Hacking I., The Taming of Chance Cap1

7/30/2019 Hacking I., The Taming of Chance Cap1

http://slidepdf.com/reader/full/hacking-i-the-taming-of-chance-cap1 1/10

1

The argument

The most decisive conceptual event of twentieth century physics has been

the discovery that the world is not deterministic. Causality, long the

bastion of metaphysics, was toppled, or at least tilted: the past does not

determine exactly what happens next. This event was preceded by a m or e

gradual transformation. D uring the nineteenth century it became possible

to see that the world might be regular and yet n ot subject to universal laws

of nature. A space was cleared for chance.

This erosion of determinism made little immediate difference to

anyone. Few were aware of it. Something else was pervasive and every-

body came to kno w about it: the enumeration of people and their habits.

Society became statistical. A new type of law came in to being, analogous

to the laws of nature, but pertaining to people. These new laws were

expressed in terms of probability. They carried with them the conno-

tations of normalcy and of deviations fr om the no rm . T he cardinal conceptof the psychology of the Enlightenment had been, simply, hum an nature.

By the end of th e nineteenth cen tury, it was being replaced by som ething

different: normal people.

I argue that these two transformations are connected. Most of the

events to be described to ok place in the social arena, n ot t ha t of the natural

sciences, but the consequences were momentous for both.

Th rou ghou t the Age of Reason, chance had been called the superstition

of the vulgar. Chance, superstition, vulgarity, unreason were of on e piece.The rational man, averting his eyes from such things, could cover chaos

with a veil of inexorable laws. The world, it was said, might often look

haphazard, bu t only because we d o not kn ow the inevitable workings of its

inner springs. As for probabilities - whose mathematics was called the

doctrine of chances - hey w ere merely the defective bu t necessary tools of

people w ho kn ow too little.

There were plenty of sceptics abo ut determinism in thos e days: those

who needed room for freedom of the will, or those who insisted on theindividual character of organic and living processes. N o n e of these though t

for a mom ent that laws of chance would provide an alternative to strictly

causal laws. Yet by 1900 that was a real possibility, urged as fact by an



2 The taming of chance

adventurous few. The s tage was se t for u l t imate indeterminism. H o w did

that happen ?

This is not a quest ion about some sort of decay in knowledge or

management. The erosion of determinism is not the creat ion of disorder

and ignorance - uite the contrary. In 1889Francis G al ton , founder of the

biometr ic school of statistical research, not to mention eugenics , wrote

that the chief law of probabil i ty ‘ reigns with serenity and in complete

effacement amidst the wildest confusion’.’ By the end of the century

chance had attained th e respectability of a Victorian valet, ready to be the

loyal servant of the natura l, biological a nd social sciences.

There i s a seeming paradox: the mo re the indeterminism, the m ore the

control . Th is is obvious in the physical sciences. Q u an tu m physics takes

for granted tha t natu re is at botto m irreducibly stochastic. Precisely tha t

discovery has immeasurably enhanced our abi l i ty to interfere w ith and

alter the course of nature. A moment’s reflection shows that a similar

s tatement may be at tempted in conn ection with people. Th e parallel w as

noticed quite ear ly. Wilhelm Wundt, one of the founding fathers of

quantitative psycho logy, w rote as ear ly as 1862: ‘It is statistics that first

dem on strate d tha t love follow s psycho logical laws.’’

Such social and personal law s w ere to be a matter of probabilit ies, of

chances. Statistical in nature, ‘these law s were none theless ine xorab le; th eycould even be self-regulating. People are normal if they conform to the

central tendenc y of such laws, while those at t he extrem es are pathological.

F ew of us fancy being pathological, so ‘most of us’ t ry to mak e ourselves

normal, which in turn affects what is normal . Atoms have no such

inclinations. T h e hum an sciences display a feedback effect n o t to be foun d

in physics.

Th e t ransformat ions tha t I shall describe are closely conn ected w ith an

event so all-embracing that w e seldom pause to notice it : an avalanche ofprinted num bers. Th e nation-states classified, co un ted an d tabulated their

subjects anew. Enu merat ions in som e form have been w ith us always, if

only for the two chief purposes of government, namely taxat ion and

military recruitment. Before the N apole onic era m ost official coun ting had

been kep t privy to adm inistrators. After it, a vast am ou nt was printed and

published.

