. Strategy in Bao
1. Methods of research
Apart from general conversations with players, there have been two exercises that
encouraged me to write a Chapter on technical mastership in Bao in all its forms.
Firstly, we conducted seven long interviews on video with masters of the game, i.e.
Mkiwa, Shamte, Nasoro, Kijumbe, Rajah, Masoud, and Majaliwa. Together with
short conversations off camera, these interviews have presented an interesting range
of experiences and theories concerning Bao mastership. Secondly, during the
tournament, there was always a group of masters around that discussed games and
other issues after the matches. Since this continued for three weeks, there was ample
opportunity to discuss rules, elements of mastership, characters of players, and
games. Each master contributed in this way to the theories presented here. I
specifically asked some masters whether I was in a position to discuss matters of
mastership, or whether there was too much else to know before I could enter into
such a venture. There seemed no serious problem; however, I will refrain from any
speculation as far as specific skills are concerned. The annotations to the games only
illustrate superficial, i.e. one or two moves deep, observations or remarks by the
players themselves.
usually not successful, however, considering several plans in Chess does not usually
waste a Chess master's time.
3. Fani
Strategic skills in Bao consist of fani, made with many or few duru, and is by nature
a long term objective, looking several moves ahead. We can distinguish two kinds
of fani, general fani used in both stages of the game, and fani restricted to either the
namua or mtaji stage. I will list the important plays with their Swahili terms
illustrated with games from the tournament:
General fani
The verbs weka or miliki, describe a Bao situation in which the opponent is trapped,
i.e. he has lost control over the game, his moves are dictated by the opponent. In
Chess such situations are sometimes referred to as a grip or a bind by a player.
Nasoro trapped Abdu, in a game where the mtaji stage was fully
controlled by Nasoro. For six moves, Abdu had only one capture to play,
dictated by Nasoro's moves. Nasoro won the game (Day 5). Such a sequence
of forced moves is called lazimisha (to force). Since capturing is obligatory,
this is a tactic often used in Bao. ·
Kitakimbi is a specific trap that forces the player to defend his kimbi hole.
This a good example of a weakness, i.e. a position difficult to defend (Hooper
and Whyld, 1992). When this continues for several moves, one speaks of
kitakimbi. This is considered a disadvantageous situation. The kimbi hole is
more difficult to defend than others, since it is less likely to be emptied in a
combination move. Therefore, the kimbi hole is often sacrificed, even if it
contains many seeds.
The saying goes: Nyumba ya kimbi hainamuki. Which means that the
kimbi hole with many seeds should not be played into (namua-ed).
Othman was forced in this way against Shamte (Day 4), but was able to
empty his kimbi hole with 16 seeds into his own row in the 22nd move, the
last move before entering the mtaji stage. A similar thing happened to him
against Kijumbe (Day 1) where he played a kimbi hole eight times in a row. Ft
He finally plays the 28 seeds gathered in his kimbi, but Kijumbe's trap makes TI
him resign before the next move is played. otl
st~
Tapisha Bao is a most spectacular skill in Bao. There are few players who master
this 'vomiting of Bao '. The moves consist of a series of sacrifices, depleting one's
, , 'I,
~:I
,1'
'I
1!:
own rows while the opponent captures many for his own. Suddenly, a move is
created which captures everything back and changes the situation dramatically and
usually decisively. Many times, the opponent captures many seeds which become
piled up in his front row. Usually, the players who gives many seeds, has some
seeds placed in such a way that travelling the back row is still possible. There are a
few examples in the tournament that illustrate the art.
Day 7:
87654321
b 13203101 b
right a 00121020 a left Abdu
left A 21u79050 A right Shamte
B 21051210 B
12345678
Shamte has to play his only mtaji at this point, A1R. Abdu strikes with a2R
after which the situation is as follows:
87654321
b 04014121 b
right a 20360555 a left Abdu
left A 101200000 A right Shamte
B 21051210 B
12345678
Shamte is forced to takasa, Abdu controls and finishes the game. Similar
feats have been accomplished by Shamte himself, as well as Kijumbe, and
Nasoro.
Takasia is a specific skill that only occurs in the mtaji stage. It occurs mostly when
a player has a position that has deteriorated, although not necessarily. The move
114 LIMITS OF THE MIND
Kitakomwe is, among other things, a special trap in the namua stage, in which the
opponent's house is captured without emptying one's own house. The special moves
used for this game are disadvantageous when your opponent is familiar with them.
The kitakomwe game usually ends quickly and used to be very popular. Among
others, Nahodha is said to have been an expert. Nowadays, .the game is not
considered as being very good and has largely been abandoned.
Namua nyumba, or to play the house, is sometimes avoided by certain players, like
Kijumbe. They prefer openings where the house is emptied early in the game.
Others, like Rajab, prefer the game around the house, or will start building a new
pile of seeds when the house has been emptied. It appears to simplify the moves,
however, the game acquires a certain climax of complexity when the house obtains
dangerous proportions.
A game on Day 8, Shamte vs. Nasoro, shows clearly that Shamte is building a
large hole on A4. On moves 10 to 16 he enters only into A4 in order to defend and
increase it. Such moves usually consist of gosa, which is to place a captured seed in
an empty kichwa (outermost) hole, thereby ending the move.
Takasa nyuma, to play into the back row is considered a different fani. This skill is
not considered a trap, and should be referred to as tegua.
Majaliwa prefers to play into the back row from the opening onwards. Others,
like Rajab and Kijumbe sometimes play an opening where they resort to playing into
the back row, as on day 6 when Kijumbe plays the back row in an answer to Abdu's
sixth move, or on day 12 when Rajab does the same against Shamte. It shows
clearly in the record of the games, since the namua stage appears with many takasa
moves.
The absence of such, moves with the other players - personal communication
confirmed this- shows that they are not comfortable with this kind of play.
The strategic skills described above are difficult to generalise into strategic rules.
