Professional Documents
Culture Documents
Toward Learning Organizations:: Integrating Total Quality Control and Systems Thinking
Toward Learning Organizations:: Integrating Total Quality Control and Systems Thinking
Total Quality Control and systems thinking have complemen- Juran first spoke those words in 1981, and we see in ret-
tary strengths that can greatly enhance an organization’s abil- rospect that he was overly optimistic. The Japanese overtook
ity to improve its performance through a more balanced learn- the U.S. sometime in the 1980s by improving their manufac-
ing process. As Daniel Kim explains below, the integration of turing capabilities at a faster rate than the U.S.—and they
the two approaches can provide the synergistic boost that will show no sign of slowing down. Ray Stata, chairman and
help U.S. firms reassert their competitiveness and build the
foundations for a new type of organization—a learning organ-
ization, where front-line people work in self-managed groups,
managers develop their research skills and take on the role of
theory-builders, and leaders become more like philosophers
who inspire the human spirit. At the core of these learning
Contents
organizations will be learning systems and processes firmly Total Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
rooted in the two disciplines of TQC and systems thinking.
Organizational Learning . . . . . . . . . . . . . . . . . . . . . . 2
Operational and Conceptual Learning
Japan’s success in wielding Total Quality Control1 (TQC) as a Systems Thinking and Organizational Learning
competitive weapon jolted American managers out of their
complacency and illustrated the need for drastic changes in
Beyond TQC: Systemic Quality Management . . . . . .6
management practice. We have had to discard old ideas that
cost and quality are an either/or decision, recognize the need Tackling Organizational “Messes”
for closer worker involvement, and assume that management The “Laws” of Systems Thinking
is responsible for defects arising from faulty systems.2 The
The Ten Tools of Systems Thinking
Japanese, however, began their TQC activities in the early
1950s and took nearly four decades to attain their current level The Manager’s Dilemma: Reflection vs. Action
of worldwide prominence. Companies in the U.S. often over-
look this fact as they embark on TQC programs of their own.
From Learning Systems Toward
Many firms have discovered just how painfully slow the
Learning Organizations . . . . . . . . . . . . . . . . . . . .12
process of instituting TQC can be: Time frames of eight to 10
years are quite common. Joe Juran, an internationally recog- Managerial Microworlds as Learning Systems
nized expert on Total Quality Management, underscored this Toward Learning Organizations
point at a European conference on quality control addressing
Managers’ New Roles: Researcher and Theory-Builder
the subject of “When Can the West Catch up with Japan”:
Japan has done its QC education well. But it took ten
years for this education to show results, for quality to Toward Learning Organizations:
improve, and for productivity to rise. No matter how hard Integrating Total Quality Control and Systems Thinking
Western nations try to engage in QC education, they may @1997 by Pegasus Communications, Inc.
not catch up with Japan until the 1990s, since it requires All rights reserved. No part of this publication may be reproduced or transmitted in
any form or by any means, electronic or mechanical, including photocopying and
ten years for the QC education to take effect. recording, or by any information storage or retrieval system, without written
permission from the publisher. For additional copies contact:
IMS003E permissions@pegasuscom.com
Toward Learning Organizations
2
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
president of Analog Devices, presented performance of a particular system. It often synonymous with individual learn-
a provocative argument that, unlike can involve physical changes in a ing, for it usually involves a small group
“the Boston Consulting Group’s experi- machine setting, procedural changes in of people and has minimal structure. As
ence curve theory, which says that a production step, or a psychological an organization grows, however, a dis-
learning is a function of cumulative change in a worker’s attitude about his tinction between the two levels of learn-
volume, independent of time . . . learn- or her job. Learning at the conceptual ing emerges. Somewhere in that process,
ing, properly managed, occurs as a level means changing one’s mental a system for capturing learnings from its
function of time, independent of models about how the world works. It individual members evolves. The way in
cumulative volume.” includes changing the way one thinks which an organization learns through
This statement carries a message far about a problem by reframing it in a individuals is a topic of growing interest,
more startling than Juran’s prediction: new context and exploring the implica- but has not yet reached consensus. One
We will never catch up with the tions. Learning at one level without the of the main dilemmas shared by all who
Japanese because they have a perma- other is like trying to run a marathon tackle this issue was posed by Chris
nent head start on us. In other words, with one foot nailed to the starting line; Argyris and Donald Schön:
we can’t catch up by simply imitating you can be off to a quick start, but you There is something paradoxical here.
what they have done; we must innovate won’t get very far. Organizations are not merely collec-
and improve upon TQC. tions of individuals, yet there is no
Peter Drucker wrote in 1990 about organization without such collec-
Organizational tions. Similarly, organizational learn-
the “postmodern factory” whose essence
Learning4 ing is not merely individual learning,
will be defined by the synthesis of four yet organizations learn only through
principles and practices—Statistical All organizations learn, whether they the experience and actions of individ-
Quality Control (SQC), new manufac- consciously choose to or not—it is a fun- uals. What, then, are we to make of
turing accounting, “flotilla” or module damental requirement for their sustained organizational learning? What is an
organization of manufacturing processes, organization that it may learn?
