Turin

:.~-J;~'~-~
• '11

IS
Digital
fi:.e:ality IS digi'tal • Tur[ngl comp..._ti:lbilit~
rutes,
·i::"'..~:";;.""'f:P-

Why Universal

Universal
:t"'L'"-:::T''::':'?:~-

........................,.._ ..... ..
h .. ",~_ ... i ...... _

..... ...... ... ~7;:::::=,, .. .
"""-'_",

__-

=~~~c.;1!"
1I'o!:~~E~
'"*~ijo!:!"'L:(~

~~.;:.:.-.'="!:~~r.:.~ nr..:Iru':!,,,,",,"

Paul. Cockshott

~ University

S!, if Glasgow

-II;:.~-::
Why Computer
....
l-...o... ..__ .

:O;;:.,!11=:-....,."

--.. .•.~-

Pi lot Ace Console

,.~
.~_, J

'

Turinqs idea wath~5 ~Qm~hfng much more

tu nd(! menta l
;;I,11d lastillg than that of

ChUhOh

Turing's
rsal

Universal
U;ol!lr1:f
~I)o

Digi

Cheapness
EUill\r'E!J':KIlmJcl1olJJeRas !! ;adv;aRt;Jg~ t~ cne
!d~r'I ~r'I ue I!Ipplll!d

litilEl'iI. This 1linI.~~ hllQE!
(lm;Im"f'iJf:;«lLIJ iriuf;Jcruril1g.