Apart from sayings, like Nyumba ya kimbi hainamuki, it appears difficult to give
general rules for strategic play. The only tangible agreement on strategic rules is, as
in Chess, the opening moves. For many different reasons, and after many years of
play, masters prefer certain openings, playing the same answers several moves long.
Perhaps contrary to Chess, opening moves are played rapidly, i.e. very rapidly. In
the game Nasoro - Abdallah (Day 7), the first ten moves were played in only 31
seconds. To play 20 single-moves of Chess in 31 seconds is fairly easy. However, in
Bao an average move contains many changes. The reader is challenged to play the
20 single-moves of this game in this record time, and will find it surprisingly
difficult.
Apart from being played rapidly, many opening moves are standardised, i.e.
played in the same way whenever they occur. The opening A7L is especially long
with few variations. It is said that the opening A6L is most diverse, not only
concerning the different answers to certain moves, but also to the different fani that
can be used in it.
Many skills in Bao, and especially opening theory, can be compared to Chess.
Apart from generalising the theory above, the openings of Bao will give us more
insight in Bao from a Western perspective. This is perhaps useful for those
Westerners who doubt the expertise shown by the Bao masters. However, certain
common elements in Chess and Bao are also meaningful for the purposes of this
study. Bao mastership consists of several elements, as was pointed out before. If
Chess skills, like memory for openings, position play, material gain, etc., are present
in Bao but not discussed as such by the masters, then a short analysis of the
tournament games and examples from play should make clear the possibilities and
usefulness of generalising Chess theory to Bao. For such a study we need a more
elaborate analysis by Bao masters. I will restrict my comments to a short discussion
of opening play in Bao.
1. Openingplay
Openings can only be established by proving their soundness. Such proof is the task
of a Bao analyst, in our case, a Bao master. Before establishing the soundness of an
opening it is first necessary to identify them. It is only on this level that the following
study can make a contribution.
Openings in Bao can be discovered in three ways:
1. With the help of the opinion of masters.
2. With the help of an inventory of openings played in the tournament
3. With time records of the tournament that show the number of beginning moves
that are played without hesitation.
I will only discuss certain repeated openings in the game. The relative speed with
which these openings are played may indicate that we are not analysing a
coincidence. Beyond the statistics of the inventory there are no means of analysis
available except the opinion of masters.
As in Chess, certain games are sometimes repeated well beyond the opening
moves. Therefore, we define openings in this tournament as; beginning moves that
were repeated by different pairs of players. Unusual openings are sometimes shown
as well, even if they were not repeated, in order to illustrate the variation.
116 LIMITS OF TIIE MIND
Table 13: Preferred openings of masters (only) in the 1994 Masters Tournament
In the tournament league there were 92 games played involving the above masters (four regular players at
the tournament have been excluded). These games provided 64 openings by the masters listed above.
(1) Masters are listed in alphabetical order. The lesser players of the tournament have been excluded.
(2) This column lists the favourite opening move, i.e. first move, of each master.
(3) This number indicates the frequency of their opening moves, with a maximum of 10, 9 (Nasoro and
Rajah), or 6 (Majaliwa).
(4) The master opened the game and answered to each move in the listed way. The opening moves are
only listed if this sequence of moves occurred more than once. Only the longest repeated opening
move is shown in this way, for each opening (6L, 6R, 7L, or 7R). Kijumbe, Rajah and Nasoro have a
single preference and two examples of repeated openings are then provided in the table.
(5) 'A' indicates that the master opened the game (White in Chess). 'a' indicates that the opponent
opened the game (Black in Chess).
(6) Indicates the number of single-moves (plies) or minutes that a master played. The numbers are split
between games opened by the master and games opened by the opponent.
(7) Indicates the number of wins for games begun by the master or the opponent.
The averages and the repeated openings mentioned for Majaliwa .(and, for instance, Maulidi's second
opening) are much less interesting because of the relatively small corpus of data available.
MASTERSHIP IN BAO: CONFLICT AND STRATEGY 117
In general, opening theory in Chess does not exceed 11 moves. This is similar in
Bao, however, it is not always possible to determine whether a game is repeated or
whether they are playing opening theory.
Hooper & Whyld (1992:279)state that the rules of Chess in 1475 in some way
prevented the development of opening theory. This changed in the 19th century,
after the London Rules were internationally accepted, and there came a great
advance in opening studies. In the 20th century, knowledge of opening games often
became decisive in games of Chess.
In Bao, decision complexity (see also Appendix III) is small, therefore, by
definition, there is a small number of initial openings possible. However, with each
move there is a seed added to the game. The game, therefore, gradually becomes
more complicated and remains the same until the end.· In Chess, this is the other way
round. The game starts with all the pieces on the board and the end game is defined
by a number of exchanges that reduces the number of pieces of the board.
The tournament shows that the player who started a game (A) did not have a
serious advantage. On the contrary, 41 out of 64 games that were opened by masters
were lost, as opposed to 32 games that were lost when the opponent started. Since
this is the only statistic available on opening play, we can only conclude that in this
championship opening play did not form a serious disadvantage but rather an
advantage to the opponent.
The opening of 7L
This opening was played fairly often in the championship. Nowadays, it is preferred
by many experts, K.ijumbe preferring it to any other opening. The game has certain
characteristics when 7L is followed by 6L in the next move. Firstly, opening theory
(i.e. an often played variant) covers twelve moves, i.e. up to the point where the
players capture each other's house (cf. exchanging Queens in Chess). The game is
considered even for both players at that point and continues with a high turnover of
seeds with each move. This kind of play fits in nicely with the expertise of Kijumbe.
He does not excel at calculating long duru, but complex duru are plentiful because of
the many combination captures that are involved. A second advantage for Kijumbe
is the relatively long opening theory, which will make the players reach the mtaji
stage relatively sooner, ;1 stage which is also much preferred by Kijumbe.