existence. The extent to which companies
and a systems approach. According to consciously manage the learning process, Clearly, organizations learn through
Drucker, the integration of these four however, varies greatly. Some organiza- their individual members and, therefore,
concepts will build a new theory of man- tions are quite deliberate, developing are affected either directly or indirectly by
ufacturing in which every manager “will capabilities that are consistent with their individual learning. Argyris and Schön
have to learn and practice a discipline objectives. Others make no focused present a theory of action perspective in
that integrates engineering, management effort, and inadvertently acquire habits which organizational learning takes place
of people, and business economics into counterproductive toward achieving through individual actors whose actions
the manufacturing process.” Implicit in stated goals. For example, superstitious are based on a set of shared models. The
Drucker’s theory is the underlying goal of learning can occur when an experience is authors argue that most organizations
creating an environment conducive for particularly compelling but the connec- have shared assumptions that preclude
continual learning. tions between actions taken and out- challenging people’s attributions and that
Being TQC-driven means creating comes produced are misunderstood or provide self-sealing affirmation of those
such an environment by advancing con- far apart in space and time. Organi- attributions. Engaging in such actions,
tinuous improvement at every level of zations can also suffer from competency which Argyris and Schön refer to as sin-
the organization. From factory workers traps when they accumulate experience gle-loop learning, ensures that very little
to CEOs, the goal is to become better in an inferior procedure with which they learning occurs. The authors present an
learners. But to endure, organizational have had initial favorable results. In the alternate model of learning, double-loop
learning must advance on both the grip of this kind of trap, a company may learning, in which people openly inquire
operational and conceptual level.3 either fail to notice a superior procedure about the company’s prevailing assump-
Learning at the operational level entails or dismiss it as unattractive. tions and are, in turn, open to having
changing behaviors or methods of In the early stages of a company’s them challenged, tested, disconfirmed,
doing things in order to improve the existence, organizational learning is and replaced.
Toward Learning Organizations
3
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
After the devastation of World War II, activity that involved all divisions and all Computer and semiconductor man-
Japan had to rebuild its industrial base employees. ufacturers have also instituted TQC in
almost from scratch. With the help of QC Circles grew out of the impor- their organizations. Analog Devices, a
U.S. occupation forces, the Japanese tant role that workers played in the leading manufacturer of linear inte-
began applying the modern techniques of actual manufacture of products. The grated circuits, included Quality
quality control in rebuilding their indus- cornerstone of the QC Circle lay in edu- Improvement objectives in their strate-
tries. A group of engineers and scholars cation mixed with on-the-job applica- gic planning and bonus incentives. At
formed the Union of Japanese Scientists tion of tools. In addition, the idea of the Hewlett-Packard’s Lake Stevens, WA,
and Engineers (JUSE) to engage in QC Circle was based on a strong belief facilities, workers cut failure rates for
research and disseminate knowledge in voluntarism. This belief led to a slow their 30 products by 84 percent, and
about quality control. The concept of start, but the movement later mush- manufacturing time by 80 percent. In
quality control was introduced to Japan roomed rapidly as early success stories recognition of their crusade for quality,
in 1950 when JUSE invited Dr. W. spurred other companies to follow suit. Motorola received the first annual
Edwards Deming, a recognized expert in The results of Japanese TQC activi- Malcolm Baldrige National Quality
the field of sampling, to give a seminar ties require little elaboration. In the Award—the American equivalent of
on statistical quality control for managers space of only a few years, Japanese car Japan’s prestigious Deming Prize.
and engineers. makers grabbed an ever-increasing share Florida Power and Light became the first
Although Deming’s tools proved of the U.S. automobile market. Since the non-Japanese company ever to win the
valuable for production problems, man- early 1970s, Japanese car makers have coveted Deming Prize in Japan.
aging the process of getting workers to produced more and more fuel-efficient The tools and methodology of TQC
use them effectively was difficult. Dr. J. and higher quality cars. They have gained widespread acceptance because
M. Juran’s visit in 1954 shifted Japan’s soundly bested many U.S. industries they fit in with the traditional model of
quality control emphasis from the fac- such as steel, semiconductors, motorcy- problem solving, which is based on
tory floor to an overall concern for the cles, televisions, and cameras, and they reductionism and analysis. It is indis-
entire management. The initial concept threaten the entire U.S. consumer elec- putable that the application of tradi-
of Total Quality Control emerged from tronics industry. They have penetrated tional TQC tools7 to manufacturing has
that shift in thinking. every market they have entered with proven highly successful. Steadily, and
In TQC, all quality efforts are car- superior quality products in workman- with remarkable efficiency, the TQC
ried out with the purpose of improving ship as well as design. As one striking process has reduced defects, shortened
the product or service provided as seen example of their success, Japanese semi- cycle times, and increased throughput in
by the customer. The definition of cus- conductor manufacturers currently sup- the manufacturing process.
tomer includes internal as well as exter- ply 85 percent of the worldwide mem- TQC embodies both a holistic phi-
nal buyers. Quality control (QC), or ory chip market. losophy about the enterprise of running
quality assurance (QA), began with an Although it has taken some strong a business and a set of statistical tools
emphasis on inspecting-out defects and convincing, many U.S. manufacturing applied at the lowest levels of an organi-
evolved into the concept of controlling firms have embraced the TQC way of zation. It is this blend of the macro and
manufacturing processes in order to conducting business and have made the micro that makes it such a potent
prevent defects in the first place. This significant strides toward improving discipline. Without its philosophy, TQC
idea was later extended to include the quality. AT&T, for example, cut the would be simply a bag of tools applied
product development process—to development time for their model 4200 to mere firefighting, or solving problems
design-in quality from the very begin- cordless telephone from two years to only as they arise. Without its statistical
ning. The inclusion of the product one year while improving quality and tools, TQC provides a guiding light to a
development process produced the need lowering costs. All three U.S. auto goal but no way to navigate the terrain.
to involve the entire company in quality makers have undertaken TQC activities TQC’s success lies in linking the lofty
control activities. QC was no longer the that have helped reduce defects, cut goals for top management with a set of
province of inspectors performing an cycle times, and increase customer tools for operators to achieve those
isolated function, but a companywide satisfaction. goals.
