Customer Reviews

The Taming of Chance (Ideas in Context)

5 star:		(3)
4 star:		(1)
3 star:		(1)
2 star:		(1)
1 star:		(1)

 

 

This is a fascinating book, which charts the gradual development of statistical ideas in the nineteenth century, along with associated concepts, such as normalcy, chance, and determinism.

However, a few criticisms are in order. Hacking reports that there was a certain conceptual incoherence surrounding ideas relating to statistics in the 19th century, especially concerning ideas relating to determinism and chance. But I'm not quite sure that Hacking has been able to find the thread out of this confusion, as some confusions appear to remain rather resistant in spite of the narrative, which in general is admiringly clear. Three points will serve as examples:

Eastern and Western: Hacking describes two broad classes of reaction to the development of statistical reasoning; 'Western' (U.K. and France) and 'Eastern' (centred on what was then Prussia) approaches. Western thought, which was largely open to statistical reasoning, is described by Hacking as 'atomistic, individualistic, and liberal'. Eastern thought on the other hand was 'holistic, collectivist, and conservative', and critical of the developing trends of statistics.

Geographical and political issues aside, this characterization almost at once falls apart. For instance, slightly later in the book we are told that statistical methods were resisted in (French) medicine, as medicine was concerned with the individual case, not the average or normal, and hence statistical data was of no use. Immediately after reporting this, Hacking queries, without irony, 'how then could there be a use of statistics in human affairs? In the very institution designed to strip away the individuality of man, namely the court of law'. To add to the confusion, we later find out that Engel, the Prussian apparatchik, and hence 'Eastern' thinker, considered statistical reasoning to be part of a certain mentality he wished to avoid, that of determinism, which denies individual freedom. Likewise, the economist Wagner, Hacking reports, also adhered to this view. In fact there appears to have been a general resistance to statistical methods in the 'East' precisely because the so-called individualistic methods of the statisticians were seen as a threat to the concept of human freedom and individuality.

Durkheim, the French sociologist, whom we are at one point told was 'immersed' in the Western mentality, nevertheless ascribed the functioning of statistical laws to 'collective tendencies', in fact to 'social forces', rather than to the 'underlying little independent causes' of Quetelet, the French pioneer of statistical methodology.

No doubt there was some sort of difference at play here between East and West, but it strikes me that trying to distinguish these two cultures by calling one individualistic and the other collectivist does little to help.

The Title of the book: 'The Taming of Chance', especially if one recalls the title of Hacking's earlier book, 'The Emergence of Probability' which dealt with the preceding era, leads one to think that there are two parts to the development of the ideas mentioned in the book - initially, the emergence of ideas relating to chance and probability, and later their gradual 'taming'. But that would be a mistake. Hacking makes it quite clear that probabilistic and statistical laws were not initially seen to be in conflict with necessity or determinism. Hume, and other enlightenment thinkers, regarded chance as unreal, as merely an illusion caused by lack of knowledge. There was simply no chance around to be tamed, before the nineteenth century. It would appear instead that chance and its 'taming' emerged at the same time - the book then might have been more aptly titled 'The Emergence of Chance'. The idea of 'taming' seems to have slipped in from one of the book's sub-themes, the idea that statistical methods led to greater institutional control of human affairs, or from a certain conception of causality that I shall mention below.

Multiple causality, or causal sets: Quetelet, the French astronomer turned statistician, proposed a theory of 'little independent causes' to explain statistical regularity. The causes of individual cases of, say, suicide, or coin tosses, work independently of each other, but taken as a group, over all cases, they total up in a way predictable by probabilities. As far as Hacking is concerned this explanation 'does not hang together'. Be that as it may, it strikes me that there is an important aspect of Quetelet's purported explanation that deserves attention, and that is the idea that causes are best understood as existing in sets or groups. This idea is reinforced by similar attempts to explain the workings of statistical regularity later mentioned in the book - the holism of Boutroux and Durkheim, as well as the ideas of Peirce and Nietzsche. The latter two, for instance, tried to accommodate probability within causality by claiming that while causality itself may act in a determinate manner, the existence of specific causal laws themselves are a matter of chance. This explanation is not meaningful, it seems to me, unless one brings a prior notion of possibility to play in the existence of particular causal laws - not simply their actuality - as is done with the contemporary notion of possible worlds. To say that laws p, q, and r are possible in certain situations, but only p is actual in this case, is to use the idea of sets or classes of laws which are compossible with certain situations.

There is certainly an ambiguity in the concept of chance; there are at least two ideas involved: chance as opposed to causality, as pure chaos; and chance in consort with causality, 'tamed' chance, so-to-speak, chance that can '[bring] order out of chaos', chance that can support 'laws of chance' (quoted from the book's chapter summaries). Perhaps the idea of causal sets can bring some clarity to a familiar but nevertheless obscure concept, and to help to distinguish between different kinds of indeterminism that are often conflated.

This is an extended essay by the distinguished philosopher Ian Hacking on the theme of probability versus determinism. Something of a hybrid, this is not strictly a philosophical work but a historical commentary on how ideas of chance and probability developed in the 19th century. As Hacking points out at the beginning of the book, a deterministic view of the world was demolished by the emergence of quantum mechanics in the 20th century. In The Taming of Chance, Hacking covers the development of statistics as a discipline and ideas of probability to demonstrate the gradual undermining of the notion of determinism in the course of the 19th century. Hacking opens with Enlightenment ideas of causation, very much under the influence of Newtonian mechanics, and concludes with the thought of CS Peirce, whom he sees as exemplifying acceptance of essentially stochastic views of causation.