Other variants of this opening also exchange the houses or make the house leave
at an early stage. Therefore, this opening is preferred by specialists of games with
complex duru.
The opening of 6L
This opening consists of many variants. It is the most all-round opening in the game.
Players who prefer the game of defending and playing the house, usually prefer this
opening. However, opening theory of certain variants (see Nasoro's opening, the two
arrows in the notation show that both houses have been played/emptied) sometimes
lead to an early departure of the house, and again complex duru follow. The plays
are so diverse that mastership of all aspects of this opening should bring you all-
round knowledge of opening theory, stated one of the masters. It is not surprising
that this opening was played most often in the championship. Depending on the
opponent, one can choose the direction of the game with the kind of opening, as in
Chess and other board games.
118 LIMITS OF TilE MIND
Due to an error the match between Maulidi and Omar was first given as 2 - 0 for Maulidi, in this table, as
opposed to the announcement at the tournament, this mistake has been corrected.
D. Conclusion
An important part of strategic skill in Bao is difficult to· quantify with experiments.
Although duru skills are at the basis of all Bao skills, fani and khatima appear
MASTERSHIP IN BAO: CONFLICT AND STRATEGY 119
• I.-
~'
':~
~e'
i
,-;_,
~
I'! I
,:-~
I'
1
~~
~r~ ~ 1·,
~ [j
-
~~
~ I
~~--
'~ ~~
,,
•J' \'
"i.
l :ili
'.i I
':~t 'I
1'1:
I
1
,., ('I
¥
'"?
1111
11 ~ 1
.1, I!
~ I,'I
~
~
-~i
'I I!
t ,I I'I
1'1
:f I:
I'
~
fj !
!
i:
t
':'o~ ,,I•
':I :II
•~ II
,,
"r
~
":i•
il
I
"' II
11
I
. ~
Ii
I
35
30
"'
Q)
25
~
01)
....... 20
0
....
Q) 15
z~
~'
10
5
0
1-5 6-10 11-15 16-20 21-25 26-30 31-35 36-40 41-45 46-50 51-55
Moves
I
I
l
I; Length of Bao games
'!\
J[!
li
i.,jl: 35
l 30
s"'
Q)
;.'
I 25
1·'
j,
"'
01)
....... 20
0
j: .... 15
I" Q)
~
~ I•
10
z
5
0
1-10 11-20 21-30 31-40 41-50 51--60 61-70
Minutes
IX: CONCLUSIONS
1. Practice
Investigations into the skills of Chess masters (de Groot, 1946; Simon, 1979) have
led to explaining the differences between masters and novices in terms of experience
(de Groot, 1946) or practice:
Hence, the overriding factor in chess skill is practice. The organization of the
master's elaborate repertoire of information takes thousands of hours to build
up, and the same is true of any skilled task (e.g., football, music). That is why
practice is the major independent variable in the acquisition of skill.
Simon (1979:426)
Such an explanation fails to explain the differences between masters and novices in
Baa. Baa players do not seem to have obvious or sufficiently frequent opportunities
either to practice separately from other players, or to study in order to become
masters.
The present study does confirm the conclusions of de Groot (1946) that masters
cannot be defined as more or less intelligent in terms of general education, or
professional/social successes. In Zanzibar, the education and profession of Baa
masters appears to be defined by their time and place of upbringing. Their
educational and professional backgrounds are manifold. The only commonality that
masters seem to possess is that almost all learned the game at a young age, reaching
mastership in their twenties, as seems to be the case in Chess according to de Groot
(1946).
This thesis presents a definition of mastership in terms of limitations. A master is
someone who is that much closer to reaching the limits of the human mind, not
someone who exceeds those limits. This definition contrasts, therefore, with those
that imply that certain masters are essentially super-human. Practice, only in
combination with procedural knowledge, can lead to mastership. The individual
limitations of the masters appear to determine the result in master tournaments and
these limitations can fluctuate per match, per person, and per skill.
.!
122 LIMITS OF TilE MIND
2. Masters cannot calculate unlimited simple, short, blind simple, blind short, or
complex duru. All these calculations are limited by Miller's constraint and a
newly introduced dynamic constraint (Chapter VI & VII).
3. Masters possess a limited number of fani. They specialise in certain fani and in
certain openings (Chapter VIII).
4. Masters are vulnerable to setting change and illegal deceit, which may worsen the
limits in their playing skills (Chapter VIII).
3. Dynamic constraint
The average state-space and game-tree complexity in Bao, for which Chess and Go
are famous, is not impressive. Instead, masters require the skill to calculate duru.
The move of spreading seeds together with the complex rules create a mutational
complexity. This is the essence of the difficulty of Bao for humans; however, it is
not 'complex' for a computer. On the contrary, a computer can deal with an almost
unlimited mutational complexity without any serious memory problem while
humans are severely limited. It is exactly this maximum which is investigated in the
experiments with Bao masters.
In all the experiments reported in the literature, subjects stored increasing
amounts of information in memory. In the duru experiment, one can simply replace
the information stored in memory in order to perform well on the task. However,
replacing memory information appears problematic. If masters are presented with
duru experiments with only eight holes, within the limits of Miller's memory
constraint, they perform worse than when they have double the number of holes.
This is initially surprising, because a model with eight pieces of information needs
little more than eight pieces of information to accomplish a perfect performance on
this task. The information is simply replaced with each round.
Since the subjects are limited in their performance on this task with increasing
study times in each round, and since their performances are better explained with a
theory of effective memory allocation, it was argued that human memory has a
dynamic memory constraint as well as a storage memory constraint (Miller's
magical number of seven plus or minus two). This dynamic memory constraint
prohibits rapid replacement of information
This constraint disappears in slow change procedures, like games of blind Chess,
or blind Bao. In the latter case the number of changes exceeds the number of
changes in the best performances on duru tests. Within the skill of blind Bao, the
player is again limited when he calculates duru of one move. However, the master is
perfectly able to use pattern recognition in the sequences of moves in blind play,
whether Chess or Bao, but which appear impossible in the experiment of duru.
