From Wikipedia, the free encyclopedia

Contents 1 me 2 The Schizophrenia of Crowds 2.1 Wise, Mad, or All of the Above 2.2 When Crowds Work 2.3 How Crowds Fail to Impress 2.4 Group Collaboration and Net Crowds 2.5 References

[edit] me

After hearing Jimbo [1], I am was inspired to become a more active Wikipedian. Unfortunately, Wikipedia does not currently allow [Tor] users to edit. Perhaps I will be able to contribute in the future. 128.227.185.89 14:49, 3 October 2006 (UTC)

[edit] The Schizophrenia of Crowds

This is an essay written for EGN1935 which I am taking at the University of Florida discussing crowd behavior in general and in the context of Wikipedia. Hi Bill! Hi Jennifer! Hi Mom!

Hi Nile! (from Jennifer)

by: Nile / John Li

[edit] Wise, Mad, or All of the Above

A crowd of humans is extremely complex. Molded by years of socialization, thousands of years of evolution, and a variable amount of personal contact, their group behavior is highly unpredictable. From this complexity, both "wise" and "mad" behavior can arise.

In The Tipping Point, Malcolm Gladwell argues that large, seemingly chaotic changes can be explained by a small number of principles that are dependant on a small number of people [1]. This point of view suggests that crowds may be controlled by elite groups and thus makes them seem rather sheep-like and unintelligent.

Yet there are many striking examples of crowds defying entropy and making sense and progress out of chaos. From traffic flows to free markets, families to nation states, human interactions are deeply nuanced and highly variable.

New technologies that have enabled a greater level of interconnectedness among humans are wonderful for testing theories of crowd behavior. In particular, the growth of the Internet in the late 20th century has provided interesting data.

[edit] When Crowds Work

James Surowiecki advances the view that decisions made by groups are often better than those made by any individual member in his book The Wisdom of Crowds [2]. His examples include economic markets that exist without centralization, coordination among loosely-connected groups of people, and optimization by "dead reckoning" of groups as opposed to deliberation by "experts".

In "prediction markets", an aggregate answer to a query often turns out extremely well [3]. Whether the prediction is the weight of an ox, prices in the stock market, or the winner of an election, group guesses are almost always better than the guesses of expert committees or individuals.

On a busy Manhattan sidewalk, pedestrians do not resort to fist fights to make their way or throw their hands up in frustration and give up. As viewed from above, humans navigating the sidewalk traffic flows seem like a well-organized, ant-like super-organism. People are also expert psychologists, endowed by evolution with the uncanny ability to empathize with and understand others. Thus, a society of many hundred thousand operates relatively smoothly.

[edit] How Crowds Fail to Impress

It is necessary to examine the situations in which crowd intelligence fails. Surowiecki's requirements for "smart" behavior in crowds limits the usefulness of his theories. Members of the group are required to have diverse opinions and backgrounds, independent thought processes, and a method to pool their disparate opinions into an average. These requirements almost never hold in natural social networks. There is certainly a large degree of heterogeneity in opinions and backgrounds in most societies, but people are not islands --- they communicate their opinions in the open with one another. In general, humans have the tendency to want to trust fellow members of the group; this bias means that they do not have truly independent and unique opinions.

The problems that Surowiecki suggests that crowds handle well are also not particularly impressive. While pedestrians are able to manage the flow of sidewalk traffic well enough to avoid hand-to-hand combat, the problem simply not remarkably difficult. Humans have a relatively small profile (compared to cars and shopping carts, for example) and are able to maneuver with high precision. A more interesting example of crowd optimization is street traffic. Cars are far more difficult to control than one's own body and the effects are clear: fatal collisions, road rage, and hundreds of thousands of dollars lost to body shops and traffic cops.

Similarly, the fact that large groups of people can make better aggregate economic decisions than a single expert economist is not surprising. A single mind, no matter how brilliant, is a single point of failure. A group of minds is dynamic and adaptable. The Law of Large Numbers suggests that a large sample of economic guesses from a crowd would be more accurate than a single guess, regardless of the level of expertise and intelligence supporting it [4]. The Law of Large Numbers is also why a crowd at a county fair can come up with very accurate guesses for the mass of oxen.

Examples of crowds failing are as copious as bad similes used in student papers written hours before a due date: the perpetuation of the Eskimo "words for snow" myth [5]; Kitty Genovese's murder and the bystander effect; the Cultural Revolution; the Salem Witch Trials; the Crusades; and the dot-com Bubble and Tulipomania [6].

[edit] Group Collaboration and Net Crowds

Can a crowd --- which have a history of being dysfunctional --- really publish a completely free compendium of human knowledge or build a free operating system? The existence and success of Wikipedia and Linux suggests that crowds may indeed be capable of such great feats. But there are a number of characteristics that distinguish Wikipedians and Linux kernel hackers from an average crowd.

The Internet actually helps facilitate the creation of Surowiecki's smart crowd behavior. The global nature of the Net enables an incredibly diverse collection of environments and opinions. It allows instant communications while keeping the thought processes collaborators separated by time and space (not being able to go to a bar and chat results in the insulation of ideas and opinions). However, simply having a framework to encourage the creation of smart crowds is not enough.

Projects like Wikipedia and Linux are not powered by democratic crowds. They are hierarchical meritocracies. Linux kernel patches will not get accepted on a "first come, first serve" basis for the sake of fairness --- it is preferable to reject cruddy code and be less functional. With Wikipedia, long-time editors are treated with respect and trust while anonymous edits are eyed suspiciously carefully vetted. More and more of Wikipedia has gone into various levels of "protection" as it has grown in popularity. Some are completely locked, but many more are just restricted to registered users (the Wikipedia and Elephant articles, for example). Wikipedia is often thought of as the ultimate in democratic publishing; indeed --- if one's willing to accept that some Wikipedians are more equal than others.

Literal democracy (literal in the sense of "literal interpretation of the Bible") is not necessarily a desirable state. Just as freedom of speech defends against the "tyranny of the majority" outshouting minority voices, meritocracies can defend against the anarchic nature of the semi-anonymous Internet and allow people to get real work done. Wikipedia and Linux are wildly and surprisingly successful volunteer-based projects --- but they are nimbler, specialized crowds with no pretenses of total egalitarianism.

[edit] References

[1] "The Tipping Point." Wikipedia. http://en.wikipedia.org/wiki/The_Tipping_Point_(book) [2] "Wisdom of Crowds." Wikipedia. http://en.wikipedia.org/wiki/Wisdom_of_Crowds [3] "Prediction market." Wikipedia. http://en.wikipedia.org/wiki/Prediction_market [4] "Law of Large Numbers." Wikipedia. http://en.wikipedia.org/wiki/Law_of_large_numbers [5] "Eskimo words for snow." Wikipedia. http://en.wikipedia.org/wiki/Eskimo_words_for_snow [6] "Tulip mania." Wikipedia. http://en.wikipedia.org/wiki/Tulip_mania