a very well written introduction to a few important ideas from physics that have been used to model and explain facts in the social sciences. I liked the author's pedagogical insistence on certain themes. for example, I had never understood that phase transitions can be seen as the signature of processes where a lot of items almost simultaneously interact with their neighbors and so transitively with one another. similarly for power law distributions. the second part of the book (the internet and its patterns of connectivity) veers into territory that is very extensively covered by a large number of very good recent books and so I found it somehow less interesting/informative. ( )

This is a super book.

Philip Ball, a self-confessed liberal - more on that later - is first and foremost a scientist (a former staffer on Nature magazine), and his brief here is to canvas the application of statistical scientific explanations of physical phenomena, such as phase transitions in liquids and solids, by analogy to human behaviour.

This is a splendid enterprise, not just because it is a very imaginative application of established knowledge to novel fields of enquiry, which to my mind is always a worthwhile endeavour (whether or not the results are useful, we're better off if someone has done the intellectual exercise than if they haven't), but also because it grasps a fundamental point which social scientists almost always miss: It is what a population will *actually* do which matters, not what it *ought* to do, or what *we'd like it* to do.

Politics is the pursuit these latter questions, and it is almost always pursued in ignorance of scientific data describing the former, and a central point of Ball's book is that this is a dreadful shame. And so it is.

A complaint one sees levied against Ball's book is that it misses the critical distinction between physical particles - which are all identical, except for a few key easily measured properties - and humans who, in almost every respect, are entirely different from each other. But this misses the point: the very beauty of statistical physics is that you can draw inferences about how a large mass of particles behave without knowing or measuring *anything* about the behaviour (vector, spin, magnetism, whatever) of any given particle. And so it is with people: Ball's argument is to say, on the basis of the statistical evidence, from the markets, from patterns on university lawns, from trajectories of individuals navigating a corridor: we can make inferences about what a group of people will do knowing nothing about their individual motives, in the same way we can about particles without knowing their vector or spin. At that level, people are *not* significantly different: people *do* behave like particles. Therefore these fundamental differences between people, which may be real (but may be not - for all we know, these "special qualities" we cherish may be a product of human chauvinism) are not material to how we behave en masse.

As Ball moves on, his subject resolves slowly to focus on social interaction within a society, and the interesting work on game theory and iterated prisoner's dilemma by Axelrod and others, all of which tend to suggest, in spite of centuries of supposition to the contrary, that if left to their own devices and allowed to act selfishly, folks will tend to get on with each other - in life, co-operative strategies will tend to be more successful than uncooperative ones, so people naturally inclined to cooperate will tend to flourish. This is contra Marx, Hobbes and so forth, but stands to reason when you think about it: if this principle were not true at the most fundamental evolutionary level, it is hard to see how we would be here to argue about it.

Despite that, Ball's liberalism does show through, and in odd ways, in a couple of places. One result suggested by research is that many distributions in society - sizes of incomes, cities, businesses and so on) will tend to be arranged according to a power law, rather than a normal distribution (that is, there will be a large number with a broadly similar size, and a very few with a very much larger size). Traditional social-liberal orthodoxy is that this is a bad thing, and by implication Ball thinks so:

"This is not to say that power law disparities in the free market are inevitable. But it does suggest that, if we decide they are undesirable, we shall probably need to restrict some of the freedom with which the market operates."

Unobjectionable, centrist sentiment you might say. But hold on: a free market assumes the free participation of everyone in the market (otherwise, it wouldn't be a free market). Now, if that market arranges itself according to a power law, then must that not be precisely what "we" - the participants in the market - have decided, by our very own actions, *is* desirable? We have, all by our own actions, unwittingly colluded to make one city very big, or one company very rich - if that is truly not want we want, we can move, or we can buy a different product. By Ball's own argument, there is no better indication of what "we" decide is what we want. As soon as someone starts talking about what "we" want, overriding the judgment of the market (which, statistically, describes how we collectively behave without needing to measure individual vectors, spins or magnetisms, remember) it seems to me we are in very dangerous territory.

Ball, I think, realises this and never dares more than a wistful look in this direction. In any case, it is certainly not enough to deprive this book of five stars: a thoroughly enjoyable and enlightening read. ( )

Very interesting overview of the application of scientific methods used in explaining physical phenomena to also explain mass animal and human behaviour. Historical background is thorough, but was a bit surprised that the original greek philosophers were not included in background, such as Plato, who I believe already had already discussed political science. ( )

born 1962, an English science writer. He holds a degree in chemistry from Oxford and a doctorate in physics from Bristol University. He was an editor for the journal Nature for over 10 years.

Pros: a relatively complete compilation of related topics; clear writing
Cons: no central thread or the proposed thread is stretchy and not well developed; superficial connections between topics; organization on the chapter level is also messy and all over the place; almost all well-known and old materials ( )

I love reading about complexity science, and this was all pretty interesting stuff. But, there wasn't really anything new, and he jumped from topic to topic so much that it all doesn't really hold together very well. ( )