The enthusiasm for numerical data is reflected by the United States

census. Th e first Am erican census asked fo ur questions of each household.

The tenth decennial census posed 13,010 quest ions o n var ious schedules

addressed to people, firms, farms, hospitals, churches and so for th. This

3,000-fold increase is striking, b ut vastly und ersta tes the rate of growth of

printed n um bers: 300,000 would b e a better estimate.

The pr in t ing of numbers was a surface effect. Behind it lay new



The argument 3

technologies fo r classifying and enu m erating , an d ne w bureaucracies w ith

the au tho r i ty and cont inui ty to deploy the technology . There is a sense in

which many of the facts presented by the bureaucracies did no t even exist

ahead of t ime. Categories had to be invented into which people could

conveniently fall in ord er to be cou nted. T he system atic col lect ion of data

about people has affected not only the ways in which we conceive of a

society, but also the ways in which we descr ibe our neighbour . It has

profoundly t ransformed what we choose to do, w h o w e t r y to be, and

wha t we th ink of ourselves. M arx read the m inutiae of official statisti cs, th e

repor ts f rom the fac tory inspectora te and the like. O n e can ask: w ho had

more effect on class consciousness, Marx or the au thors of the official

reports which created the classifications into which people came to

recognize themselves? These are exam ples of quest ions about wha t I cal l

‘ mak in g u p pe op le’. T h i s b o o k t o u ch e s o n t h em o n l y i n d i r e ~ t l y . ~

W ha t has the avalanche of pr in ted numbers to do with m y chief topic,

the erosion of determinism? O n e answer i s imm edia te. D eterminism was

subver ted b y laws of chance. To believe there were suc h laws o ne needed

law-like statistical regularities in large popu latio ns. H o w else co uld a

civilization ho ok ed o n universal causality g et the idea of som e al ternat ive

kind of law of nature or social behaviour? Games of chance furnished

initial illustrations of chance processes, as did birth an d m ortality da ta.Those became an object of mathematical scrutiny in the seventeenth

century . Without them we would not have anyth ing much l ike our

m ode rn idea of probabil i ty . But it is easy for the determinis t to assume that

th e fall of a d ie or the sp in of a roulet te w ork ou t according to the s imple

and immutable laws of mechanics. N ew ton ian science had n o need of

probabilities, except as a tool for locating underlying causes. Statistical

Iaws that look l ike brute, irreducible facts were first found in human

affairs, but they could be noticed only after social phenomena had beenenum erated, tabulated an d m ade public. Th at role was well served by th e

avalanche of printed num bers at the s tar t of the nineteenth century.

O n c loser inspect ion w e f ind tha t not any n um bers served the purpose .

M o s t of the law-like regularities were first perceived in connection with

deviancy: suicide, crime, vagrancy, madness, prostitution, disease. This

fact is instruct ive. I t is now co m m on to speak of in fo rmat ion and con t ro l

as a neutral term embracing decision theory, operations research, r isk

analysis and the broader but less well specified domains of statistical

inference. W e shall f ind that t he ro ots of the idea lie in the notio n that o ne

can improve - contro l - a deviant subpopula t ion by enumerat ion and

classification.

W e also f ind that routinely gather ing numerical data was not eno ugh to

m ak e statistical la w s rise to the surface. T h e laws had in the beginning to be




read in to the data. Th ey w ere not s im ply read off them. Th rougho u t th is

book I make a contras t of a rough an d ready sor t between Prussian (and

other east European ) at t itudes to num erical data, and those that f lourished

in Britain, France, and other nations of we stern E uro pe . Statistical laws

w ere fou nd in social data in the W est, wh ere libertarian, individualistic an d

atomistic con ceptions of the person and the s tate were ram pant . This didno t happen in the East, wh ere collectivist an d holistic atti tudes were m ore

prevalent . Th us the transforma tions that I describe are to be unders tood

only w ithin a larger contex t of w hat a n individual is, and of w hat a society

IS.

I shall say very li t t le about mathematical conceptions of probabil i ty .

T he even ts to be described are, nevertheless, ingredients for u ndersta nding

probability and for grasping why it has been such an incredible success

stor y. Success stor y? A quad ruple success: metaphysical, epistemological,

logical an d ethical.

