Customer Reviews

Critical Mass: How One Thing Leads to Another

5 star:		(15)
4 star:		(6)
3 star:		(5)
2 star:		(1)
1 star:		(2)

 

 

Readers of "Critical Mass" by Philip Ball will learn many new concepts and ideas from a skilled science writer with a doctorate in physics. His book opens with brief historical account that weaves the political confusion that engulfed Britain in the seventeenth century into early developments of science, but it is with the work of Thomas Hobbes that the author is particularly concerned. Although others had imagined ideal societies - Plato's "Republic", Thomas More's "Utopia", and Francis Bacon's "New Atlantis" come to mind - Hobbes attempted to deduce the laws of society from basic postulates in the manner that Isaac Newton had recently managed to explain planetary motion. In other words, Hobbes sought to establish a "physics of society" which is also the aim of Ball's book.

Sensitive to charges of "arrogance", Ball asserts that his work is "not an attempt to prescribe systems of control and governance, still less to bolster with scientific reasoning prejudices about how society ought to be run." Rather he would help us to understand how "patterns of behavior emerge - and patterns undoubtedly do emerge - from the statistical melée of many individuals doing their own idiosyncratic thing." Thus he uses the tools that have recently been developed in nonlinear science to understand collective social behavior. To this end, the historical introduction is followed by a discussion of the concept of probability and the corresponding growth of statistical physics that developed in the nineteenth and early twentieth centuries. The general reader who would understand these important ideas will benefit from the early chapters which clearly expound the notion of a phase change (think of boiling water or melting ice). As a central metaphor for much of the book, Ball carefully presents the Ising model, which comprises a two-dimensional array of rotating magnets (think of small compasses) each influencing the orientations of its nearest neighbors. Below a certain "temperature" (random vibrations of the magnets), the magnets all "freeze" into a certain orientation - a global effect that stems from local (nearest neighbor) interactions. To what extent, the author asks, do local interactions among people lead to the emergence of global social phenomena?

Beginning with discussions of snowflake growth, the formation of complex patterns in bacterial colonies, and the dynamics of flocking birds (in which the interactions are local), the author turns to collective phenomena involving humans, including the organization of passing rules on sidewalks and corridors, tragedies stemming from inept crowd control, path formation in parks, and the nonlinear dynamics influencing the growth of cities. These fascinating discussions are followed by a chapter on traffic flow (in which the dynamics of jamming are clearly explained) and several chapters on economics.

In the first of these, Ball considers fluctuating price levels, which Adam Smith deemed to be governed by the collective effect of an "invisible hand" as far back as the eighteenth century. An important aspect of price variations, well laid out in this book, is their statistics. If all the influences on prices were random, the variations would be governed by Gaussian statistics with large variations falling off as a negative exponential of the square. In fact, large variations are often found to be much more likely than in a random process, suggesting the statistics of Lévy flights used unconsciously both by foraging bees and also by Jackson Pollock in his famous drip paintings. Interestingly, an analysis of the S&P 500 market fluctuations shows a power-law distribution lying between Gaussian and Lévy statistics in which the likelihood of a variation is inversely proportional to a power of its magnitude. Power-law distributions have been found to govern many phenomena including the probabilities of avalanches and earthquakes, sizes of individual incomes, and growth rates of firms. From economics, Ball segues into the more slippery area of politics. Appealing to the Ising model, he considers analytic descriptions of the possible international alliances prior to the Second World War, statistics of recent voting patterns in Brazil (which are also found to follow a power law), and various models for investigating balances between social order and justice. Final chapters discuss the nature of interconnecting networks, the World Wide Web (in which the number of links to a site are governed by a power law) and analytic evaluations of strategies for international relations. Surprisingly, Ball ignores the application of collective dynamics to the human brain even though physicist John Hopfield has famously based such a description on the Ising model.