Toward Learning Organizations
5
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
TQC is particularly well equipped to learning at a deeper level, where man- marketing and claims. The above exam-
advance learning at the operational agers can gain a better understanding of ples all lead to functional gridlock—each
level. The seven tools of TQC (Pareto their organization and improve the way function striving to optimize its own per-
chart, cause-and-effect diagram, stratifi- they manage. That is, TQC is valuable formance while the organization as a
cation, check sheet, histogram, scatter for enhancing learning at the opera- whole is grossly suboptimized.
diagram, and control charts) are rela- tional level but is limited in its ability to Systems thinking helps break through
tively easy to understand and use. advance management thinking at the functional walls of isolation by providing
Through the use of control charts and conceptual level. a framework for understanding the
Pareto analysis, for example, operators importance of managing the intercon-
can understand and improve their pro- nections among the various functions. It
Systems Thinking and
duction steps. Under the TQC umbrella, also provides a methodology for thinking
Organizational Learning
engineers can design experiments and about the ways in which prevailing men-
collect data on the factory floor to better Compared to TQC’s emphasis on oper- tal models may restrict learning, gaining
understand and improve manufacturing ational learning, systems thinking’s deeper insights into the nature of com-
processes. Improvement through opera- underpinnings are more conceptual in plex systems, finding high leverage points
tional learning involves an incremental nature. Systems thinking8 approaches in the system, and testing one’s assump-
process whereby a particular problem is problems from the basis of the whole, tions about the efficacy of various policy
whittled down bit by bit. At the heart of rather than breaking things up into choices.
this learning process is Deming’s PDCA individual pieces and trying to under- System dynamics is a field within
(Plan-Do-Check-Action) cycle, which stand each part. Where TQC focuses on systems thinking that is particularly
promotes continual improvement by analysis of the separate parts that make rich in the area of conceptualization
cycling through the PDCA problem- up the whole, systems thinking strives and synthesis of complex systems. It
solving loop. Transfer of learning from for synthesis of the constituent parts. also provides a methodology for syn-
individuals to organizations is managed According to systems thinking, if a thesizing disparate kinds of variables
through an organization-wide TQC system is decomposed into its compo- that have traditionally been considered
effort designed to facilitate the sharing nents and each component is optimized, too “fuzzy” to measure. By focusing on
of learning in one setting with the rest the system as a whole can be guaranteed making mental models explicit, expos-
of the organization. not to be optimal. A common character- ing them to challenge, and building
As quality continues to improve at istic of many complex systems is that new models based on insights gained
companies engaging in TQC activities, a they are often designed with the inten- from this process, system dynamics
great deal of learning doubtless contin- tion of optimizing the parts rather than helps managers gain a more systemic
ues to take place at the operational level. the whole. In a typical company, the view of their organization.
New methods of soldering a joint or manufacturing function is expected to In contrast to TQC, systems thinking
assembling an engine, for example, are operate as efficiently as it can. The same offers no simple tools to date that can be
tried, tested, and absorbed into the goal holds for marketing, accounting, used at the operational level to actually
organization’s memory. Aside from the engineering, and so forth. At a leading make the improvements indicated by a
initial mental breakthrough required at manufacturer of linear integrated cir- system dynamics study. Systems thinking
the outset of instituting TQC, however, cuits, initial attempts to use TQC to is often weak in specifying operational
new learning opportunities at the improve on-time delivery performance procedures or methods to effect a desired
conceptual level become less available. A led to each function jockeying to change in procedure or structure.
manager can go on advocating improve- improve its own performance measure. Managers may gain a terrific insight that
ments within the current framework of This outcome did not ultimately benefit easing time pressure is a high leverage
organization policies and traditions customers. In the insurance industry, point for reducing turnover and, in turn,
without gaining much insight about the underwriter departments tend to see increasing productivity, but systems
whole system. TQC provides limited optimization of their function as inde- thinking gives them little guidance for
methods and tools for organizational pendent from other functions such as how to accurately measure each of the
Toward Learning Organizations
6
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
variables, implement changes, and methods. The two approaches form a dynamics approach of gathering data,
monitor progress. synergistic pair whose individual conceptualizing, building a model, run-
strengths complement each other and ning simulation analyses, and proposing
provide a balance of operational and policy changes. An implicit assumption
Beyond TQC: conceptual learning (see “Systemic of this process is that the insights alone
Systemic Quality Quality Management Model”). Each would be compelling enough to produce
Management process informs and enhances the other. action. In reality, however, such policy
As I argued earlier, we must do more Together, they advance organizational change recommendations are seldom
than play “follow the leader” if we hope learning by helping to build a shared implemented, because building shared
to regain and sustain a competitive understanding of conceptual insights and understanding traditionally has not been
advantage in the global marketplace—we operational processes, and create a pow- part of the process. Clearly, more could
must innovate beyond TQC. Integrating erful new model I call Systemic Quality be done on implementation.
TQC and systems thinking can accelerate Management (SQM). The bottom box represents a typical
organizational learning beyond the cur- In the SQM model diagram, the top TQC process of quality improvements,
rent capabilities of traditional TQC box represents the traditional system the so-called PDCA (Plan-Do-Check-
Act) cycle, which should be carried out
at every level of an organization.