Metaphysics is the science of the ultimate states of the universe. There,

the probabilit ies of qu an tum me chanics have displaced universal C artesian

causation.

Epistemology is the theory of know ledge and belief. N ow aday s w e use

evidence, an alyse data, design expe rime nts and assess credibility in term s

of probabilit ies.Logic is the th eory of in ference and argument . Fo r th is purpose we use

the deductive and often tautological unravelling of axioms provided by

pure mathematics , but also, and fo r m ost pract ical affairs , we n ow em ploy

- sometimes precisely, sometimes informally - the logic of statistical

inference.

Ethics is in part the study of w h a t to do. Probabil i ty cannot dictate

values, bu t it now lies at th e basis of all reasonab le choice m ad e b y officials.

No pub lic decision, no risk analysis, n o enviro nm ental impa ct, n o miIitarystrategy can be conducted without decis ion theory couched in terms of

probabilit ies. By covering opinion w ith a veneer of objectivity, w e replace

judgement by computat ion.

Prob ability is, then, the philosophical success stor y of th e first half of

the twentieth century. To speak of philosophical success will seem the

exaggeration of a scholar . T ur n then to the most worldly affairs . Prob-

ability and statistics c row d in up on us. T he statistics of o u r pleasures and

o u r vices are relentlessly tabulate d. S po rts, sex, drin k, dru gs, travel, sleep,

friends - nothing escapes. There are more explicit s tatements of prob-

abilities presented o n A me rican prim e time teievision tha n explicit acts of

violence (I’m c ou ntin g the ads). O u r pub lic fears are endlessiy deba ted in

terms of probabilit ies: chan ces of meltdowns, cancers, muggings, earth-

quakes , nuclear winters, AID S, global greenhouses , what next? The re is



The argument 5

nothing to fear (it may seem) but the probabilit ies themselves. This

obsession with the chances of danger , and with treatments for changing

the odds, descends direct ly f rom the forgotten annals of nineteenth

century information and control .

This imperialism of probabilit ies could occur only as the world itself

became num erical. W e have gained a fun dam entally quantitative feel for

nature , ho w it is and how i t ought to be. Th is has happened in par t for

banal reasons. We have trained people to use numerals . The abil i ty to

process even qu ite small nu m be rs was, until recently, the prerogative of a

few. Toda y w e hold numeracy to be at least as im po rtan t as li teracy.

But even compared w ith the num erate of old there have been remark-

able changes. G alileo tau gh t tha t God w rote th e w orld in th e language of

mathematics . To learn to read this language w e w oul d have to measure as

well as calculate. Yet measurement was long mostly confined to the

classical sciences of astronomy, geometry, optics , music, plus the new

mechanics. T.S. K uh n has iconoclastically claimed that m easure me nt did

not play much of a role in the ‘Baconian’ sciences that came to be called

chemistry and physic^.^ H e u rg ed tha t measurement found i ts place in

physics - the s tudy of light, sound, heat, electricity, energy, matter -dur ing the n ineteenth century . Only around I840 did the pract ice of

measurement become ful ly es tablished. In d ue cou rse measuring becamethe o nly experimental thing to do.

Me asurement an d positivism are close kin. Aug uste C o m te coined the

w ord ‘positivism’ as the nam e of his philosophy, holding that in all the

European languages the w ord ‘posi tive’ had goo d connotat ions . H is ow n

philoso phy d id no t fare especially well, b ut the w ord caug ht on . Positive

science m eant num erical science. N ot hi ng better typified a positive science

than a statistical one - an irony, for Comte himself despised merely

statistical inquiries.The avalanche of numbers , the erosion of determinism, and the

invention of norma lcy are em bedded in the grand er topics of the Industrial

Revolution. The acquisition of numbers by the populace, and the pro-

fessional lust for precision in mea surem ent, w ere driven by familiar them es

of m anufa cture, m ining, trade, health, railways, war, em pire. Similarly the

idea of a norm became codified in these domains. Just as the railways

demanded t imekeeping and the mass-produced pocket watch, they also

mandated s tandards , not only of obv ious things such as the gauge of thelines bu t also of the height of the buffers of successive cars in a train. It is a

mere decision, in this book, to focus on the m ore nar row aspects tha t I

have m ention ed, a decision that is wilful b ut no t a rbi trary. M y project is

philosophical: to grasp the condit ions that made possible our present

organizat ion of concep t s in tw o domains. O n e is that of physical indeter-




minism; the o ther is that of statistical inform ation developed f or purpo ses

of social control.