4. Summary
At the basis of expert skills in Bao lies the ability to calculate duru. The outcome of
various experiments on duru skills introduce the dynamic constraint in human
memory, unknown in computer memory, which limits continuous change of memory
information. Since practice alone does not explain differences in Bao performance,
the allocation model defined within the limits of Miller's magical number seven and
the dynamic constraint proves more accurate. Procedural knowledge then appears
essential in the acquisition of duru skills.
In order to excel in Bao, masters also obtain_ knowledge about certain
combinations and certain opening moves. They acquire procedural knowledge on
duru calculation skills which they practice in play. They experiment with fani in
CONCLUSIONS 123
The following glossary gives a list of Swahili words often heard among Bao players.
The words do not necessarily occur in the text. Most words have similar meanings in
daily life. If appropriate, their regular meaning is presented as well. A few words
(in italics) seem to occur only in Bao. Special attention should be paid to the word
kitakomwe which seems to have several meanings within the game of Bao. Note that
nouns are translated without articles preceding them. Verbs are indicated in the
translation by using 'to'.
Shamba inland
Taasisi Institute
takasa (or takata) to play without capturing
takasia to play without capturing and forcing the
I opponent to play without capturing, after
which it is possible to capture
! tanga mtini not to know where one's move is going
I (literal meaning 'to climb a tree')
I tapisha Bao to play an elaborate gambit (literal
II tega
meaning 'to vomit Bao')
to trap
I tegua 1. to neutralize traps; 2. to counter-trap
to boa 1. to play a special opening move (now
obsolete) where seeds of a/A6 and a/A7
I are lifted together and played to the left; 2.
to bore a hole
I! 1'
tobwe 1. novice; 2. fool
i Tumbatu name of island belonging to Zanzibar, off
II the coast of Northern Zanzibar (opposite
I
Mkokotoni)
Unguja Zanzibar
1:1 1 1. play of singletons (7 or 8) in the front
,,,1
1'1
utitiri
II row; 2. chicken lice
weka 1. to be in power of the opponent's game;
'I 2. to put
zamu turn (of a player in a game)
lI
I
li
'II
1'1
,,
t
;I
,,r:
I'I~
]\
i
r I''I'
11
iii
I''
:i,
1
il ,-_; -,
I
:I'
r:
1:
I,
1.
1,-;.
I I
LITERATURE
Allis, L.V., H.J. van den Herik & I.S. Herschberg, Which gemes will survive? In: D. II
Beal & D. Levy (Eds.), Heuristic Programming in Artificial Intelligence 2: the I
Ji
~--~
128 LIMITS OF THE MIND
\.
I'
il
II,
LITERATURE 129
I
Muller, H.R., Warri: a West Mrican game of skill, Journal of American Folklore
1930, 43:169
Murray, E.K.M., Caught in the web of words: James A.H. Murray and the Oxford
English Dictionary, New Haven: Yale University Press 1977.
Murray, H.J.R., A history of chess., Oxford Press, 1913.
Murray, H.J.R., A history of board games other than chess, London: Oxford at the
Clarendon Press, 1952.
N'Guessan Assande, G., L'apprentissage de l'awele: Etude du processus
d'acquisition des tactiques et strategies, Presented at Colloque de Cerisy-la-Salle,
1986, June.
N:simbi, M.B., Omweso, a game people play in Uganda, Occasional Paper 6,
University of California: Mrican Studies Center, 1968.
National Museums of Tanzania, How to play Bao? DarEs Salaam, 1971.
Newell, A. & H.A. Simon, Human problem solving, Englewood Cliffs, N.J.:
Prentice-Hall, Inc, 1972.
Ochse, R., Before the gates of excellence: The determinants of creative genius,
Cambridge: Cambridge Univerisity Press, 1990.
Odeleye, A.O., Ayo: A popular Yoruba game, Ibadan: Oxford University Press,
1977.
Peterson, L.R. & M.J. Peterson, Short term retention of individual items, Journal of
Experimental Psychology, 1959, V.58, pp.193-198.
Philidor, F.A., L'analyze des echecs, London, 1749.
Potter, S., The theory and practice of Gamesmanship, Middlesex, England: Penguin
Books, 1947.
Raabe, J., L 'Awele, Paris: Ed. de la Courtille, 1972.
Retschitzki, J., Strategies des joueurs d'awele, Paris: L'Harmattan, 1990.
Reysset, P., Les jeux de reflexion pure, Paris: Presses Universitaires de France
1995.
Russ, L., Mancala games, Michigan: Reference Publications, 1962.
Sacleux, Ch.C.S.Sp., Ancien missionnaire Dictionnaire Swahili-Fram;ais., Paris:
Institut D'ethnologie, 1939.
Sanderson, M.G., Native games of Central Africa, Journal of the Royal
Anthropological Institute of Great Britain and Ireland, 1913, V.43:726-736.
Sawyer, W.W., The game of Oware, Scripta Mathematica, 1949, V.15.
Shackell, R.S., Mweso, Uganda Journal, 1934, V.II. iII,
Simon, H.A., Models of thought, New Haven: Yale University Press, 1979.
Simonton, D.K., Genius, creativity, and leadership: Historiometric inquiries,
Cambridge, Massachusetts: Harvard University Press, 1984 .
Stoep, A. van der, A history of draughts: with a diachronic study of words for
draughts, chess, backgammon and morris, Rockanje: Vander Stoep, 1984.
iI
I
li
II
Sun Tzu, The principles of war, Translation by E. Machell-Cox. Ceylon: Royal Air 1i I
Force publication, 1943. (original510 B.C.) r I
Sun Tzu, The art of war, Translation by Cheng Lin. Shanghai: The World Book
Company, Ltd, 1945. (original510 B.C.) I
Tarrasch, Dr., The game of Chess: A systematic text-book for beginners and more
experienced players, London: Chatto and Windus, 1935 .