SYSTEMIC QUALITY MANAGEMENT MODEL Requests from a higher level are inter-
preted and translated into a plan of
action with the appropriate check
CONCEPTUAL LEARNING points identified for monitoring
Policy progress relative to the plan. The plans
Implementation
are incorporated into the budgetary
cycle and implemented. The check
Gather points identified earlier are tracked, and
Policy Change
Hard and Soft deviations are observed. The data is
Recommendations
Data Traditional model then analyzed, and actions are taken to
of system dynamics
organizational correct any discrepancies. Although the
intervention PDCA cycle can help in implementing
new requests given from above and in
Synthesis Conceptual
maintaining control over current
Insights
Analysis processes, it is relatively weak on identi-
fying the high-leverage areas that can
Shared drive the whole process.
Understanding
Combining these processes means
Check
integrating conceptual and operational
learning by blending the two into a
seamless process. For example, building
Total Quality
Control's Do shared understanding through the use
Analyze / Act
PDCA cycle of management flight simulators and
learning labs9 can enhance the
Plan PLANning and DOing steps by provid-
ing a common base of conceptual mod-
OPERATIONAL LEARNING els. Having greater shared understand-
ing can also facilitate buy-in of policy
change recommendations. Conceptual
Toward Learning Organizations
7
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
National Standard
Cross-Industry
ORGANIZATIONAL COMPLEXITY OF A "MESS"
of Living
Supplier-Distribution
Cross-Organizational Networks
ity
Cross-Functional
Product
plex
Development Cycle om
cC
mi
Dyna
Multiple-Intrafunctional
Manufacturing
sing
Cycle Time rea
Inc
Scrap and
Unifunctional Repair Costs
insights such as “eroding goals” and are inadequate. In particular, TQC is ill- (current cycle time of projects). “Fuzzy”
“worse-before-better behavior” can help equipped for tackling a class of problems variables often accompany Ackoff ’s
those involved in the DOing to see how that Russell Ackoff labels as a mess: “messes” because such systems are
their actions relate to the overall sys- What is a mess? That’s the significant rarely clear cut or well defined.
tem. The analysis and action produced thing—a mess is a system of prob- Organizational efforts that grapple with
through the PDCA cycle should gener- lems. Now, the significance of this is problems involving a high degree of
ate new data that would feed into the that the traditional way of managing organizational complexity, long time
data gathering process as well as the is to take a mess and break it up into delays, and fuzzy variables provide
problems and solve each problem
next cycle of the PLAN. opportunities for improvement and
separately, with the assumption that
Through SQM, organizations can the mess is solved if we solve each enhanced learning through SQM.
identify high-leverage points and act part of it. Organizational Complexity. Many
upon them. Because there is an abun- But remember . . . if you break a
initial improvements gained in most
dance of written materials available on system into parts and make every
part behave as effectively as possible, companies come rapidly and with rela-
TQC and limited systems thinking liter-
the whole will not behave as effec- tive ease, either because the situation
ature, I will concentrate on illuminating
tively as possible. Therefore, the solu- was so bad that almost any concerted
the ways in which systems thinking can
tion to a mess does not consist of the effort would have yielded quick results,
contribute to the SQM model.
sum of the solutions to the problems or because most early projects revolved
that make it up. And that is abso- around single functional units requiring
Tackling Organizational lutely fundamental. minimal cross-functional cooperation.
“Messes” “Complexity and Time Lags of These initial projects are often part of a
Although TQC seems to have been Organizational ‘Messes’” is an attempt bootstrapping strategy, advocated by
applied to everything from reducing to classify “organizational messes” into Juran, in which other functional units
defects to solving inventory control a matrix based on Organizational in the organization slowly buy into the
problems and designing customer- Complexity (the number of units and TQC philosophy after its value has been
oriented products, there are cases for level of complexity of their intercon- proven. This buy-in process progresses
which current TQC tools and methods nections) and Time Lags of Process slowly, taking longer and longer as it
Toward Learning Organizations
8
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
involves increasing levels of cross-func- means getting real-time data and ana- spans several months from the very
tional cooperation. lyzing it to see what the data has to say. beginning rather than retrospectively.
One study shows that the rate of The process step usually takes minutes
improvement in a wide range of TQC or hours, not days or months. Thus, it Fuzzy Variables. According to Juran,
projects is primarily a function of the is feasible to collect data and be confi- the three laws of TQC are “look at the
organizational complexity of the project, dent about causal conclusions drawn data,” “look at the data,” and “look at the
not the specifics of the project itself. This from the data. When the time delays of data.” Hard, measurable data is the fuel
finding isn’t too surprising if you listen a project are extremely long, such as a that drives the TQC engine. Getting
to Edward Baker, Ford’s Director of product development process that takes good, reliable data is not always easy,
Quality Planning and Statistical several years, running real-time experi- even on mechanical processes that have
Methods, explain the difficulties of com- ments becomes impractical, and current universally accepted units of measure
municating in top-down organizations: data is of limited usefulness. One can such as units per minute, or pounds per
tweak individual steps within the unit. The task of collecting data becomes
Consider an organization with six lev-
els below the senior executive and a process but cannot gain much insight much more difficult as the measured
span of control of three. This makes about the implications of this tweaking item becomes much less clear. Fuzzy
1,093 people. More importantly, there on the process as a whole. variables include such notions as Effect
are 586,778 potential two-person inter- of Time Pressure on Productivity, Effect
A research project studying the
faces that represent potential internal
dynamics of software development, for of Delivery Delay on Demand, and other
supplier-customer relationships. These
1,093 people depend on one another example, revealed that an emphasis on information that may be available only at
to get their job done, but their interde- accuracy in estimating project comple- the intuitive level. Issues stemming from
pendencies are not explicit. tion led to increased accuracy but at sig- fuzzy variables are likely to increase as
nificantly higher costs and longer com- the service industries begin looking for
Managing even a few of those sup-
pletion times. The research findings ways to apply TQC.