Hacking presents this change in world view as driven by a number of intersecting, complex, and unexpected phenomena. A major one was the expansion of the state and systematization of government, particularly associated with Napoleonic France. This leads to the generation of large demographic and social datasets that often reveal unexpected regularities, such as the persistent uneven male to female birth ratio. These datasets not only require new methods of analysis but led to the idea of the idea of statistical 'laws.' Hacking emphasizes that the collection and analysis of this kind of data was driven in part, and in turn fed, by a desire to achieve a higher level of social control. The emergence of social statistics combines with a number of other trends, such as the idea of organ pathology in medicine, the increasingly physiological orientation in biology, and a general emphasis on quanitification in the sciences, to generate a set of new attitudes towards data and ideas of causation. In several aspects, Hacking presents a series of apparently paradoxical or perhaps ironic events, such the desire for improved social control contributing to recognition of stochastic causation, which ultimately transform ideas of causation.

In many respects, this is a somewhat exploratory essay and Hacking's narrative is not laid out smoothly. This book presupposes some prior knowledge of 19th century science and philosophy. It is also dense in the sense that Hacking has compressed a great deal of analysis into a relatively short book. Nonetheless, its worth taking the effort to read it carefully because of Hacking's insightful analysis and knowledge of a broad range of 19th century intellectual history. His reconstruction of how we got from the Enlightenment to Peirce is really impressive.

This book is notable also for Hacking's interesting comments on a number of other features. He has an interesting discussion, for example, of the development of the concept of normality and its consequences. His brief comment about the relationship between Peirce's pragmatism and 19th century idealism is really illuminating.

This may be the greatest work of non-fiction I have ever read. It explains so much about Western society in such an understated way that it is shocking. It remains a must read!

Some works of non-fiction manage to be engaging throughout. Others, like the Taming of Chance, are important but can be tough to read through much of the text. Hacking takes on the history of probability; which he describes as "the philosophical success story of the first half of the 20th century." The taming of chance refers to the way apparently irregular events have been brought under the control of natural or social law.

Hacking takes us through the 19th century intellectual battle between adherents of determinism and probability's champions. The book devolves at times into more of a history of thought than a discussion of the implications of these changes in thinking. In fact, the author admits late in the book, "My chapters have become successively more removed from daily affairs."

He describes chance first as a concept that had no place in reasonable discourse during the Enlightenment. With the development of measures of probability around 1830, chance is condemned by "statistical fatalism" to irrelevance. Finally, with the development of quantum mechanics in 1930, chance becomes the critical element of life with which we are all too familiar.

Along the way, we learn that some proponents of probability helped create the idea that free will existed only in theory (from 1830 until 1930). Thus, criminals are behaving predictably and the degree of their personal responsibility is at issue. Hacking concludes, "we have not made our peace with statistical laws about people. They jostle far too roughly with our ideas about personal responsibility."

While I would not consider this book as light summer reading, it will reveal to the determined reader changes in historical thought with which he is not likely to be previously familiar.

If you look at the table of contents in this book, you'll see that it has 23 chapters. Nearly all of them are less than 10 pages long, with uninformative headings like "Bureaux", "Regimental chests" and "The mineralogical conception of society". This is in fact a good indicator of how fragmented and confusing this book is.

You wouldn't know it from the strange headings, but each chapter deals with some particular field of inquiry where statistical methods and statistical thinking became prominent in the 19th century. That sounds good, but unfortunately the author just dumps a truckload of names on the unsuspecting reader in every chapter. Person A studied this and that, worked here and there and was in regular contact with persons B and C. They in turn studied this and that and thought so and so about chance and statistics. It goes on like this for the entire book, detail after detail after detail. But these collected details aren't very interesting without a broader perspective. A good account of the history of ideas should include some generalizations too.

The author actually states on page 1 that his intention is to argue that two 19th century transformations are connected. The first transformation is the idea of a world which is regular without being causally determined. The second transformation is the emergence of a society governed by statistical information. So apparently his intention was to make a general argument, but he really fails miserably in putting it together. I just finished the book and I don't recall a single passage where the author would have clearly argued the connection he claims on page 1. Whereas good books in the history of ideas contain a general analysis supported by selected examples, this one seems to contain only examples. There's no general argument so the reader is forced to draw his own conclusions.

Maybe this book will have some appeal to people with a biographical interest in history who want to read primarily about influential persons and their deeds. But I like books that concentrate on broader developments which took place above and beyond any particular individual. Such books require some theorizing, and I was disappointed to find that this book did not offer anything on that front. So if you're looking to understand the growth of statistics as a general phenomenon, you'll have too look somewhere else.

but a mind-opener for those who are ready, an awesomely rewarding book for those who are willing to make the extra effort

If, somewhere, deep within the tortured bowels of this book, there is a central thesis that could be stated in a few short sentences and comprehended by most educated English-speaking peoples, I have yet to find it. Endless restatement, obfuscated in painfully cultivated strings of verbiage, of trivial fact is used to document an hypothesis that if stated clearly could be supported or refuted in about a page-and-a-half and then likely consigned to the graveyard of such endeavor. The prose is a true caricature of Derrida's; the logic is a laTour de force. Typical of such works, the author begins with a premise and then selectivley seeks textual support. Of course, such an approach can be conveniently utilized to support any premise and if written with sufficient opacity will pass for scholarship and great insight.

The book is an unreadable bloody bore; its value is restricted to its caloric content relative to the market price of a barrel of Texas sweet.