I took a few notes when reading the book “The Wisdom of Crowds” that I post here.
James Surowiecki’s book is about how smarter are groups than the individuals than compose them separately, under certain conditions that the author enunciates. For instance, a crowd can calcutate the weight of an ox, the outcome of a general election or of a drug test more accurately than the best experts, consistently over time.
The production of superior average judgement of a number of people regardless of the best ability, engagement and information of the individuals, or a small subset of them, composing the group. In other words, the averaging the answers of the individuals of larger groups result is answers of counts or any quantity more accurately and precisely than their brightest of its individual components.
This wisdom works particularly well for simultaneous, well-defined problems and with a limited and pre-determined set of solutions determined like polls or contests. The book comes short of proposing new practical applications of the wisdom. Which practical problems can be solved with this newly found Wisdom of Crowds? The author claims that “the implications [of the wisdom of crows] for the future are immense”. He suggests that security intelligence would benefit from probability and decision markets.
Physics as a Social Science
I read The Wisdom of Crowds after Philip Ball’s Critical Mass book. Critical Mass describes how physics is helping understand some phenomena of social science. I enjoyed reading both immensely but I feel that what I learned from Critical Mass influenced the way I feel about the Wisdom of Crowds. I miss in this one the approach of asking and researching the why, not only the how, of the collective phenomena described on the book. I also would have enjoyed a pulse to cross over the science of sociologists and social psychologists with the ones of mathematicians, ecologists and statisticians.
Crowds solve only a few types of problems, and some better than others
- cognition: weight of an ox or where to build a public swimming pool
- coordination: stock markets, traffic, organization in companies. Coordination is well harnessed by companies like Zara, which is basically a logistics-centric organization.
- cooperation: paying taxes, fighting pollution, democracy. Cooperation looks like the type of problems that crowds usually fail at from the corruption of the Italian football league, measurement of TV audiences by Nielsen
Conditions of the individuals that make groups intelligent
- private judgement: information, analysis or intuition
- decentralization, eg. scientific efforts to fight SARS epidemia
Diversity and independence are important because the best collective decisions are the product of disagreement and contest, not consensus or compromise.
Factors that affect negatively decision making by groups
Sequential decision making. The best technology will not necessarily win in a market-driven selection process. It is a wasteful process
Drift towards consensus over dissent
Verdict-driven juries over evidence-based juries
Armstrong’s seer-sucker theory: “Not matter how much evidence exists that seers do not exist, suckers will pay for the existence of seers.”
New information by a few is ignored, misinterpreted or modified to fit old messages
Group polarization, sequence or status deference. Talkative people are not necessarily well liked but they tend to be influential. Very few human beings perform consistently well in an environment of negative reinforcement
Summary of the ideas of the book
- The Difference Difference Makes:
- Waggle Dances: send out as many scout bees as possible when the alternatives are unknown. Examples of gasoline-powered car reaching mass production status over steam-powered cars.
- The Value of Diversity: Expertise beyond a minimal level is of little value in forecasting change
- Monkey See, Monkey Do: Imitation, Information Cascades and Independence
- Putting The Pieces Together: The CIA, Linux and the Art of Decentralization?
- Shall We Dance? Coordination in a Complex World. Imperfect markets composed by irrational people can still produce near-ideal results
- Society Does Exist: Taxes, Tipping, Television and Trust. Willingness to punish bad behaviour even when you get no personal material benefit from doing so. Be nice, forgiving and retaliatory if you expect successful cooperation.
- Traffic: What We Have Here Is a Failure To Communicate. Congestion charging leaves the decision to drive or not in the hands of the individual.
- Science: Collaboration, Competition and Reputation
- Committees, Juries and Teams: The Columbia Disaster and How Small Groups Can Be Made To Work. A successful face-to-face group is collectively intelligent, it makes everyone work harder and think smarter: the intellectual swing.
- The Company: Meet The New Boss, Same as the Old Boss? Decentralization allows to make decision and become engaged. It also makes coordination easier. Irrational people can add to collective rationality.
- Markets: Beauty Contests, Bowling Alleys and Stock Prices? Shorting stocks is riskier than buying them. Investors are concerned nut just with what the average investor thinks but with what the average investor thinks the average investor thinks. A crash is the inverse of a bubble, although more sudden. We do not know why crashes occur or why bubbles start.
- Democracy: Dreams of the Common Good?: a healthy democracy inculcates the virtues of compromise – which is, after all, the foundation of the social cotntract
Some of the things the book made me think about
The rule by a technocratic elite fails because of small-group dynamics, groupthink and lack of diversity. Groupthink works not so much by censoring dissent as by making it seem somehow improbable.
The judicial system is a particular case of non-elected, insulated elites. Journalists employed by media powerhouses in representative democracies. Scientists and mathematicians also suffer from similar problems but this is partially mitigated by a few factors. Incentives and reputation rewards are usually available to those scientists turn curiosity into results and challenge established views. Reputation is not the only transaction of the scientific market: eventually the value of new ideas and empirical evidence is also part of the transaction. Part of the job of scientists, at least nominally, consists in verifying the rigour and robustness of the work of other scientists before building further work upon it.
My questions about some points of the book
Surowiecki is more interested in the how than in the why. In this regard, his book reads like a descriptive essay. I miss a bit more reflection into why groups work well solving some problems and not others. The author simply mentions that “this is the way the world works”. It is funny how smart authors have different approaches to their ideas. Philip Ball’s Critical Mass book on one hand goes, in my opinion, too far in using Hobbes’ as a base to explain many ideas, by consistency or contrast. Surowiecki is not less profound in its ideas, but it touches many of them without historical references.
Is the efficiency of groups at solving numerical problems yet another evidence of phenomena involving systems of minimum energy? The experiments of Epstein with inter-dependent agents prove that agents are lazy; they want to do as little thinking as possible. I wonder if independent agents share with dependent ones the spontaneous conformity to whatever is minimal energy, which also happens to be, on average, the “right” strategy, solution or state of a physical system?
Why is the arithmetic average the best metric to assess the wisdom of crowds? Is the distribution of the answers normal? Are the mode or median even better estimates? This question reminds me of my excitement about getting the book “The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty” by Sam L. Savage in a few days time.
Wisdom might not be the right term for the phenomenon of crowds being accurate and precise. Wikipedia defines the term as “Wisdom is a deep understanding and realizing of people, things, events or situations, resulting in the ability to choose or act to consistently produce the optimum results with a minimum of time and energy. …” Can the phenomenon be defined as “deep”? The crowds seem to be good at solving some quantitative problems, but can we assuming that that is deep understanding or is it simply the law of mechanical physics at work?