plier-customer relationships can be a
indicate that people’s decisions are In the case of Hanover Insurance, a
daunting task. Without a full under-
greatly influenced by the project sched- property and casualty insurance com-
standing of the interdependencies, people
ule itself. Thus, owing to the pressures pany, exploring the dynamics of man-
can take actions that produce undesirable
and perceptions that a given schedule aging a claim office highlighted the link
results. For example, managers at Xerox
produces, different estimates create dif- between the amount paid in claim set-
who bought into the concept of Just-in-
ferent projects. Using a system dynamics tlements and the quality of adjusting.10
Time (JIT) manufacturing “solved” their
model, researchers ran two different esti- By interacting with a decision-making
inventory problems by demanding that
mates of a 64,000-delivered-source- game (based on a system dynamics
their suppliers hold inventory until they
instruction software project. In the base model), claim managers experienced
were ready to accept it. By “injecting” a
simulation, Method A produced a 2,359 how their decisions led to the erosion of
good idea into one part of the system
person-day estimate. The simulated quality and resulted in higher settle-
without a full appreciation of the whole,
project actually consumed 3,795 person- ments. In the absence of hard, measura-
they severely strained the good supplier
days, leading to an error of 38 percent. ble data, the model framework allowed
relations that had been painstakingly
Method B’s estimate of 5,900 person- managers to visualize and experience the
developed over many years. In the SQM
days (150 percent higher than A’s esti- interconnectedness of such concepts as
model, building a shared understanding
mate) resulted in a project that con- time pressure, work intensity, time
about the whole system lies at the core of
sumed 5,412 person-days, which has a effectiveness, and quality standard.
organizational learning.
9-percent error factor. Although Method
Time Delays. Another attribute of B’s estimate appears more accurate, it The “Laws”
early TQC projects is that the time leads to a much more costly result—the of Systems Thinking
delays within those systems are rela- project consumes 43 percent more per- Systems thinking can help make sense
tively short. For example, reducing son-days. With such simulation testing, of messes by providing a framework for
defects at a specific production step one can gain insight into a project that understanding the interconnected
Toward Learning Organizations
9
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
nature of systems and how they interact is multiple “compensating feedback” tively easily. By translating sophisticated
in the short and long term. Over the relationships that attempt to maintain statistical methods into easily under-
years, certain systems principles have internal balances despite external inter- stood tools like control charts, his-
been identified that provide a frame- ventions. It is this feedback that pro- tograms, check sheets, and scatter dia-
work for understanding the dynamic duces the pattern of better behavior grams, JUSE has made Statistical Process
implications of a whole system without before worse behavior. The principle of Control accessible to the masses.
requiring a detailed knowledge of its “shifting the burden to the intervenor,” Similarly, a number of tools have
individual components. Peter Senge, for example, results from the intended been developed over the years for
author of The Fifth Discipline: The Art action of helping someone. However, it applying systems thinking to organi-
& Practice of the Learning Organization, also produces the unintended conse- zational “messes” (summarized in
states that complex systems are subject quence of undermining that individual’s the table on pages 10–11). The tools
to the following laws: ability to help him- or herself.12 The ele- fall under four broad categories:
● Today’s problems come from yester- phant analogy represents the notion of Brainstorming Tools, Dynamic
day’s “solutions.” “indivisible wholes”; that is, dividing an Thinking Tools, Structural Thinking
● The harder you push, the harder the elephant in half doesn’t produce two ele- Tools, and Computer-Based Tools.13
system pushes back. phants—it produces a mess. Although each of the tools can be used
● Behavior grows better before it A common characteristic of the separately, they also build upon one
grows worse. “laws” listed above is that they appear another and can be used in combina-
counterintuitive at first glance. This tion to achieve deeper insights into
● The easy way out usually leads
back in. is more than mere coincidence— complex systems.
counterintuitive behavior is
● Faster is slower.
fundamental to the nature of complex Brainstorming Tools. The table shows
● The cure can be worse than the
systems. Applying linear thinking and the double-Q diagram, which is based
disease.
static tools to nonlinear and dynamic on what is commonly known as a fish-
● Cause and effect are not closely
problems often leads to solutions that bone or cause-effect diagram. The Q’s
related in time and space.
produce tomorrow’s problems. Thanks stand for qualitative and quantitative,
● Small changes can produce big
to delays and multiple feedback loops and the technique is designed to help
results—but the areas of highest
leverage are often the least obvious. within a system, our well-intentioned users begin to see the whole system.
“cures” can often produce results that During a structured brainstorming ses-
● You can have your cake and eat it
too—but not at once. are much worse than the disease we sion with the double-Q diagram, both
sought to cure. In such situations where the “hard” and “soft” sides of an issue
● Dividing an elephant in half does
not produce two small elephants. our intuition is a poor guide, we need a remain equally visible. The diagram
new set of tools. also provides a visual map of the key
● There is no blame.
factors involved.