While this book is highly recommended, the author seems unaware of a seminal study of living systems published by Manfred Eigen and Peter Schuster a quarter century ago on how the first biological structures might have first become organized, which showed that three or more interacting hierarchical levels of organization are necessary for self-reproduction. In addition to being important for the emergence of life, this result has deep implications for the emergence of consciousness in our brains. Why? Human brains are organized into cognitive hierarchies, just as living organisms are organized into biological hierarchies, and cities are organized into social hierarchies. To better understand the dynamics of such intricate systems, we must move beyond the concept of emergence at a particular level of a nonlinear dynamic hierarchy to appreciate the possibilities of downward causation and positive feedback networks that extend over several hierarchical levels. Also, the author ignores the vast amount of work in cultural anthropology produced by physicist Franz Boas and his many brilliant students at Columbia University over much of the twentieth century, including Ruth Benedict's classic "Patterns of Culture".

Alwyn Scott
http://personal.riverusers.com/~rover/

This is a sometimes dense, often rambling and always interesting book about the history of science, the history of social philosophy and many points of congruence between the two, from how traffic jams happen to how communities self-organize. Author Philip Ball seems to include almost every notable physical scientist since Sir Isaac Newton as he traces how key scientific theories have influenced or been influenced by the speculations of economists and political scientists. Anyone whose acquaintance with science is minimal, but whose curiosity is deep, will find that reading this book is something like floating down a river that is a sometimes windy, sometimes swampy, sometimes roiling stream of discoveries, ideas, broken hypotheses and curious characters. There are two small flaws. First, the author identifies almost every scientist who ever worked on a problem remotely related to the book's subject and sometimes he does not clear the path through the thicket of names and experiments. And, second, in a social science discussion toward the end, Ball permits his political biases to color his story with occasional, apparently heartfelt, denunciations of right-leaning politicians. These quibbles aside, we say buy this book and enjoy an intriguing raft ride through interesting intellectual waters.

This book is an excellent historical look at how scientists and social scientists have attempted to measure,analyze and discuss the effects and causes of group interactions,be they the interactions of atomic particles or speculators operating on the New York stock exchange.The author provides a superb overview of herd effects,cascades, and other types of crowd effects,as well as a good discussion of how economists have attempted to model the interactive effects of crowd behavior.Readers who are interested in this topic will find a much more detailed discussion in"The Wisdom of Crowds",by J Surowiecki(2004).John Maynard Keynes and Benoit Mandelbrot are both given appropriate recognition for their pathbreaking contributions in this area.Ball recognizes,as did Keynes and Boltzmann before him,the faddish nature of much of the social sciences , economics in particular ,in attempting to mimic mathematical physics in its approach to the use of formal mathematical methods.In many cases this leads to fads which emphasize the mere use of the technique,irrespective of any quantifiable scientific results.Ball points out that the overuse of the normal(Gaussian)probability distribution among economists is an attempt to obtain the self ordering and equilibrating structure of gas particle models within the human domain even if there is no empirical support for such a distribution.Here both Pareto,Zipf,and Mandelbrot receive credit.

My feeling on Critical Mass is that it is a mixed bag. When Philip Ball discusses physics his prose is, to borrow a United Kingdom phrase, "spot on" (I particularly liked his description of entropy). However, when he attempts to extend these descriptions to the social sciences, he sometimes (in this reviewer's opinion) simplifies the line between cause and effect as well as exaggerates foibles of those that came before him.

In some chapters (e.g., "On the road") I would argue that at best, all Critical Mass is doing is importing names from physics to describe similar appearing phenomena in our macro world. However, as the great Richard P. Feynman once said: "simply knowing the name of something is not knowledge". To me at least, there should have been more discussion on experimentation to back up the assertions that the similar appearing phenomena are in fact the same thing. Then it would truly illustrate something deeper. Again, to borrow from Feynman using his famous license plate analogy, if you have already observed the results and then develop a theory it is not science.

I could also have done without some of the condescending comments on the some of the great men that came before that apparently disagreed with the author's politics. For example, when discussing Adam Smith's theories on economics (Rhythms of the Marketplace), the book belabors his theories (e.g., page 180 "...even on its own term's Smith's economic theory was too simplistic to cover the whole story..." or page 184 "...Smith does not endorse the grinding poverty implicit in his words...") In contrast, the author is positively gushing when in the same chapter he describes Karl Marx's theories as "...the most influential of `scientific' economic theories in the nineteenth century..." (page 183) and "Marx's economic vision contained the crucial concept of a market that was potentially unsteady and to oscillate between boom and bust..." (page 186).