Townshend, P ., Les jeux de mankala au Zaire, au Ruanda et au Burundi, Les Cahiers
du CEDAF, 1977, V. 3:1. II
Townshend, P., Mankala games, International Committee on Urgent
Anthropological and Ethnological Research, Bulletin, 1977, V. 19, pp.47-54. 'If
::1
130 LIMITS OF THE MIND
c
ri
11
l
!
;:
i
I,. I
~'""
>: . -_=~,-
Ii
--~
!
~,!,_-_-
! ,,
~=..·
("
I! ~;i~
I --;.,·
,,._,=s
ll!'
,,,, ''t'-.=:
/ :~-1
1
11 .;o-.'
.,.,
,,;I[~
f:
j[:" _fj,_
llf'
.'!'f,
~,t'l
c~
'~:
;tc
I'' ~-;t:'.
:•
li
1:.•'l
I'
t
'"'·
'II
t:
~:
'.: .
~:
I
lr
I!
l
!I:-:.
II
I
l
,.
L .
I i·
: ~
Station Cnd.P Round Cnd.( Round Cnd.R Round Cnd.S Round Cnd.T Round
1 4 1 2 1 4 1 2 1 2 1
2 7 1-2 3 1 5 1-2 3 1 2 1
3 4 2 2 1 2 2 4 1-2 5 1-2
4 2 2-3 4 1-2 2 2 5 2 4 2
5 3 3 5 3 3 3 4 2-3 3 3
6 2 3 2 3 2 3 2 3 2 3
7 4 3-4 2 3 5 3 5 3-4 2 3
8 5 4 3 3 7 3-4 2 4 2 3
9 2 4-5 2 3-4 4 4-5 5 5 5 3-4
10 2 5 5 4 2 5 4 5 7 4-5
11 3 5 7 4-5 4 6 2 5 2 5
12 2 6 4 5-6 3 6 3 5-6 2 5
13 5 6 2 6 4 6 2 6 2 6
14 7 6-7 4 6 3 6-7 2 6 3 6
15 4 8 3 6-7 2 7 5 6-7 2 6
16 2 8 4 7 2 7 4 7 4 6-7
17 4 8 3 7-8 3 7-8 2 8 5 7
18 3 8-9 2 8 6 8 3 8 2 8
19 4 9 2 8 2 8-9 2 8 2 8
20 3 10 3 9 2 8 3 8
21 2 10 6 9 8 8-9 2 8
22 2 10 2 10 8 8-9
23 3 10
24 6 10-11
25 2 11
132
Tables 16, 17, 18, & 19: Results of the short duru experiment, Ta
conditions Q, R, S, and T.
Cnd. Round Station ROR MJS SMS AMF Cnd.R Round Station SMS
2 1 1 4 1 1
3 1 2 5 1-2 2
2 1 3 2 2 3
4 1-2 4 2 2 4
5 3 5 3 3 5
2 3 6 2 3 6
2 3 7 5 3 7
3 3 8 7 3-4 8
2 3-4 9 4 4-5 9 X
5 4 10
7 4-5 11 X X
4 5-6 12
2 6 13
4 6 14 X
3 6-7 15 X
Cnd.S Round Station ROR SMS NOA AMF Cnd.T Round Statiou NOA
2 1 1 2 1 1 't
3 1 2 2 1 2
4 1-2 3 5 1-2 3
5 2 4 4 2 4
4 2-3 5 X 3 3 5
2 3 6 X 2 3 6
5 3-4 7 2 3 7
2 4 8 2 3 8
5 5 9 5 3-4 9
4 5 10 7 4-5 10 X
2 5 11 X
3 5-6 12
2 6 13 X
il
I
133 I'
II
,I,
Tables 20, 21, & 22: Results of the blind simple durn experiment, l !l
conditions B, C, and D. 'I
Ill
2
Cnd.B Round Station AMF
1 1
SMS Cgd.C Round Station ROR
13 1 1
SMS
,I
6
4
1
1
2
3
2
4
1
1-2
2
3 II
: r
5 1-2 4 4 2 4
8 2 5 4 2 5 I'
4 2 6 7 2-3 6 ''
I I
'I
10 2-3 7 8 3 7 II
3
4
3
3
8
9
15
4
3-4
4
8
9
i i
I i
6 3-4 10 3 5 10 blind blind 'I
5 4 11 blind 4 5 11 X
2 4 12 X 8 5 12
7 4-5 13 5 5-6 13 X
4 5 14 blind
6 5 15
4 6 16
5 6 17
8 6 18
4 6-7 19
5 7 20
2 7 21 X
j
Tables 23, 24, 25, & 26: Results of the blind short duru experiment,
conditions Q, R, S, and T.
1
Cnd. Round Station ROR Cnd.R Round Station NOA
2 1 1 4 1 1 (
3 1 2 5 1-2 2 (
2 1 3 2 2 3
4 1-2 4 2 2 4
5 3 5 3 3 5
2 3 6 2 3 6
2 3 7 5 3 7
3 3 8 7 3-4 8
2 3-4 9 blin 4 4-5 9 blind
5 4 10
7 4-5 11 X
4 5-6 12
2 6 13
4 6 14
3 6-7 15
iI
I',
I,
Day one
"'
-{ It
1,,
~· I'
~-
'I
136
L
137
Day two
7; (14) 1 3R; (15) 5L 6L; (16) 6L 4R; (17) 2 6L; (18) 4R 2; (19) 3R 7; (20) 5L
6L; (21) 4L 2; (22) 5L 3R;
(23) B4L a3R; (24) B2L blL; (25) A2R (tak) a8L; (26) B8R b5L; (27) B1L a4R
(tak); (28) A5L b8R; (29) A4R a7L; (30) B1L b6R; (31) A2R a5L (tak); (32) A2R
b1R (tak); (33) ASR b8L (tak) (This is not takasia since there are mtaji's possible on
more than two opposite holes); (34) A7L (tak) b2L; (35) A2R a3R; (36)A1R b1L;
(37) B8R b5L; (38) A7L b7R; (39) A6L b8R; (40) B7L ... Bao hamna! (A) wins.
somewhat overplayed.) (28) A4L (tak) a8L; (29) B1L · a5L (tak); (30) A2L (tak)
139
a2R; (31) B3R (tak) b6L; (32) B6R a7L; (33) B8R b1L; (34) A7L (tak) a2R; (35)
A2R b3L; (36) (A) resigns.