The above laws hold true because in The Ten Tools A double-Q diagram begins with a
complex systems cause and effect gener- of Systems Thinking heavy horizontal arrow that points to
ally are not close together in space or One of the greatest strengths of TQC lies the issue being addressed. Major quan-
time. Yet we often assume and act as if in its ability to educate people in using titative factors branch off along the top
they were. We take action on what the tools. Each of the Seven Tools of and major qualitative factors run along
appears to be the obvious culprit, only to Quality and the Seven Management the bottom. Arrows leading off of the
find that our solutions have exacerbated Tools for quality is well developed and major factors represent subfactors.
the very problems we intended to solve.11 documented. Under the guidance of These subfactors can, in turn, have sub
These unintended consequences are the JUSE (Japanese Union of Scientists and subfactors leading off of them. Many
result of complex systems’ tendency to Engineers), training in the use of the layers of nesting, however, may signal
resist changing their behavior. The most tools has been standardized so that large that a subfactor should be turned into a
common cause of such policy resistance groups of people can learn them rela- major factor. Although double-Q dia-
Toward Learning Organizations
10
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
From Learning tasks, research revealed that varying the involvement in conceptualizing the
Systems Toward strength of feedback (in either the posi- model may result in a shallow under-
tive or negative direction) had a detri- standing of the model’s dynamics,
Learning Organizations
mental effect on performance. which are gained primarily through
Managerial Microworlds trial-and-error experimentation in mul-
as Learning Systems Management Flight Simulators. A tiple game plays. A deeper understand-
There are three important elements of management flight simulator is a man- ing of the underlying structure requires
learning: a set of tools appropriate for ager’s equivalent of a pilot’s flight simula- more than repeated plays of the game;
the task or concept to be learned, a tor—with its analogous gauges, lights, participants need more explicit discus-
framework that provides a context for throttle, steering wheel, and so on. The sion of the theory underlying the MFS.
the learning, and a playing field in which first flight simulator developed at the Thus, the value of the flight simulator
to practice and learn with those tools. MIT Sloan School of Management was as a learning tool can be extended
Baseball teams have their equipment, the People Express Simulator.14 People through the development of an envi-
rules of the game, and a field on which Express has been used with entering ronment within which one can replicate
they can practice. Airline pilots have classes of masters students at the MIT the conceptualization phase of the sim-
flight simulators, grids of the air space, Sloan School of Management, who spend ulator development, design various sce-
and simulated flight conditions. No team a day engrossed in strategizing and oper- narios to highlight specific dynamic les-
would dream of playing a regular season ating the simulated airline.15 The simula- sons, and create an environment in
game without having practice games. tor requires each team of players to make which the simulator serves as an experi-
Airlines would never risk multimillion- up to five decisions on a quar- mental tool for learning.16
dollar airplanes and the terly basis as they try
lives of hundreds of to manage the Learning Laboratories. Extending the
passengers for a boundaries of the microworld beyond
Managers, on the other hand, growth of a
pilot to learn by start-up air- the computer simulator itself, the learn-
do not have comparable tools and
trial and error. line (whose ing-lab concept can be viewed as a
environments in which to practice
Managers, on structure is manager’s equivalent to a sports team’s
and learn—initiation by fire
the other hand, do modeled after practice session. The learning lab is a
is the rule.
not have comparable tools the now-defunct managerial “practice field” with an
and environments in which to prac- People Express). Spreadsheets, emphasis on team learning; in the lab,
tice and learn—initiation by fire is the graphs, and internal management reports one can test new strategies and policies,
rule. Experimental data on decision- containing competitor and market infor- reflect on the outcomes, and discuss
making performed within complex mation are provided on a quarterly basis. pertinent issues with others in the
feedback environments underscores the Through repeated trials of launching a group. In a learning lab, a management
need for a manager’s equivalent of a start-up using various strategies, the stu- team can accelerate time by simulating
pilot’s flight simulator. Professor John dents gain simulated experience of the a model (or virtual world) of a real-life
Sterman of the MIT Sloan School of dynamics that People Express actually system quickly and slow down the flow
Management has demonstrated how experienced. of time at each decision point to reflect
subjects are insensitive to feedback Although the management flight on their actions.
from their own decisions in a produc- simulator by itself can be valuable for The learning lab provides a unique
tion-distribution simulation game, bringing an experiential and dynamic forum in which participants can ques-
leading to grossly suboptimal behavior. aspect to an otherwise static case, its tion operating norms and assumptions
An analysis of managers’ decisions in an usefulness as a stand-alone tool has safely, via the game model. In an insur-
insurance claims game revealed misper- limits. For example, a player is strictly a ance company implementation of a
ceptions of the time delays involved consumer of the model on which the learning lab, for example, although the
with their decisions. In experiments on simulator is based, and does not partici- company itself emphasized pursuing
human control of stock-adjustment pate in the creation process. This lack of high quality standards, the behavior in
Toward Learning Organizations
13
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
the games showed that controlling thinking” and break away from out- standing can we focus on a small set of
expenses dominated people’s actions.17 moded frames and perceptions. high-leverage points to produce
One manager remarked that while play- changes that self-reinforce and endure.
ing the game, Toward Learning To acquire such an understanding
“I kept telling myself, ‘Don’t add to Organizations requires an ongoing management edu-
staff, don’t add to staff,’ even though Organizational learning is the root from cation process to develop a new style
there is no one telling me not to, and I which all competitive advantage stems. of thinking with the right blend of
know that I really need to!” In many The level of advantage depends on the analysis and synthesis—to be able to see
cases where there was extra speed and quality of the analogy of the two small elephants
people capacity, man- learning, in all complex systems. The unique
agers chose to either The learning-lab whether organizational realities of the ’90s and
cut staff or push for behavior beyond will require managers to take
environment helps develop
more production to change is on new roles and acquire new skills.