Just one man's opinion.

Books about complexity and self-organization might be a dime a dozen, but this book by Philip Ball is unique. Underneath its popular style is a serious analysis of the science that underlies the concepts of complexity and collective dynamic phenomena. The analysis of specific examples of self-organizing systems (e.g., traffic models) is quite professional, with good referencing of the original literature. At the same time, the writing is clear and understandable to a non-specialist, and the examples given in the book will stimulate the interest of most readers. Another strength of this book is its historical approach; the author shows how the dynamic concepts evolved from 19-th century thermodynamics, to non-equilibrium thermodynamics, to the science of non-integrable dynamic systems. I really enjoyed reading it.

Well, this was a good book, but not a great book. The central thesis of this book is that we can apply the kinds of statistical models seen in physics to social situations. This is because many of the population-level properties we see in large groups of people are emergent and primarily the result of interactions amongst the "particles".

Right up to here I agreed with everything the book claims. Unfortunately I just don't buy the physics fetish quite as much as the author wants to. Phillip Ball seems to be arguing that in the future we should be able to predict the behaviour of societies as reliably as we can predict the behaviour of large quantities of gas and that we can do this via an approach very much like that of statistical physics. This argument does not hold for two very closely related reasons. Firstly, a key assumption of statistical physics is that the particles are all identical, except for a few key easily measured properties such as velocity. This is clearly not the case for humans. In almost every respect, every human is different from every other. Furthermore, even assuming we know which properties we need to measure and which we can ignore, the properties themselves are not trivially measurable, especially for the large populations which we need to model events accurately. Any suggestions as to how we might measure how quickly each individual in a population of 10,000 will panic if a few others around them are? Thus the statistical approach starts to break down. The second problem is largely a consequence of the first. In such a varying population of "particles" stochastic randomness will take over. Chaos theory slips in and all of a sudden we can't make reliable long term predictions. A couple of the infinitely many little things we didn't model because they seemed irrelevant turn out not to be. Because we didn't take them into account our model now can't predict past its nose. In chaotic situations every butterfly is relevant.

The end result of all this is unfortunately very simple: the statistical sociologists are left with a devil's choice. They can accept that their models can only be rough generalisations, which means that they will typically only allow weak predictions to be made. This is no worse than the current approach, and is really more scientific and better, but it is not what the book aims for. Alterntatively the "fundamental particles" modelled by the system must be made heterogeneous and just as complex as the models developed by standard sociology and psychology.

If we choose the former than statistical sociology is really something of a dead end, adding mathematical rigour for little benefit. And if we choose the latter than it becomes meaningless, claiming only that "If we have an accurate model of the human mind, then we can accurately model how humans will behave." This letdown is unfortunate because the idea put forward by the book is a good one, and can hopefully be modified into something more general and useful. If that can be achieved it will revolutionise the social subjects, turning them from arts into sciences. If it can't - well, we'll still be able to do the odd useful thing with it, but it will hardly be groundshaking.

I am an academic economist who frequently collaborates with scientists and find that interdisciplinary collaboration is often fruitful and that there are clear synergies from letting scholars from different fields interact. I therefore purchased this book with the expectation of being exposed to useful ideas. I was sadly disappointed. The book starts with stating the need for behavioral physics. I smiled a little at this, since economists have been doing this for 150 years, but I guessed that you can't know everything. Then the author started writing about economics. Oh boy! Either the author is truly ignorant of economic research the last 50 years or this guy is to scientific writing what Ann Coulter is to punditry. I doubt that there is a single economic concept that the author has understood. A fundamental concept such as Pareto optimality is botched. Apparently there is evidence that prices fall when demand grows which would be news to most economists (including Adam Smith who is supposed to have expressed this interesting idea.) The author states that economists look for theories that enables prediction of market movements whereas real traders know that this is not possible. Any economist will tell you that the truth is the other way around. In fact, economists know that unless there is asymmetric information among traders it is not possible to consistently outperform the market. Real traders make a living trying to do so. And the statement that microeconomic models ignore interactions between agents is just laughable. In fact economics is a holistic science in the sense that it is explicitly acknowledged that actions are interconnected. This is what general equilibrium analysis is all about. (I don't recall the publication date of Cournot's market model but it was a long time ago.) I could come up with more examples of nonsense but if you want a laugh, just borrow the book at the library. I am concerned with the disingenousness of the writing style. There is no end to the number of statements like "Some economists believe," "mosts economists will" and "many think that." Since I don't recognize any of this I can only assume that he made these things up. The other approach is to cherry pick bad ideas from the past and present them as current practice. (Kontradieff cycles!!!)

