Excellent book if you're looking to explore the theoretical underpinnings of living in the world. Very engaging - think of "Freakonomics" on steroids. A good science background in the reader leads to lots of 'a-ha' moments. ( )

I'm not really sure how I came to get this book but I must have assumed it was a bit like "Tipping Point" by Gladwell. Though as the author, Philip Ball, describes it very late in the piece it is a book about the value of physics-based models in the social sciences. He certainly packs a lot in. He starts with an overview of Hobbes political thinking ....that the idea society is covered by scientific reasoning and the people elect a leader and endow that leader with huge authority.
He then digresses into a discussion of thermodynamics and the law of large numbers....essentially the idea that you don't need to know in detail the behaviour of each "particle" (or person) it can be sufficient just to know what happens on average. And what happens on average very often fits a bell shaped probability curve. Ball then moves into the area of his specialty...phase changes and there is a very good section of the physics of phase changes between say water into ice and into a gas. (I think he did his PhD at the University of Bristol in phase change physics). A basic point made in these relationships is that as you approach the critical point (for either a liquid or a magnet being heated to the point at which it is demagnetised) the percentage RATE at which it approaches the critical point is the same for fluids and a certain class of magnets. The phase transition is a sudden jump between different stable arrangements of particles. Developments in this area have been termed statistical physics.
He then moves on to discuss snowflakes and other kinds of structures such as "diffusion limited aggregation" that you can get with mineral deposits in rocks ......it looks like fossilised ferns but follows fractal laws.the snowflakes hexagonal symmetry reflects an underlying unity at the molecular level. But for non equilibrium growth (eg convection patterns in heated liquids) there is no overarching theory comparable to the statistical physics that describe states and phases in equilibrium.
He then digresses into a discussion of biological organisation such as with bacteria growing in spiral patterns, slime mould acting in a coordinated way, birds forming flocks and discusses the simple rule developed by Craig Reynolds...a software engineer; match the speed of birds nearby and move towards the centre of mass of the flock and avoid collisions. Visek and Czirok came up with a similar formula for bacteria; travel at same speed and in average direction of all others within a certain distance....but they added in some random "noise". At low noise levels the cells displayed collective behaviour...raise the noise and the degree of coordination declined. At some critical level all coherence was lost. Apparently when you run similar trials on models of people moving in confined spaces ...such as a corridor but with no rules about avoiding etc, certain sorts of behaviour emerge spontaneously..and some sorts of crowding etc occur too. By putting some barriers to guide people at intersections or providing two exits instead of one the natural behaviour can be made much more efficient. This sort of planning can be very important in situations where people can panic and jam exits.
He briefly plays around with various models (including fractal models) governing the growth of cities and explains that the models can provide a fairly good fit with actual cities. OK, maybe so but it is retro-actively fitting the model to actual data and I remain unimpressed. He seems to force fit a lot of his theories to real world data.
Moving right along from people crowding, Ball looks at various traffic models and notes that a kind of hysteresis applies with vehicular traffic; the vehicles flow well up to a certain critical density ....and provided nobody slows down or does anything unusual ....the traffic will flow smoothly beyond this point into the metastable zone. But once there is a disturbance in the flow, the volumes will crash and then only very slowly recover. So the impact of a traffic jam moves upstream, (against the traffic flow). The traffic flow models seem to be proving their worth in real life according to Ball. It's been used to plan traffic flow in Dallas, for example.
He then commences a section looking at economics and the hidden hand of Adam Smith. He considers a variety of economic models (Marx, Schumpeter, Keynes) but generally he seems to adopt a very sceptical attitude towards economists and their application of models. He questions the whole notion of business cycles. And demonstrates that stock market fluctuations do not follow a Gaussian curve...but per Mandelbrot, they execute a "Levy flight...or random movements punctuated by sudden large leaps. (Like animals searching for food). The Black-Scholes model (Nobel prize awarded) more or less defines the state of play for options but it assumes gaussian fluctuations in the market and doesn't account for volatility..therefore can produce nonsensical results.There seems to be some evidence that traders and firms take note of what others are doing in the market and revise continuously...but this merely then reflects the messy, unpredictable nature of the real market. Ball is critical of the blind adherence to a free market philosophy.
He claims that critical points are at least a good metaphor for the strange combination of the unpredictable and rule-bound influences that govern our lives
If you make a sand pile and measure the probability of an avalanche vs the size of the pile it follows a power law...at least up to around 800 grains. (His diagram seems to falter about this level...and that's not a very big pile.....apparently it works better for rice grains....presumably polished rice!). The pile is in a critical state but is not in an equilibrium state. And this self organised criticality model has been applied to earthquakes, forest fires, conflicts such as wars. George Zipf suggested that the power law with a slope of -1 is characteristic of of phenomena where people act in groups rather than as individuals.
The size of firms and the growth of firms also appear to fit power distributions but this knowledge doesn't really tell you if Brian's butchery will survive a new supermarket opening up next door.And he describes Axelrod's model of firm alliances which pushes firms in a direction of stability. (It's like a certain model of the ferromagnet). He claims that this predicts the alliances formed in the second world war.....But I'm not really impressed. Spain is left out of the equation altogether as is Switzerland, Belgium, the Netherlands. Portugal's alliance is wrongly predicted. And the whole exercise really depends on a kind of arbitrary allocation of values to the measures used.
Undeterred by any of such criticism, Ball suggests that we can now start to understand how a society creates its leaders, its customs, its fashions and its problems through a mass of mutually interdependent decisions. A first past the post democratic voting system has a lot of weaknesses (but he doesn't discuss Australia's compulsory, preferential voting system...which also has its problems but gives a better reflection of the real popular sentiments). Joshua Epstein of the Brookings Institution has developed a model where each "agent" makes up their mind according to the majority view of those surrounding it. If you introduce some noise into the system it makes the boundaries less likely to resolve into distinct blocks. Ball says that "as a metaphor for social decision making, the model is absurdly simplistic - and irresistibly attractive. Schelling used a similar model in predicting patterns of racial segregation and Campbell and Omerod have postulated that you can get sudden jumps in criminality due to small changes in social circumstances. He points to the case history of crime dropping in New York under Mayor Rudy Giuliani. Ball suggests this is very similar to phase transition in liquids/gases. And a curve developed plotting the economic incentive to stay married against the proportion of people married shows a kind of of "S" bend which is very like the curve showing first-order liquid gas transition, vanishing into the critical point. If one includes "strength of social attitudes towards marriage" as another dimension you end up with a surface with a critical point....akin to a similar plot of temperature, density and pressure for fluids. It's this sort of exercise that I started to find a bit annoying. He poses the model from physics and suggests that it fits some social phenomena ...where the social variables are either very hard to actually measure (eg social attitudes towards marriage) or are just guessed at and then doesn't really actually draw the conclusion that they are equivalent or simply suggests that the social world can learn from physics. I guess it is reasonable to conclude that reduction in crime rates is due to more than repression. But does one really need a model from physics to tell us this? And is this the complete model for crime in society ...or are there other important variables not included in the model. He is very weak when it comes to plugging real data into the models and many of these models seem to be "proven" with the benefit of retrofitting (eg the models of alliances in WW2).