an inquiry mode of learning that accom-
reduce expenses
challenges managers to “think panied by
rather than to work Managers’ New Roles:
on improving quality. about their thinking.” cognitive Researcher and Theory-Builder
Another set of operating change, and
assumptions that were brought to the whether continual education As the role of workers increasingly
surface and challenged were the notions is emphasized over sporadic training. By becomes one of self-managing work
of “proper” workload and productivity. emphasizing the importance of trying to groups, the role of managers will
In one scenario, managers concluded understand a problem, not simply solve undergo redefinition as well. Given
that work backlog per employee and it, systems thinking attempts to trans- today’s pace of change and organiza-
productivity were such that they labeled form problem-solving organizations into tional complexity, managers need to
it a “country club,” or an office with too learning organizations. A learning organ- know how to apply the research skills of
little to do. Many responded by reduc- ization is one that consciously manages a scientist to better understand their
ing the number of personnel or push- its learning process to be consistent with organizations. The old paradigm of
ing for more production from each per- its strategies and objectives through an experiments in organizations being fed
son in order to cut expenses per claim. inquiry-driven orientation of all its into research institutions, receiving the
This practice inevitably led to increases members. That is, learning organizations output and feeding it back into the
in settlement dollars in excess of any actively and explicitly encourage the organizations, is no longer adequate.
savings in expenses. Most participants learning process to ensure that areas of Intragenerational change means that
acknowledged that they had made their strategic importance are not neglected. the research cycle must happen faster
decisions in the game (and admittedly, This is accomplished, in part, by embed- than ever, otherwise solutions (in the
in real life as well) on the basis of ding learning systems that foster the con- form of research results) will be still-
assumed acceptable numbers, without tinual enhancing of the organization’s born; the problems they were address-
questioning the decisions’ appropriate- capabilities through its members. ing will no longer be relevant. The
ness for a specific situation. Management flight simulators and learn- dichotomy between manager and
In the learning lab, the process of ing labs are two such learning systems. researcher must end, because the pace
making operating assumptions explicit Building a learning organization of change is such that one can no
and testing those assumptions encour- means continually developing the longer separate the two functions—
ages managers to reflect not only on the capacity to create one’s vision of the managers must wear both hats
decisions they make, but on the process future. Designing and implementing simultaneously. Ed Baker’s proposal
by which they make those decisions. The effective policies to create desired that the CEO’s new role should be that
learning-lab environment helps develop results requires an understanding of an of Head of Research and Development
an inquiry mode of learning that chal- organization and its environment as a for the Enterprise is the type of leader-
lenges managers to “think about their unified system. Only with this under- ship needed for supporting such a shift.
Toward Learning Organizations
14
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
Managers must also become theory- who are interested in building Peter curve theory. (The case of Texas
builders within their organizations. Drucker’s “postmodern” factory, for Instruments and the personal-computer
They must create new frameworks example, must translate his ideas into a debacle is one example.) Others who fol-
within which they continually test their framework that can be applied in their lowed the BCG business-portfolio theory
strategies, policies, and decisions so as own organization. As theory-builders, also suffered their share of problems by
to inform themselves of improvements managers must have an intimate knowl- either giving up entire markets or not
on the organization’s design. It is no edge of how their organization works taking full advantage of synergies among
longer sufficient to apply generic theo- together as a whole. their different businesses. Theory build-
ries and frameworks like Band-Aids to There is no “golden formula” ing should not be done as an academic
one’s own specific issues. Managers that will hold for all time, or even for exercise but as a process, grounded in
must take the best of the new ideas one’s tenure in a present position. reality, that continually helps provide a
around and then build a workable the- Companies who lived by the learning- framework for interpreting one’s com-
ory for their own organization. Those curve theory almost died by the learning- petitive environment.
References Dumaine, Brian (1989) “How Managers Can Morecroft, John D. W. (1988) “System Dynamics
Succeed Through Speed,” Fortune, February and Microworlds for Policymakers,” European
Abdel-Hamid, T. K. and S. E. Madnick (1986) 13, pp. 54–59. Journal of Operations Research, 35, pp.
“Impact of Schedule Estimation on Software Fiol, C. M., & Lyles, M. A. (1985) “Organizational 301–320.
Project Behavior,” IEE Software, July, Learning,” Academy of Management Review, Ozawa, Masayoshi (1988) Total Quality Control
pp. 70–75. 10 (4), 803–813. and Management. Tokyo: JUSE Press.
Ackoff, Russell L (1981) Creating the Corporate Forrester, J. W. (1971) “Counter-Intuitive Behavior Roberts, Nancy (1983) “Testing the World with
Future: Plan or Be Planned For. New York: of Social Systems,” Technology Review, 73 (3), Simulations,” Classroom Computer News, 3
Wiley & Sons. pp. 52–68. (3).
Alfeld, Louis Edward and Alan K. Graham (1976) Garvin, D. (1988) Managing Quality. New York: Schneiderman, Arthur M. (1988) “Setting Quality
Introduction to Urban Dynamics. Cambridge, Free Press. Goals,” Quality Progress, April, pp. 51–57.
MA: Allen–Press. Hedberg, Bo (1981) “How Organizations Learn Schön, Donald (1971) Beyond the Stable State.