'I
Ramadhan (A)- Abdu (a) 12-10-94
(1) A7L (tak) a5R; (2) 6L (tak) 5R; (3) 2 6L; (4) 1 5R; (5) 3R 1; (6) 5R 3L; (7)7
3L; (8) 2R (tak) 5L; (9) 5R 3R; (10) 3L 3R (tak); (11) 3L 4R; (12) 4R 6R; (13) 5R
3L; (14) 4R 2; (15) 5L 4R; (16) 5L 6R; (17) 4R 8; (18) 6L 8; (19) 5R 2; (20) 6R
5L; (21) 6R 2; (22) 5R 7;
(23)B3L b2L; (24) B7R a4L (tak); (25) A5R b7L (tak); (26) A2R BSR (tak); (27)
B8R -(a) resigns.
Day three
Day four
more than 5 minutes here.) 2; (21) 2 4R (This move takes more than 4 rounds); (22)
7 4R;
(23) A3R b2L; (24) B8R a4L (tak); (25) A8L b1L; (26) B5L b7R; (27) A5L (tak)
a4R; (28) A7L (tak) a5L; (29) B1L b1L; (30) A3L b7R; (31) A6L a6R; (32) A1R
(Thinking time about 5 minutes.) b4R; (33) A2R b6R; (34) A4R (tak) a6L; (35)
B7R (Thinking time about 4 minutes) a6R; (36) A7L b2L; (37) A5L b1L; (38) B1L
b7R; (39) B2L a4L; (40) A1R b8R; (41) A4L a2R; (42) B4L (tak) a3R; (43) B1L
b4L; (44) A1R a3R; (45) A2L (tak) b7R; (46) A7L (tak) b3L; (47) A6L a7L; (48)
A4R (tak) b8R (In this move, Kijumbe makes a mistake in his back row. He skips
b4 when spreading seeds, which makes the move end in a8 instead of b8. Since he
'corrected' himself at that point, he either corrected himself wrongly, or he made a
miscalculation and thought the move had to end at a8. The influence of the mistake
is more significant than mistakes encountered in other matches.) (49) B8R a6L; (50)
A5R a8L (This move would have been a takasa move had his previous mistake not
been made.) (51) A 7L b1R; (52) (A) resigns.
Day five
+
l
145
8L (tak) 3R (tak); (10) 7R (tak) 6R (tak); (11) 8L (tak) 3R (tak); (12) 7R (tak) 7R
(tak); (13) 8R (tak) 8; (14) 5L 7; (15) 1 7; (16) 4L 8; (17) 4L 6L; (18) 5R 7R
(tak); (19) 1 7; (20) 8 2; (21) 4L 1; (22) 5R 1;
(23) A5L (tak) a5L; (24) B6R b6R; (25) B8R b8L; (26) A6L (tak) a4R; (27) A3L
b1L; (28) B1L b2L; (29) A2R (tak) b3L; (30) (A) resigns.
_;;,L__ -_
146
(23) A7L b7R; (24) B8R b5L (tak); (25) B6R b4R (tak); (26) A3R b1L; (27) A3R
a2R; (28) A5R a5L; (29) A6L a4L; (30) B4L b8R; (31) A4R a5R; (32) A6L b7R;
(33) B7R b1L (tak); (34) A3L b8L (tak); (35) B3L - (a) resigns.
Day six
3L (tak); (15) 3L 1; (16) 3R 2; (17) 1 5R; (18) 8 8; (19) 3L 8; (20) 8 4L; (21) 4R
4L; (22) 3R 4R;
(23) B2L b3L; (24) (A) resigns.
Day seven
t1
150
(23) A6L (tak) a6R; (24) B1L a6L; (25) A3R (tak) b1L; (26) B7R a2R; (27) A5L
b6R; (28) B3L b8R; (29) B5L (tak) - Shamte points to B4R (= Bao hamna) and
Abdu (A) resigns.
Day eight
~l
----------------------------------~~
154
(15) 4L 6R; (16) 1 6R; (17) 6L 2; (18) 4L 6L; (19) 8 2; (20) 8 6R; (21) 1 2; (22)
5R 5R;
(23) B2L b8R; (24) A2L a2R; (25) A4R a5L; (26) B3L b2L; (27) B8R b2L (tak);
(28) B1L b6R; (29) A1R -(a) resigns.
Day nine
-----------------------.L~
APPENDIX III: TIME MANAGEMENT AND POSSIBLE MOVES
23.25 5 5L 24.00 4 3L 18
24.20 7 5L 28.50 8 7 19
31.00 4 6L 34.25 4 1 20
37.25 7 3L 40.10 2 5L 21
40.10 6 6R 42.40 5 2 22
Mtaji
44.40 2 A6R 49.45 3 b7R 23
49.35 2 A5R 50.00 1 a8L 24
50.05 1 A6L 50.15 2 a4R 25
51.20 3 B7R 54.45 5 b6R 26
55.20 4 B8R 57.45 2 b8R 27
57.50 2 A3R 58.25 2 b7R 28
59.05 4 A5L (tak) 1.00.20 h/m/s 6 a7L (tak) 29
1.04.30 h/m/s 5 B4L 1.04.35 1 b2L 30
1.07.50 4 B3R 1.08.10 1 a2R 31
1.08.30 6 A8L 1.08.50 1 a6L 32
1.09.12 3 A5R 1.09.20 1 a4R 33
1.09.55 3 B7R 1.11.35 4 a6R (tak) 34
1.14.10 3 A6L 1.14.20 2 a5R (tak) 35
1.14.40 2 B2L 1.14.50 1 b1L 36
1.15.10-1.15.50 3 B5R resigns 1 (b7R, A1R, Bao hamna!)