This study can be used to illustrate a number of more general philo-

sophical themes. I have mentioned one above: the idea of making up

people. I claim that enu meration requires categorization, and th at defining

new classes of people for the purposes of statistics has consequences for

the ways in which w e conceive of others and think of o u r o w n possibilit ies

and potentialities.

A no ther philosophical them e is reasoning. In thinking ab ou t science we

have become familiar with a number of analytic concepts such as T.S.

Kuhn’s paradigms, Imre Lakatos’s research programmes and Gerald

Holton’s themata. Following A.C. Crombie I have thought it useful to

employ the idea of a style of reasoning.’ C ro m bie had in mind e nd ur ing

ways of thinking such as (a) the simple postulation and deduction in the

mathematical sciences, (b) experimental exploration, (c) hypothetical

construction of models by analogy, (d) or de rin g of variety by comp arison

and taxonomy , (e) statistical analysis of regularities of populations, and (f)

historical derivation of genetic development.6

Each of these styles has its o w n sources and its o w n pace. Th ose w ho

envisage continuity in the gro w th of knowledg e see each style evolving at

its own rate. Catastrophists see sharp beginnings and radical mutations.O n e need n ot dogmatically adhere to either extreme in order to see styles

of reasoning coming together. Each contribu ted to wh at Cr om bie calls

‘the growth of a research mentality in European society’.

M y topic is Crombie’s style (e) wh ich, of the six that h e distinguishes, is

qu ite the most recent. Despite various discernible precursors and anticipa-

tions, o u r idea of probability came into being only a roun d 1660, and the

great spurt of statistical thinking did not occur until the nineteenth

century. T he statistical exampIe makes plain that t he g row th of a style ofreasoning is a matter n ot o nly of thought but of action. Take so seemingly

unproblematic a topic as population. We have become used to a picture:

the number of people in a city or in a nation is determinate, like the

num ber of people in a room at noon , and no t like the num ber of people in a

riot, o r the nu m ber of suicides in the w orld last year. But even the very

notion of an exact popu lation is one which has little sense until there are

institutions for establishing and defining what ‘population’ means.

Equally there must be ways of reasoning in order to pass from cumber-som e data to sentences with a clear sense abou t ho w m any were such and

such. Most professionals now believe that representative sampling gives

more accurate information about a population than an exhaustive census.

This was unthinkable durin g most of the nineteenth century.’ T h e very

thought of being representative has had to come into being. This has



The argument 7

required techniques of thinking together w ith technologies of da ta collec-

t ion. An entire style of scientific reaso ning has h ad to evolve.

I ts development was int imately connected w ith larger quest ions abo ut

wh at a society is, and th us leads to speculation and historical stud y of the

formation of the western conc ept of acommuni ty . ' Bu t it also invites m or e

abstract analytical philosophy, because styles of reasoning are curiously

self-authenticating. A proposition can be assessed as true-or-false only

when there is some style of reasoning and investigation t ha t helps

determine its t ruth value. W hat the proposi t ion m eans depends upo n th e

ways in which w e m ight set tle its t ruth. T hat innocent ob servation verges

nervously o n circularity. We c ann ot justify the style as the w ay best to

discover the tru th of the propo sition, because the sense of the proposi t ion

itself depend s up on the style of reasoning by w hich its t ruth is settled. A

style of thinking, it seems, can no t be straightforwardIy w rong, on ce it has

achieved a status by w hich it fixes th e sense of what it investigates. Such

thoug hts call in que stion th e idea of an independ ent world-given cr i ter ion

of t ruth. So the seemingly innocent notion of a style of reason ing can lead

to deep w aters, and it is wiser to enter them by w ading into examples than

by a high dive into abstract ion. The development of statistical thinking

may be o u r best exam ple available - ecause m ost recent and end uring and

now pervasive.Historians will see at once that what follows is no t h i s to ry . O n e may

pursue past knowledge for purposes other than his tory of science or

his tory of ideas. A noncomm it ta l account of what I am at tempt ing might

be: an epistemological study of the social and behavioural sciences, with

consequences for the conce pt of causa lity in th e na tura l sciences. I prefer a

less expected description. This book is a piece of philosophical analysis.