Argyris, Chris & Donald Schön (1978) and Unlearn,” In Nystrom, P.vC. & Starbuck, London: Temple Smith.
Organizational Learning: A Theory of Action W. H. (eds.) Handbook of Organizational _____. (1983) The Reflective Practitioner. New
Perspective. Reading, MA: Addison-Wesley. Design. London: Oxford University Press. York: Basic Books.
Baker, E. M. (1989) Dialogue, Winter. Hutchins, D. (1986) “Having a Hard Time with Schonenberger, R. (1982) Japanese Manufacturing
_____ (1989) “The Chief Executive Officer’s Just-In-Time,” Fortune, June 9, pp. 64–66. Techniques. New York: Free Press.
Role in Total Quality: Preparing the Ishikawa, Kaoru (1982) Guide to Quality Control. Senge, Peter M. (1989a) “Organizational Learning:
Enterprise for Leadership in the New Ann Arbor, MI: UNIPUB. New Challenges for System Dynamics,”
Economic Age,” Conference on Quality, _____ (1985) What Is Total Quality Control? D–memo 4023, System Dynamics Group,
Madison, Wisconsin, April 19–21. Englewood Cliffs, NJ: Prentice-Hall. MIT Sloan School of Management,
Bakken, Bent (1989) “Learning in Dynamic Juran, Joseph M. (1964) Managerial Breakthrough. Cambridge, MA.
Simulation Games: Using Performance as a New York: McGraw-Hill. ______ (1989b) “Catalyzing Systems
Measure,” Computer-Based Management of _____ (1985) Management of Quality. Juran Thinking Within Organizations,”
Complex Systems: Proceedings of the 1989 Institute. Organization Development, F. Masarik, Ed.
International Conference of the System _____ (1974) Quality Control Handbook, 3rd ______ (1990) The Fifth Discipline, New
Dynamics Society. Edited by P. Milling and E. Edition. New York: McGraw-Hill. York: Doubleday.
Zahn, Berlin: Springer-Verlag. _____ (1982) Upper Management and Sherkenback, William W. (1986) The Deming
Canegelosi, Vincent E. & William R. Dill (1965) Quality, Fourth Edition. Wilton, CT: Juran Route to Quality and Productivity.
“Organizational Learning: Observations Institute, Inc. Washington, DC: Cee Press
Toward a Theory,” Administrative Science Kim, Daniel H. (1989) “Learning Laboratories: Shiba, Shoji (1989) Lectures given at
Quarterly, 10, 175–203. Designing Reflective Learning Massachusetts Institute of Technology, May
Crosby, P. B. (1979) Quality Is Free. New York: Environments,” Computer-Based Management 9, 11, & 12.
McGraw-Hill. of Complex Systems: Proceedings of the 1989 Stalk, George (1988) “Time: The Next Source of
Daft, R. L., & K. E. Weick (1984) “Toward a Model International Conference of the System Competitive Advantage,” Harvard Business
of Organizations as Interpretation Systems,” Dynamics Society. Edited by P. Milling and E. Review, July–August, pp. 41–51.
Academy of Management Review, 9, 284–295. Zahn. Berlin: Springer-Verlag. Stata, Ray (1989) “Organizational Learning: The
Davis, Stanley (1987) Future Perfect. Reading, MA: Levitt, Barbara & March, James G. (1988) Key to Management Innovation,” Sloan
Addison-Wesley. “Organizational Learning,” Annual Review of Management Review, 30 (3).
Deming, E. D. (1986) Out of the Crisis. Sociology, 14, pp. 319–340. Sterman, John D. (1989) “Modeling Managerial
Cambridge, MA: MIT Center for Advanced Mizuno, Shigeru, ed. (1988) Management for Behavior: Misperceptions of Feedback in a
Engineering Study. Quality Improvement. Translated by Dynamic Decision Making Experiment,”
Diehl, Ernst (1989) “A Study of Human Control Productivity Press, Cambridge, MA. Original Management Science, 35 (3), pp. 321–339
in Stock-Adjustment Tasks,” System Japanese edition, Kanrishi to Sutaffu no Shin-
Dynamics Group D–memo 4030, MIT Sloan QC-nanatsu-dogu, published by JUSE Press,
School of Management, Cambridge, MA. Ltd., Tokyo, Japan, 1979.
Drucker, Peter (1990) “The Emerging Theory of Moissis, Alexander A. (1989) “Decision Making in
Manufacturing,” Harvard Business Review, the Insurance Industry: A Dynamic
May/June. Simulation Model and Experimental Results,”
unpublished Sloan School of Management
Master’s Thesis at M.I.T., Cambridge, MA.
Toward Learning Organizations
17
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com
Acknowlegements
I am grateful for the insightful and thought-provoking comments of Peter Senge,
which were instrumental in shaping my thoughts on this paper. Comments and
feedback from John Sterman, Alan Graham, and Bent Bakken on earlier drafts are
also greatly appreciated. Toward Learning Organizations: Integrating Total Quality
Control and Systems Thinking was edited by Lauren Johnson and Kellie Wardman
O’Reilly and designed by Cia Boynton.
Daniel H. Kim is cofounder of the MIT Center for Organizational Learning and
Pegasus Communications, Inc., and publisher of The Systems Thinker®. This
volume was formerly published as part of the Special Report Series, and has
been reformatted for this edition.
Toward Learning Organizations
18
Pegasus Communications, Inc. (tel) 781-398-9700 www.pegasuscom.com