In my interactions with scientists I have often despired about the calvalier attitude they have to economics. They seem to think that having a Ph.D. in e.g. physics means that they are super-smart (which they often are) so something as trivial as economics must be dead simple to them. This leads them to say the dumbest things like claiming that energy should be the unit of account when measuring value. It seems to me that having a brain that understands quantum mechanics and relativity makes you supremely unable to understand elemenetary aspect of behavioral science. (Although you may be able to predict behaviour in queues and other not very interesting patterns.) The contrast with biology is interesting. The flow of useful ideas between biologists and economists has been very beneficial to both sciences and promise to be more so in the future.

This well-written, informative, and provocative book raises almost as many questions as it answers. Again and again, Ball walks us through a relatively simple physical phenomenon and shows that an ostensibly more complex human phenomenon follows the same underlying mathematical patterns.

As intriguing, uncanny and even jaw-dropping as this gets, it's not always clear what it means. Is it sheer coincidence? Just a point of explanatory departure? A metaphor? A heuristic? Is the apparent connection nothing more than a byproduct of the math? Or is it a genuine case of unexpectedly like causes generating like effects?

Careful not to bloat his conclusions, Ball does not always provide a working theory that connects the simple and the complex phenomena. This can be unsatisfactory, as we are left to marvel at the parallels without gaining a fresh insight into the nature of the phenomena on either side of the comparison.

I would like to say that this book is the basis for Asimov's "psychohistory". It isn't, but it is a great collection of social systems and ideas presented as physical and mathematical models. Starting with simple phyiscal models such as spin glasses and ending with game theory, this book is a treasure of lucid explanations about how these models apply to the system being modeled and their limitations.

I found this a very interesting read and I appreciated that the author does not breathlessly hype the models, but explains the reasoning behind each and details the results, where they work and sometimes where they fail.

About a month ago I read "Critical Mass: How One Thing Leads to Another" by Doctor Philip Ball, who is a physicist by education and a science writer for Nature. I read it because of my interest in networks and complexity, and this book promised some insights. It was an entertaining read, but was hard going in a lot of places. Despite this the book was well written and flowed naturally.

I did enjoy the first chapter on Thomas Hobbes "Leviathan", probably because I like philosophy! The point of the chapter was that mankind has been trying to discover universal laws of society since at least 1651, and continues to do so. In subsequent chapters at times I was disappointed and found myself saying "So what!" or "Where is the evidence?" - cause and effect where not always clear to me.

What did interest me was the convergence of the hard sciences with the soft arts. By this I mean Ball takes proven examples from physics and applies them to social phenomenon. Sometimes this worked and sometimes it didn't, but nonetheless is evidence of a growing convergence - see for example my review on "Weak Links: Stabilizers of Complex Systems from Proteins to Social Networks" by Peter Csermely. This convergence is most evident to me when small worlds and networks are examined. I always find discussions on phase transitions interesting, particularly when applied to complex human systems, although I must say I am not completely convinced. I am also cautious in applying laws of physics to human behaviour. Particles don't have conscious behaviours - human's do! In fairness Ball does caution against blindly taking scientific laws and applying them to situations like crowd behaviour.

All in all it was an entertaining and thought provoking read, providing one approaches the subject matter with an open and sceptical mind. In this case critical thinking matters!


Regards, Graham