Philoso phica l analysis is the investig ation of concepts . Con cepts are w ords

in the ir sites. T he ir sites are sentences an d institution s. I regret that I havesaid too little a b o u t institutions, and too much abo ut sentences and ho w

they a re arranged.

But w hat sentences? I use only the printed w ord , a minuscule fraction

of w ha t was said. Th e distinguished statistician I. J. G o od noted in a review

that ' the t rue his tory of probabil i ty or of science in general will never be

written because so much depends o n unrecorded ora l communicat ion , and

also because w riters ofte n do no t cite their source^'.^ T he t rue h istorian of

science is well able to solve the second problem , bu t no t the f irs t . O n e may

nevertheless make a good s tab at it by consult ing the ample Victor ian

troves of notebooks, letters and o ther ephemera. I do n o t do so, fo r I am

conc erned with the public life of concepts and the w ays in w hich they gain

autho rity. M y data are published sentences.

But which ones? I om it many per t inent words because o ne cannot d o



8 The taming af chance

everything. I leave ou t Malthus and M endel, for example, A.A. Co ur no t,

Gustav Fechner, Florence Nightingale and ever so many more modest

participants in the taming of chance. Very well: but I say nothing of

Maxwell, Bolzmann or G ib bs , altho ug h statistical m echanics is critical to

the spread of chance and probabili ty not only in to physics bu t also into

metaphysics. I say nothing of Charles Darwin, al though evolut ionary

theorizing was to imp ort chance into biology. I say nothin g of Karl Marx

fabricating an iron necessity o u t of the very sa me numerals, the identical

official statistics, t ha t I have incorporated in to an account of the taming of

chance.

Th ere is an uncontroversial go od reason for silence ab ou t these figures.

Scholars and teams of scholars dedicate their lives to the study of one or

another. It wou ld be folly to venture a short s tory here, a mere chapter.

But i t is not only prud ence and respect, but also m etho d, that makes me

hold m y tongue. Transform ations in concep ts and in styles of reasoning

are the produc t of countless trickles rather than the intervention of single

individuals. Marx, Da rw in a nd M axwell wo rked in a space in which there

was something to find out. Th at m eans: in w hich various possibil i ties for

truth-or-falsehood could already be formulated. This b oo k is abo ut that

space. So although a lot of sentences are reproduced in this bo ok, they are

the words not of heroes, but of the m ildly distinguished in their d ay, thestuff of the mo re impersonal parts of o u r lives.

Sentences have tw o powers. T hey arc eternal, and they are uttered at a

moment . They are anonymous, and yet they are spoken by flesh and

blood. I have tried to answer to these tw o facts. O n the one hand, I d o

regard the sentences as mere material objects, inscriptions. Bu t to d o tha t,

and o nly that, is to becom e lost in vain abstraction. A s counterbalance, m y

epigraphs to each chap ter are dated , to recall that o n a real day im porta nt

to the speaker, those very wo rds w ere uttered, o r are said to have beenuttered. M y footnotes (m arked with asterisks) are anecdotes that wou ld be

imp rope r in the m ore solemn text.:" Th ey give so m e tiny g limpse of w h o

the speakers were. But there is seldom anything personal about the

footn otes. Th ey address th e individual as official, as public w riter, even if

his behaviour may strike us, so m uc h later, as strange.

Th us although m any chapters have a central character or text, it is not

because Salomo n Ne um ann , A.-M. Gu err y or John Finlaison is ' impor-

tant'. Th ey are convenient and exem plary anchors for a particular organi-zation of sentences. I use the antistatistical me thod , th at of F r i d i r i c Le

Play, topic of chapter 16. After having interminably trekked across the

':. Notes at the end of the boo k provide references, and, rarely, numerical formulae. Th ey aremarked w ith numerals. A num eral after an asterisk (as :i3) indicates that no te 3 at the end of

the book bears o n the material in th e foo tnote marked ::..



The argum ent 9

written equivaIent of his Hartz mountains, I take what I think is the best

example of one speaker. Much like Le Play, I include a few stories, b ut the

personages whom I use are in some ways like his household budgets, if ,

alas, less tho rou gh .