-

ty
W

Why Universal
Predecessors
TIi~r~ were lots. of computing mamirtl;!=? Ibl?f~Hlli:Turing put to practlca l uee

~~~F-§-

~!.~;-.
Cheapness
The universal machine has the advantage that one design can be applied to any problem. This brings huge

Analogue·

Turing needed' self reference in order to address the decision problem. But this property was the key to the economical adoption of computer

economy of scale to
manufacturing.

technology. It gives Universality. "This special property of diqital computers, that they can mimic any discrete-state machine, is described by saying that they are universal machines."

Predecessors
DigitalL
perfbrms
th:E!~ ratl~ digitallil'lULtlplicatiol\5: .;mr! MI(lftions

There were lots of cornputl ng machines before Turing put to practical use '"

Analogue
HlO vearsaqo navleshad to solve .~al-tim~ vector ariUh'l'leti~ problems lind reqression problems that were w.y beyond what could tJ.e done
malnuailly.
_'

JIIIA ..

IIIIII.

p,ntjk.Yth~ra comnutcr, andenr G~ck tecMIl1c9Y. preelcrs umar ectrpses eno lunar pcsltlcns, perhaps ~i Df APJlolo.nfus

Ranges were aOO!,ilt10 milt;$.
relative velocities up to 60 mp'h

time oJ night of the order 20secs
1~p':@ttm~~Wd£dil.ll. ~lrl .. ",~(mGd.l

;-".,

--",

I

!'_

\

!

/

1111'-1

'-

y~

Digital
performs fixed ratio digital multiplications
and additions

machln

put to ~

Antikythera

computer, ancient Greek perhaps work

technoLogy, predicts Lunar eclipses

lunar positions;

andl of Appolonius

implements Appnlonius ecc e ntri c orbrtat model

~~:I' -kb-

Antikythera computer, ancient Greek technology, predicts lunar eclipses and lunar positions, perhaps work of Appolonius

performs fixed ratio digital multiplications and additions

ruus
bold lines indicate eccentric model
dashed lines the epi-cycle model

-,

-,

,
I

,

-,

J
I I

, ,

I

I

,
\

,

D
,

-,

E
,
j

implements Appolonius eccentric orbital model
,
I f

s

I

,
\
I

I

,

,

I

-,

,

!

L___

____=_

- - __

-

-

'-----

__j

Moon
Moon

KEPLER
APOLLONIUS Earth

+

Earth

nodes of ellipse center of circle

-

-

-<,

B
/' ./

/' / /'

~

....
"<,

"-

C
""
'\

./ / / /

"

\
\ \

I
/

A
I I

I
J

\ \

I \ \
/
/'

E
"
.....

I
j

"

-,

\

/

\ \ \ \

I

0
\

b
\
'

I

I

\

,

C

I'

f
J\
\ / \
\ /

I
\

I

I

r
I
\ \

" "-

./

"

,
\

I

I
\
\

f I \ \
\ / / /

" "-,
'<, ./ ./

I' / ./

-, -, "<,

/'

--

~"

groove in B

in mounted on C

re Turing l U5,e

Analogue
100 years ago navies had to solve real-time vector arithmetic problems and regression problems that were way beyond what coutd be diane

manually.

Ranqes were about 10 miles relative velocities up to 60 mph ti me of fUght of the order 20secs
A

1-::'YOrB
~f~~\'~l~A

='[L:Ulj

Rear Adrrniral Johm Dumaresq c~rnmand~r of' RQ}'al AMstralii.'ln N~vy descrlbed by his colleagues i nav3l gUrI'11B"i)' a matnemsncat as genius. and UI1SUllg pioneer of

fr'Ieeh~t'lie~~ C.:Jm:putlr.1J

100 years ago navies had to soLve reaL-time vector arithmetic probLems and regression probLems that were way beyond what couLd be done manuaLLy.

'" n
Ranges were about 10 miLes reLative veLocities up to 60 mph time of flight of the order 20secs
B A
~~-_:

-.--<~

~-"-----.

Bearing of B-relative _-to A---

Fthe order 20secs
B

A
Bs velocity

Bearing of B-felative __ A" --to

[15,-3

As velocity [0,-14]
...

Velocity of B relative to A = [15,11]

Dumaresq

l Australian
lis colleaques in nathematical pioneer of . ~ng

Rear AdrmiraL John Dumaresq commander of RoyaLAustralian Navy described by his colleagues in navaL gunnery as a mathematicaL genius, and unsung pioneer of mechanical computing

0.(
AFT

AJ 0

81."
HE

II L 0

r
lfu',I
,10 l

Pl.A
'/I£q

e

n:

SeLf reference
Turing needed seLf reference in order to address the decision problem.

But this property was the key to the
economicaL adoption of computer technotoqy. It give's Universality.
'This speciaL property of digitaL computers, that they canl mimic any discrete-state machine, is described by saying that they are universal machines."

Cheapness
The universal machine has the advantage that one design can be applied to any problem. This brings huqe economy of scale to manufacturing.

Turing's
rsal

Universal
U;ol!lr1:f
~I)o

Digi

Cheapness
EUill\r'E!J':KIlmJcl1olJJeRas !! ;adv;aRt;Jg~ t~ cne
!d~r'I ~r'I ue I!Ipplll!d

litilEl'iI. This 1linI.~~ hllQE!
(lm;Im"f'iJf:;«lLIJ iriuf;Jcruril1g.

-

ty
W

Digital
machine is digital however mas a more subtle It means firstly tlhdt numbers can be represented by strlnqs of dirghs that can be as lonrg'as one wi~hes, One can therefore work <0 any d:esired d:egree ofacouracv. This; BCCL.U'aC)! is: not obtained by more careful machimin91 of parts, control of temperature vari at ions; and such means. but by a slf.ght increase in the amount of equipment in the macntne, To double the number of si'grnffiiGlnt figures, would involve increasing the amount of the equipment by a factor definitely less than two, and would also have some enect in increasing the time taken over each job. This is. in sharp contrast with analogue ma;Ctlj'ne-s., .and continuous. varIable machines such as the dUffelienti<lll analyser, where each additionaU declrnal dligit required necessltates a complete redeslqn of the machine, and an increase irn the cost ltIy as much as a focroroflQ_1 Turing, Lecture ron the AUlOmoti,