I
I Duru Gambit, see Sacrifice
1I Experiment, see Experiment Gambling, 106
li Expert,82-84,95,101,118 Game-tree, see Complexity
~I Dynamic constraint, 3, 68, 86, 88, Genius, 68, 110
It:· 92,95,122 Go, 1,25,26,48, 122
Errors, see Irregularities God, 107
Ethnological theory, 1 Handicap-play, 46
Experience,2,20,21,83,123 Heko!, 105, 124
Experiment, see Table of Contents Hierarchy, 23, 35, 36, 39-42, 44,
Blind short results, 93, 134 46,47,101
Blind simple results, 93, 133 History
Complex, 95-100 Bao, manqala, 7, 8
Rejesha, see complex Bao masters, 10
Short duru results, 85, 132 Bao Society, 8, 9
Simple duru results, 72-75 Masters Tournament, 12-15
Expert, 2, 4, 32, 44, 48, 66-68, 79, Zanzibar, 8
92-94,117,119,121 House,6,34,36,38,40,42,45,
Expertise, 66, 68, 95, 121 46,63,64, 70,114,117,125,
I Exponential increase, 76, 87 135,157
Fani, 2, 100, 101, 110-119, 122, Human weakness, 2, 118
I 124 Ideal model, 87, 93, 95
Jenga nyumba, 124 Illegal deceit, see Deceit
Kitakimbi, 112, 115, 124, 157 Intelligence, see Masters
Kitakomwe, 63 (fani), 104 Interference, 67, 86, 87
(deceit), 108 (deceit), 114 Irregularities, 43, 104, 140, 143,
(fani), 125 144, 146, 154
Lazimisha, 112, 125 Islam,
Lengo,111, 125 Koranschool, 17, 22
Vunja lengo, 114, 125 Muslim law, 19, 88
Miliki, see weka Kenya, see Lamu
Namua nyumba, 114, 125 Khatima, 62, 101, 110 ,111, 118,
I
~~II
Opening play; 52, 115, 116, 119, 121, 124
119 Kikao, 8, 11, 124, see Clubs
Takasa nyuma, 114, 125 Kiswahili, see Swahili
~-'
i-1!
: Takasia, 33, 34, 41-46 (rules), Kitakimbi, see Fani
··'
J. 113,126,135,138,141,147, Kitakomwe, see Fani
~i
151 Knowledge, 2, 119
Tapisha Bao, 112, 113, 126 Koranschool, see Islam
Tega,64, 111,114,126,140, Lamu, 35, 104
j 149 Lazimisha, see Fani
Tegua,64,111,114, 126 Legal deceit, see Deceit
J
I~
I ,~
VVeka, 112, 125, 126
Finger movement, 70, 87
Forced moves, 52, 137, 146, 153,
Lengo, see Fani
Length of Bao game, see Average
Limitations, Limits, see also
'!
, I
157 constraints, 2, 3, 23, 48, 68, 76,
·? long flank, 59 79,92, 100,121,123
.1,
I ,~
short flank, 58 Duru, see Experiment
;i. Table 3, 60-62 Fani,_see Fani
'I
!.
Fork attack, 63, 136, 151 Khatima, see Khatima
,t' Four-row game, 7, 30 Qualitative, see Fani
'I
Ji
~
161
Bao is a board game played in East Africa that is a member of the manqala family of
games. Regular championships are held in Bao, especially on Zanzibar. These
championships have produced Bao masters, who form the object of this study. Bao
masters appear to be capable of calculating extraordinarily complex moves, difficult
for humans to emulate but easy to calculate with present-day computers. The aim of
this study is to determine those characteristics that are particular to mastership in
Bao. This will enable us to gain insight into a limitation of the human mind
identified in a game at which Bao masters excel, eventually presenting a description
of mastership in terms of human memory and problem-solving limitations. In order
to understand these characteristics and their relevance to Bao it will be necessary:
1. to understand the historical and geographical context within which the game is
played;
ii. to do cumen t and structure the rules of the game;
iii. to collect sufficient material to document Bao masters at play, illustrating
where human mental limitations constrain the way masters play;
iv. to assess empirically the limitations of human calculating and problem-solving
skills appropriate to Bao.
In Chapter II, the history of Bao masters is described and a list of all the Bao
masters living today in Zanzibar is presented. It appears that no more than ten
masters have been active at any given time in this century. Professions and
educational backgrounds vary widely among both the younger and the older
generations. Bao mastership was, therefore, impossible to link to any particular
skill, whether skills learnt at school, or social skills that make people successful in
society. The only conclusion that could be drawn was that masters appear to reach
mastership in their early twenties, starting in their teens. This confirms earlier
findings on Chess masters collected by de Groot (1946). .;1I
In the third Chapter an attempt is made to give a systematic classification of
board games based on the rules of the games. In that respect, Bao is a war game.
Bao is also a manqala game- a family of games in the classification by Murray
(1952) - these are games with a particular way of moving pieces. Such a move is
known as spreading, where a number of seeds, for instance, are spread one by one
into consecutive holes in a certain direction. Wari and Songka are games related in
that respect, Wari also being played in championships and even programmed to be
played by computers.