Th ere is one exception among these chapters. Th e final on e is twice as

long as the others, and is a rather full account of one side of one writer,namely C.S. Peirce. H e really did believe in a universe of absolute

irreducible chance. H is w ord s fittingly end this boo k, for as he wro te, that

thought had become possible. But I argue that it became possible because

Peirce now lived a life that was permeated with probability and statistics,

so that his conception of chance was oddly inevitable. H e had reached th e

twentieth century. I use Peirce as a philosophical witness in some thing like

the way that I used Leibniz in The Emergence of Probability.'O But

Leibniz was a witness to the transformation that I was there describing,

namely the emergence of probability around 1660 and just afterwards.

H ere Peirce is the witness to something tha t had already happened by the

time that he was mature. That is why he is the topic of the last chapter,

whereas in Emergence the name of Leibniz recurred throughout.

Although other philosophers are mentioned in the two books, only

Leibniz and Peirce play a significant part. T he tw o w orks do, however,

differ in structure in other ways. Emergence is about a radical mutationthat took place very quickly. Doubtless, as Sandy Zabell and Daniel

Garber have shown in an exemplary way, the book underestimated

various kinds of precursors." M y central claim was, however, tha t many

of o u r philosophical conceptions of probability were formed by the nature

of the transition from immediately preceding Renaissance conceptions.

Accounts of the m eth od olo gy have been given elsewhere." Taming, in

contrast is about a gradual change. Hence the geological metaphors:

avalanches, yes, but also erosion.Most of my selections and omissions - such as my long treatment of

Peirce and my neglect of any oth er phil oso ph er- have been deliberate. But

sloth and good fortune have also played their part. When I began work

there was hardly any recent secondary material; now there is a great deal. I

am particularly glad of new books by m y friends Lorraine D aston, Ted

Porter and Stephen Stigler, and of earlier ones by W illiam Co lem an and

Donald MacKenzie. We all participated in a collective inspired and guided

by Lo renz Kriiger. Th e joint w or k of that group has also appeared. H encethere is now a number of brilliant and often definitive accounts of many

matters that overlap with mine.13 Th ey have made it unnecessary for me to

examine a good m any matters. A nd aside fr om specific histories, there are

also points of great generality that I have allowed myself to gloss ove r in

the light of that collective work. For example, another virtue of my



10 The taming af chance

geological metaphor is that the erosion of determinism took place at

marked ly d ifferent rates on different terrains. Not uncommonly the least

deterministic of disciplines most fiercely resisted indeterminism -economics is typical. This phenomenon emerges from the individual

studies of the research group, and is further emphasized in a recent

summing up of some of its r e ~ u 1 t s . l ~

I have mentioned a num ber of m ore specific topics o n w hich I have on ly

touched, o r have entirely avoided: making up people; styles of reasoning;

great scientists; philosophers; mathematical probability. There is a more

glaring om ission. I write of the taming of chance, that is, of the way in

which apparently chance o r irregular events have been bro ug ht un der the

control of natural o r social law. Th e wo rld became no t m ore chancy, bu t

far less so. Ch ance , wh ich was once the superstition of the vulgar, became

the centrepiece of natural and social science, or so genteel and rational

people are led to believe. But h ow can chance ever be tam ed? Parallel to the

taming of chance of w hich I speak, there arose a self-conscious concep tion

of pu re irregularity, of some thing wilder than th e kinds of chance th at had

been excluded by the Age of R eason. I t harked back, in part, to something

ancient o r vestigial. It also looked i nto the future, to new , and often darker,

visions of the person than any that I discuss below. Its m os t passionate

spokesman was Nietzsche. Its most subtle and many-layered expressionwas MallarmC’s poem, ‘Un Coup de dCs’.15 That graphic work, whose

words are more displayed than printed, began by stating that we ‘N E V E R ..

will annul chance’. The images are of shipwreck, of a pilot whose exact

mathem atical navigation com es to naught. B ut th e final page is a picture of

the heavens, w ith the w ord ‘constellation’ at its centre. T h e last wo rds are,

‘Une pensee Cmet un co up de dCs’, wo rds that speak of the poem itself and

which, although they d o not imagine taming chance, try to transcend it.

Documents

Hacking I., The Taming of Chance Cap1