Review: The Perfect Swarm

日期:2013/07/18

The Perfect Swarm by Len Fisher is a model of a what a good quasi scientific book simplifying recent findings in the theory of complexity, chaos, crowd behavior, decision making, swarm intelligence, group think and social behavior in humans and insects should be. The author is a physical scientist and inventor with uncommon insight into many subjects that are relevant to our everyday and market life. Typical is his perfect one sentence summary of the Madoff fraud: "Investors collectively deluded themselves into thinking that he must be cheating on their behalf rather than his own." Exactly. There was always the hope that he would front run the limit orders of the over the counter market making operation he ran on their behalf and this led many to gloss over the complexities and fuzzy explanations of how he purportedly made profits for the investors, marks, and stooges. What's surprising is that someone not close to the markets could see this so clearly. It's typical of the Fisher insights and laser ray reporting of many of the studies and anecdotes in the complexity field.



The book opens with a discussion of how locusts and bees manage to go about their lives and find food. They follow three simple rules with wide application. Avoidance, alignment and attraction (move toward the average). These rules when quantified explain a myriad of patterns in human, fish, bird, and insect behavior. How would they apply to markets? Certainly the alignment would relate to following the trend. Attraction would be a form of moving to the limit orders and stop orders. And avoidance—reduce vig. "Computer simulations have shown that synchrony emerges because each locust acts as a self propelled particle whose velocity is determined by those of its neighbors according to a simple rule". Robots can be trained to find the right place to move to with similar rules and humans in crowds follow these rules whenever the density reaches more than 1 person a foot. We don't flap our wings but keep our arms close to our sides. Instead of diving we shorten our steps. The bees and ants get to their food sources by following in the paths of a knowledgeable leader. The investors tendency to follow Buffett, Gross, Soros, and the leaders of the Fed springs to mind.

There is a nice discussion of how averaging can lead to better estimates of things like weights, marbles, or answers to tests. Galton is treated respectfully here although Fisher is quite clear that he is a fellow traveler and agrarian who believes strongly in the idea that has the world in its grip, and is in the anti Bush, climate change is the source of all evils camp. I didn't like the discussion that follows showing how Page's diversity theorem can lead to reducing error as a function of diversity. It seems to me to be an identity similar to the corrections made for between group and person means in standard experimental designs.

The discussion of group think was particularly valuable to me. He shows how it led to the Columbia and Challenger disasters. He reveals the little known fact that Feynman's demurrals to the group think that administrative failure was the cause but that physical problems with the O Rings resilience to cold was the cause were suppressed. The leader of the report, Rogers, a flexion of the Rubin, Paulson persuasion considered Feynman a trouble maker and tried to keep Feynman's conclusions from the public and his own group members. Apparently similar suppression of divergent and unpopular views figured in various assassination reports that have come down the pike. A strong leader can cause many errors and damage by not being open to divergent views, as we know from the Lehman bankruptcy, and the mortgage crisis. I know of many who traded derivatives with big losses who also would have benefited had they encouraged a Feynman in the wings and paid more attention to such a independent thinker. Perhaps it would be advisable for the bearded chair to encourage diverse views about the wisdom of the bailout and quantitative easings, and whether there is more harm to the common man in such than the benefits to the banks whose bad assets are being purchased.

The most insightful chapter of the book, and the one with the least tarnish from hauling in the grab bag of pseudo scientific psychological studies that make up the unfalsifiable, contrived, irrational behavior paradigms that come down the pike from Kahneman and his followers in New Mexico is the chapter on decision making under uncertainty. He lists five ways of making decisions: Recognition, fluency, tallying, taking the best, and satisficing. All of them are highly relevant for market participants. The chapter is footnoted with good summaries of the literature in each field. A typical finding is that just making a list of the good and bad reasons for doing something and putting them on a balance scale without weighting leads to better predictions than regressions. There is also a nice summary of satisficing citing Mosteller's work that a good rule is to take the first third of the applicants, see which is best, and then choose the next one from the remainder that exceeds the best. This rules leads to amazingly good chances of finding the best 5 or 10% in a sample. I have always felt that applying this rule to the first 10 extremes in a day or week might lead to similar beneficent results. An interesting aside as to how the great social media companies Yahoo, Amazon, Google modify these rules for their fast moving markets to hire, fire, set boundaries, and exit ensues. Regrettably many of the star companies mentioned like Nortel, Orticon, and Enron have fallen on hard times using these heuristics.

There are too many good anecdotes in this book to pass over. For example, James Thurber reports in his bio that everyone in Columbus, Ohio, started running madly thinking that a dam had broken instigated by just one random sighting of a person running with "practically everyone in the city following him madly in the first half hour". In talking about the dangers of group think he points out that if we had not believed in basic research we would never have had x-rays, antibiotics, radio, television, or cars with inflatable tires. It is refreshing to know that in many panic situations, in crowds trying to escape fires and crushes, "the overwhelming reactions of people to the crowd pressure was to attempt to help those nearby who seemed to be in trouble".

The book is well footnoted with 72 pages of footnotes with discussions and very well indexed. If it weren't for the defects in many of the quasi scientific studies he cites, it would be useful to a wide range of investors without digging deeper into the subject. However, on the mathematical and statistical aspects of the studies he cites, the author often fall prey to the promiscuous hypothesis bias wherein whatever results come out from the study, he attributes to a real behavioral regularity rather than a random permutation of the data or a reasonable attempt at rationality when putting the contrived laboratory situations into a real life framework. One highly recommends this book for learning, stimulation, and profit.