The rules of Bao, however, appear much more complicated than in any other
manqala game. The structure of the rules is presented in detail, and the hierarchy of
these rules shows that Bao is complicated but systematic in the application of the
rules. It appears that rules for championship Bao still develop by way of consensus
among the masters. It, therefore, occurs that not all the rules of Bao are agreed
upon. The rules concerned seldomly occur but the decision on their implementation
is essential to the definition and refinement of a championship game.
In order to illustrate Bao moves, one chapter is dedicated to examples of moves
and combination moves. They show the very basic moves used in Bao and conclude
with the identification of a first limitation to human thinking, which is a forced
move, known by some masters, where the outcome can be predicted to up to
seventeen turns ahead. Masters only calculate about eight turns ahead and this
165
distinguishes him from other players and from other masters. Each player is,
therefore, limited both in the number of skills and in the skills themselves.
The study is concluded with a description of league games, played in the 1994
Masters Tournament, where masters participated. From this list of games various
examples confirm and illustrate the findings in this study.
SAMENVATIING
Baa is een Oost-Afrikaans bordspel en behoort tot de familie van manqala spelen.
Er worden regelmatig Bao-kampioenschappen georganiseerd, met name op
Zanzibar. De Bao-meesters die deze kampioenschappen hebben voortgebracht,
vormen bet onderwerp van deze studie. Bao-meesters blijken in staat om bijzonder
complexe zetten te berekenen, die moeilijk zijn te bevatten voor mensen maar
eenvoudig zijn te berekenen met de hedendaagse computer. Het doel van deze studie
is dan oak om inzicht te krijgen in de begrenzingen van bet menselijk brein die bij
dit spel aan de orde komen. De karakteristieke eigenschappen van meesterschap in
Baa worden vastgelegd in termen van menselijk geheugen en probleem-oplossings
beperkingen. Om deze karakteristieken te begrijpen is bet noodzakelijk om:
Het vierde hoofdstuk is gewijd aan het illustreren van (combinatie-)zetten in het
Bao-spel. Eenvoudige zetten worden uitgebreid gei:llustreerd; aan het eind van het
hoofdstuk wordt een menselijke rekenlimiet vastgesteld bij een gedwongen
zettenreeks, waarbij de uitkomst voor die meesters die bekend zijn met deze situatie,
wel 17 beurten vooraf kan worden voorspeld. Aangezien Bao-meesters niet meer
dan acht beurten vooruit kunnen denken moet zo'n situatie uit het hoofd geleerd
worden, daar geen meester deze reeks van beurten kan berekenen.
Het volgende hoofdstuk introduceert de vaardigheid van het berekenen van duru.
Deze vaardigheid blijkt aan de basis te staan van alle andere vaardigheden in het
Bao-spel. Een Bao-zet kan uit verschillende rondes bestaan, dat wil zeggen dat
zaadjes verspreid kunnen worden over de rijen van een speler en meermalen rond
kunnen gaan. Dum is een Swahili term dat 'ronde(s)' betekent. Een speler met duro
is een speler die vele rondes kan doorrekenen. Een experiment laat zien hoe
meesters vordering maken tijdens het berekenen van duru, waarbij het tijdsverloop
en de maximale prestatie worden vastgelegd. Het blijkt dat meesters onderling nogal
verschillen maar gemiddeld zo'n vier rondes door kunnen rekenen, dat wil zeggen
vier rondes van zestien holletjes met opeenvolgende verspreiding van zaadjes
volgens de regels van het spel. Hoe verder zij komen hoe meer tijd zij no dig hebben
om de volgende stap te berekenen.
Verschillende varianten van dit experiment zijn nodig om de variabelen te
identificeren die meesters en prestaties vergelijkbaar maken. Twee modellen zijn
vervolgens ge'introduceerd die de oplosmethodes van de meesters beschrijven. Het
allocatie model, gebaseerd op de ervaring van de onderzoeker, en het ideale model,
gebaseerd op Miller's (1956) magische nummer zeven plus of min twee, vormden
twee mogelijke verklaringen voor de prestaties. Een tweede experiment werd
ontwikkeld om deze modellen te toetsen. De ronde van zestien holletjes werd
gehalveerd tot een ronde van acht holletjes. Het allocatie model voorspelde dat de
prestaties van de meesters bij het doorrekenen van kleinere rondes zou
verslechteren. Het ideale model voorspelde dat de prestaties ongelimiteerd goed
zouden zijn, ofwel een perfecte prestatie zouden opleveren. Toen het bleek dat het
allocatie model accurater was, bleek ook dat Miller's beperking ·qp het menselijke
korte termijn geheugen, het magische nummer zeven plus of min twee, niet volstond
voor de verklaring van de prestaties van de Bao meesters. Hierom is een extra
beperking op het korte termijri geheugen geintroduceerd, de dynamische beperking,
die snelle veranderingen van gebeugen-informatie in het korte termijn geheugen
verbiedt.
Het bovenstaande experiment wordt vervolgens uitgevoerd op zo'n manier dat
bet de meester verhinderd wordt om hun berekeningen te recapituleren of te
herhalen gedurende het experiment. Deze variant geeft meer inzicht in de invloed
van de verschillende variabelen ·op de men tale beperkingen. Complexe duru-
vaardigheden zijn eveneens belangrijk in het spel; niettemin wordt het experiment
van rejesha Bao, dat bestaat uit het reconstrueren van gecompliceerde zetten naar de
oorspronkelijke positie, niet als noodzakelijke vaardigheid voor kampioenschaps-
Bao gezien en wordt daarom ook niet door elke meester beheersd.
Ten slotte heeft een meester specifiek getraind voor het spelen van blind Bao, dat
wil zeggen dat hij kan Bao-spelen met zijn rug naar bet bord en de zetten alleen
communiceert door middel van woorden. Op deze manier is hij in staat gebleken om
44 beurten te berekenen die uit een significant grater aantal positieveranderingen
bestaan dan zijn beste prestaties op het duro experiment. Blijkbaar kan een meester
bekende combinaties en gestandaardiseerde openingen herkennen of in grotere
169