Ball then moves on to a consideration of how cultures merge: how some differences persist and others disappear. He cites Axelrod, again, with a grid type model where likeness in adjacent cells increases the probability of adoption/ineration. After thousands of steps the model is essentially homogeneous ...maybe with some standout islands. If there are a large number of cultural values then the changeover to a monoculture is not gradual but happens suddenly...just like a power law covering the interaction of particles. And following this there has been a lot of work done considering how agents behave locally with each other. Ball suggests that this may hold the key to to a more rigorous science of society. But he also has a bet each way by pointing out that we don't really know wether the adoption of one set of rules or assumptions will lead to truly representative behaviour or will just be a crude caricature.
The next chapter is about ideas like the six degrees of separation and connectedness.and looks at the Kevin Bacon number (the number of movies that an actor has appeared in with Kevin Bacon or has appeared with an actor who has appeared with Kevin Bacon). According to a model developed by Strogatz and Watts, as you move from fully ordered wiring to random wiring the characteristic path length plummets. Basically you get short cuts between vertices. But the structure of a social network is not the full story on connectedness because it depends on the actions and perceptions of people. (Which indicates to me that it's not a great "social model").
As attempts have been made to model the world wide web, it appears that the probability distribution of incoming hyperlinks follows a power law in relation to the number of links (a straight line on a logarithmic plot....though appears to fall short as the number of connections rises.) Ball suggests that this is an indication that the system is scale free. And a scale free network "barely notices it if up to one in 20 of it's nodes is disabled. A relatively large connected cluster still persists even when half the nodes are knocked out. But the weakness of scale free networks is their vulnerability to planned attack. If hub nodes are removed the network rapidly falls apart. Yet with viruses in computer networks they seem to linger and last for a long time even though they can be patched quickly...so topology matters and we may need to re-think how human epidemics spread. In a sense, Ball claims, we can see network theory as a natural extension of statistical physics....a kind of "sticky" version of many-particle systems.
At this stage, in a real flight of ambition, Ball moves into political philosophy: What is the best form of government? Runs though Hobbe's version, Skirts around world government, digresses into the spontaneous truces in WWI . (I remember reading a letter in indelible pencil...slightly purple....written from the trenches to my grandmother where the soldier says something along the lines....They are just like us but I have to keep telling myself they are the Boche and we mustn't think of them as being ordinary people". One can see the conflict between one's natural humanity and the brainwashing that the soldiers were subjected to. Anyway, Ball continues with models that build in some temptation and has a lengthy discussion of the prisoner's dilemma..and Alxelrod's experiments with multiple iterations to try and come up with the best strategy for cooperation or defection. (I've read Axelrod's book...The evolution of cooperation). Pretty much the best strategy always is tit for tat...but sometimes a slightly more forgiving form of tit for tat gets better outcomes in terms of cooperation. There is speculation that such behaviour has become hard wired in us by evolution. So is reciprocity good for us? Some experiments have been run with varied strategies and tit for tat can emerge rapidly (after a period where defection rules) from the ruck. The abruptness of the change is reminiscent of a phase transition. (Ball sees them everywhere). But empires have risen and fallen over time...even the long lasting ones so Axelrod has explored the notion of territoriality in Prisoner's dilemma......and the ideas have tended to penetrate think tanks on defence strategy. Though there is a remarkable thought experiment described by strategist Herman Kahn with colleagues at the RAND Corporation during the early years of the cold war (and mutually assured destruction). Here a multi megaton bomb has been dropped on NY. So what next? asks Kahn. The US president presses all the launch buttons was the answer. What happens then? asks Kahn. The soviets do the same. And then what happens? Nothing...both sides have been destroyed. (To say nothing of the rest of the world subject to radiation etc damage). Then why did the president do this? A general rethinking then would follow with a conclusion that maybe the president should not launch an immediate retaliatory strike. The trouble with nuclear weapons is that there is nothing left to negotiate with ...nobody to "tat". Similarly with the death penalty...if you happen to get the verdict wrong first time around...you get no more "iterations". There is nothing in prisoner's dilemma to say that one player learns from the mistakes of another ...that the death of one warns the other away from defecting. But, if we know that cooperation is possible, even in a world that lacks altruism, the we have no reason to despair.
Utopias: The problem with utopias is that one can never be quite sure about how they will turn out. Human nature is the unpredictable element. The notion that we could ever construct a scientific utopia theory is doomed to absurdity. Certainly, says Ball, a "physics of society" can provide nothing of the sort. Concepts and models from physics are almost certainly going to find their way into other areas of social science, but they are not going to provide a comprehensive theory of society. One of the strongest themes to emerge from the physics of society is that individual behaviour is often no predictor of its social consequences. Society is not just a scaled up person. The physics of society is full of examples of actions that generate effects other than what was intended. The rush to escape a crowded room leads to a slower average rate of exit. Increased congestion can follow from building new roads. Ball suggests that Statistical physics may help liberate planners and policy-makers from their propensity for linear thinking and to encourage a greater sophistication in their perception of cause and effect. So it seems that Ball is suggesting that the best we can really hope for with social planning is a bit more lateral thinking by planners. OK. but that's a lot less than he seemed to be suggesting in the bulk of the book.
Many economists today (according to Ball) still have a Smithian objection to any form of market regulation. Yet there is not a single economic theory that can show that a totally free market sets the socially most beneficial price for goods or leads to optimal distribution...in fact it tends to accentuate inequalities in wealth. The real issue is whether we can trust ourselves to distinguish between what is and what ought to be...the traditional moral dilemma raised by David Hume....though Ball cites Adam Ferguson...who was an associate of Adam Smith (and Hume). The idea of using physics to empower change and to predict its limits has no intrinsic politics... Society might naturally develop along certain lines but the possibility of changing those lines by conscious and intentional intervention does exist. So the physics of society is and can only be a tool, never a moral compass. ( )