That the signifr.c:ance_ Computing, Engin@) More gener;]LLythe-rof- is a fundamental limit to spatial

acc~r~c)' provided by the Planck lengl:h M around lOh{-35} .meters whiCh limits the fundamental accuracy of ,any analoque cornputinq device .. proposals far trans: Turing ccrnputlnq are based the illusion that real numbers are 'reali' in the cntoloqical sense.
Most on

Tliley are based on continuum modets ot the world like Maxwells: equations or Newtonian mechanics.

arg~ment ~hNe is very pragmatic, There has a recent ternptanon to think that we {ian outperform d[gitaL (amputation by revertlnq to analcque <;Qmputing,
Turing. been

In the post- Turing era we have to see thecrles llke equations or Newtoniam mechanics as software packag,es rer mak.ing p:red;ction; abcut real'ty, When combined with. computer to do, the maths they allow us to build models (hat m.m.c some part ofreal'ty.
M.axwells 6~tjust like

the Antikyt~era
of
DUIi

model, tnere are Limits paC"kages. But
OIlJ r

to the a [curacy T~iS is ~"ecl

software

software paekag,~~ ate not 'eality rtsell.
in reaUty·!:'vEliythirlg is false, everyt~ing

Reality is digital, Turing computability rules.

on a philQ~Ophi<;;j1 miSQoncepliQn that is continuous, but we know:thls
Is dligitalor quanttsed.

The norton of the coottnuurn arose [n, classics! Greek geometry from the proof <If the irrationality of tile leng;tm of til. h'llpotenu,,, of a rigiht triangle with unit sides, If on.. assumes, as the ·Greeksdid, that classical ge<lmetry was il true theory of the real world, this implied mat space must be continuou,sly
subdivistble.

S Ujp pose that UlS i ng 50 me to ntln uu m model 0 ~ mecnanrcs we show that a system with computable boundary cundltlons has some points where some parameters are uncomputable, does this 'leUIus that the real world can do thinqs which are uncomputable? No .. It tells you that the software packeqe, or phvslcaf theory you are using has a buq in it
lit was jU5t such a 'bug' in, M.>well. equations, the

wo.k,in

But do we know ,f Pyth.gorus theorem ,otu,Uy the real world? (oul.d we experimentally

test it?

ultraviolet catastrophe, quanmm t~~o""

that net Einsteifl to invent the

That the machine is digital however has a more subtle significance.

It means firstly that numbers can be rep-

resented by strings of digits that can be as long as one wishes. One can therefore work to any desired deqree

of accuracy. This accuracy is not obtained by more
carefuL machining of parts, controL of temperature amount of equipment vari ations and such means, but by a slight increase in the in the machine. To doubLe the number of siqnificant fiqures, would involve increasing the amount of the equipment by a 'factor definitely less than two, and wouLd also have some effect in increasing the time taken over each job. This is in sharp contrast with analogue machines, and continuous variable machines such as the differential additional analyser, where each decimal digit required necessitates a compLete Lecture on the Automatic

redesiqn of the machine, and an increase in the cost by as much as a factor of 10. (Turing, Computing
"'F_ .. _!.

Engine)
.__ ... __ ._._ ........ 1__ ! ._._ .. _. ,._. .... __ !_

.__

"'FI __

.. __

1_. __

I I ~ L I III II ::, ::,U L I I d::' LI leu d e

I I I e I e I I Lid l d I Id l Y ::,e I, VVI lie I e ed. L I I

additional decimal diqit required necessitates a complete redesiqn of tlhe machine, and an increase in the cost by as much as a factor of 10. ( Turing, Lecture on the Automatic Cornputinq Engine)

Turings argument tlhere is very pragmatic. There has been a recent temptation to think that we can outperform digital computation by reverting to analogue computing.

This is based on a philosophical misconception that in reality everythinq is continuous, but we know this is faLse, everything is digital or quantised. The notion of the continuum arose in classical Greek geometry from the proof of th,e irrationality of the length of the hypotenuse of a right triangle with unit sides. If one assumes, as the Greeks did, that classical geometry was a true theory of the real world, this implied that space' must be continuously subdivisible.

If one assumes" as the Greeks did, that cllassicaL geometry was a true theory of tlhe reaL world, this impLied that space must be contlnuously subdivisible. IBut do we know if Pythaqorus theorem actually works in the reaL world? CouLd we experimentally test it?
B
.~

