Reinventing Discovery_ The New Era of Networked Science - Michael Nielsen [8]
Kasparov versus the World was not the first game to pit a chess grandmaster against the World. Three years earlier, in 1996, former world chess champion Anatoly Karpov also played such a game. “Karpov Against the World” used a different online system to decide moves, with no game forum or official game advisors, and giving World Team members just ten minutes to vote on their preferred move. Without the means to coordinate their actions, the World Team played poorly, and Karpov crushed them in just 32 moves. Perhaps influenced by Karpov’s success, Kasparov admitted that before his game he “was not anticipating any particular difficulties,” and was confident that he “would be able to finish matters in under 40 moves.” How surprised he must have been.
Amplifying Collective Intelligence
Examples such as Kasparov versus the World and the Polymath Project show that groups can use online tools to make themselves collectively smarter. That is, those tools can be used to amplify our collective intelligence, in much the way that manual tools have been used for millennia to amplify our physical strength. How do these new tools achieve this amazing feat? Is it just a fluke? Or can online tools be used more generally to solve creative problems that defeat the ingenuity of even the cleverest individuals? Are there general design principles that can be used to amplify collective intelligence, a sort of design science of collaboration?
A common approach to these questions is to suggest that online tools enable some sort of collective brain, with the people in the group playing the role of neurons. A greater intelligence then somehow emerges from the connections between these human neurons. While this metaphor is stimulating, it has many problems. The brain’s origin and hardware are completely different from those of the internet, and there’s no compelling reason to suppose the brain is an accurate model of how collective intelligence works, or of how it can best be amplified. Whatever our collective brain is doing, it seems likely to work according to very different principles than the brain inside our heads. Furthermore, we don’t yet have a good understanding of how the human brain works, so the metaphor is in any case of limited use at best. If we’re going to understand how to amplify collective intelligence, we need to look beyond the metaphor of the collective brain.
Many books and magazine articles have been written about collective intelligence. Perhaps the best-known example of this work is James Surowiecki’s 2004 book The Wisdom of Crowds, which explains how large groups of people can sometimes perform surprisingly well at problem solving. Surowiecki opens his book with a striking story about the scientist Francis Galton. In 1906, Galton was attending an English country fair, and among the fair’s attractions was a weight-judging contest, where people competed to guess the weight of an ox. Galton expected that most of the competitors would be far off in their estimates, and was surprised to learn that the average of all the competitors’ guesses (1,197 pounds) was just one pound short of the correct weight of 1,198 pounds. In other words, collectively, if one averaged the guesses, the crowd at the fair guessed the weight almost perfectly. Surowiecki’s book goes on to discuss many other ways we can combine our collective wisdom to surprisingly good effect.
This book goes beyond The Wisdom of Crowds and similar works in two ways. First, our goal is to understand how online tools can actively amplify collective intelligence. That is, we’re not just interested in collective intelligence, per se, but in how to design tools that dramatically increase collective intelligence. Second, we’re not just discussing everyday problems like estimating the weight of an ox. Instead, our focus is on problems at the limit of human problem-solving