Did not finish . . . ( )

As a student (not a very good one, arguably :) of social sciences, I've been more than a little disturbed by what I've seen as trying to "prescribe" how people "must" behave using statistical inferences. This book finally made me understand that social science has nothing to do with prescription, or even with figuring out what is going on in individual human minds. Rather, there are two questions: a) How people tend to behave in certain circumstances b) If they behave that way, what happens.

Also, it sparked my interest in using models to study behaviour.

The last, but not the least, it is quite easy and entertaining to read - popular science at it's finest. ( )

Why is society organized the way it is? Is it possible to use some of the laws of the physical universe to understand why and how national economies, stock and commodity markets, companies and clubs organize the way they do? Can physics provides "laws" of human nature that are as useful and universal as those of mechanics? Does the critical point in a phase diagram have analogies in human behavior? Veteran science writer and physicist/chemist Philip Ball writes very well, as evidenced by Elegant Solutions, which was one of my picks last year. He also thinks very well, as evidenced by this creative application to sociology of concepts familiar to physicists and chemists. Of course, social scientists have always tried to be as "scientific" as possible, using mathematical models, statistical analysis and, more recently, computer simulations to understand the human situation and to predict its future. Critical Mass uses a different approach: using whole concepts in physics for insight into economics, urban planning, and the self-organization of human networks. This is a very original and thought-provoking book; it has been recognized with the 2005 Aventis Prize for Science Books, bestowed annually by the Royal Society for Chemistry. ( )