~ ~ We can envisage an experimental setup that allows us to test Pythagoras theorem in the real world. Look at Figure 5.41.Suppose that we have previously measured AB and AC by means of interferometry to be the same integra]. number of wavelengths long. Clearly, if it is the case that the optical path FOund the entire system is also an integral number of wavelengths, then we would either have shown that J2 is a rational number or that Pythagoras theorem did not hold in the real world.

a lL sorts of practical and Logical A
problems, • can not simu ltaneousl measure

works In tne test it?

real WOrlO{ LOUlO we

expenrnentauy

B~

We can envisage an experimental setup that allows Us to test Pythagoras theorem in the real world. Look at Figure 5.4. Suppose that we have previously measured AB and AC by means of interferometry to be the same integral number of wavelengths long. Clearly, if it is the case that the optical path round the entire system is also an integral number of wavelengths, then we would either have shown that J2 is a rational number or that Pythagoras theorem did not hold in the real world.

v

u

A

laser

""

"
'~hlf

C
'II StJt vere d mirror _ mi

.------------,

a

detector

all sorts of practical and logical problems, • can not simuLtaneousl measure Be and AC because of mirror o rie ntati 0 n, • can not know you have a right angle without pre assuming pythagoras theorem

More generally there is a fundamental Limit to spatiaiL accuracy provided by the PLanclklength of around 1,OA{-3,S} .rneters which Limits the fundamental accuracy of any analoque cornputinq device. Most proposaLs for trans Turing cornputlnq are based on the illusion that real numbers are 'real' in the ontological sense.
They are based 'onl continuum models of the world Like MlaxweLLsequations or Newtonian mechanics.

In the' post-Turing era we have to, see theories like Maxwells equations or Newtonian mechanics as software packaqes for making predictions about reality, When comlbined with a computer to do the maths they alLow us to build models that mimic some part of reaLity.
But just like the Antikytlhera model, there are limits to the accuracy of our software packaqes, But our

software packaqes are not reality itself.

Suppose that using some continuum model of mechanics w'e slhow that a system with computable' boundary conditions has some points where some parameters are uncomputable, does this tell us that the real world can do things which are uncomputable? No. It tells you that the software' package" ,or physical theory you are using has a buq in it. It was just such a 'bug' in Maxwells equations, the ultraviolet catastrophe, that let Einstein to invent the quantum theory.

stial

sed

e

ts

ReaLity is dig ita L, Turing computabiLity ruLes.

Why Computer
Unive-r!;OlL'C(lmp uter

Equivalent to Lambda Caclulus? No! As at controlled experiment a simple tambda expression was left overnight on tOIP of book with relevant theory. but it failed to evaluate.
But pUI ,;n~o Ihe L,rned. Can it""'r~ed

Pilot Ace Console
Turings idea wa thus something much more
This 'Was s"metM~fl real, not
tJ!

Th. I..mda calculus i. only eqllivulent to tne ".i.e"aI Computer if by the La,"oo.' ",Ieu/". we mean' either • A l.arnbdlll i"mrp.rete.r .0'11' Q UiJiver!iil!J~ Computer . Or a Mathematlclo", a /l{ackb(IQrd and a' <lefioltJM o/'the ,a'co'". that the matllem.rJrlan undilmand., Tu,;n9 brings OU! the import,,",e of phy;ic,,1 embodimen.t for <alculation. Bylntro<lucil1g a Mllchi.', h. introduce> MECHANICS and indi,.clly Physic< .s a support far mathematics...

thrnlght narmitl€ ..

fundamental and lasting
than that of

'Church

10k

t

Pilot Ace Console

This was something real; not a thought machine.

\

~

tlnlversal Computer Equivalent to Lambda Caclulus? Nlo! As a controled experiment
a simple lambda expression was Left overnight on to'p of book with relevant theory, but it failed to evaLuate.
But put into the Lambda Can
it worked
Type a term, Help for info or Quit to exit. > (\x .(x+l)) 2

3
(7 Lei>
uctions, 0.005 CPU)

:lCVdIIL

LIICUI

y,

UUL I L

to eva Luate.
But put into the Lambda Can it worked

E

I'

Type a term, Help for info or Quit to exit.

> {\x .(x+1)) 2 3
(7
I

uctions, 0.005 CPU)

Lei>

Missing secret ingredient miversal computer

E

I'

Type a term, Help for info or Quit to exit.

> {\x .(x+1)) 2 3
(7
I

uctions, 0.005 CPU)

Lei>

Missing secret ingredient niversaL computer

The Lamda calculus is only equivaLent to the UniversaL Computer if by the Lambda calculus we mean either • A Lambda interpreter on a Universal Computer

• Or a Mathematician, a blackboard and a definition of the calculus that the mathematician understands. Turing brings out the importance of physical embodiment for calculation.
By introducing a Machine he introduces MECHANICS and indirectLy Physics as a supportfor

mathematics.

Sign up to vote on this title
UsefulNot useful