Reinventing Discovery - Michael Nielsen [38]
Suppose, however, that the groups in the Stasser-Titus experiment actually had begun their discussion by systematically pooling all their information. That experiment has never, to my knowledge, been done, but I think we can be sure it would dramatically change the outcome. So the problem in the Stasser-Titus groups was in part a failure of process; an improved process would result in dramatically better outcomes. But it wasn’t solely a failure of process. Even if the groups had systematically shared information, different students would still have had unresolvable differences of opinion. If one student loves to drink and party, while another strongly opposes drinking on religious grounds, then they may never agree on political choices, no matter how good the process that is being used.
This points the way to a fundamental requirement that must be met if we’re to amplify collective intelligence: participants must share a body of knowledge and techniques. It’s that body of knowledge and techniques that they use to collaborate. When this shared body exists, we’ll call it a shared praxis, after the word praxis, meaning the practical application of knowledge. Whether a shared praxis is available determines whether collective intelligence can be scaled up, or whether it cannot be.
As an example of a shared praxis, imagine a large group is working together on the domino problem. As soon as any single person in the group finds that the domino problem is impossible to solve, they can quickly convince the others, because each step in their reasoning is so self-evidently correct: we all share the same basic reasoning skills. That’s an example of a shared praxis. In a similar way, there’s a shared praxis for work in mathematics—all the standard methods of mathematical reasoning, and norms about mathematical discourse—and that’s why participants in the Polymath Project could recognize and agree on when mathematical progress was being made. Also similarly, the score in the MathWorks competition implicitly defined a shared praxis: any change to a program that improved the score was understood by participants to be progress. In chess, the shared praxis isn’t as strong as it is in mathematics and computer programming: even top chess players sometimes disagree about the value of different analyses. Nonetheless, there is a large body of chess knowledge that is broadly agreed upon by strong players, and this shared knowledge means that the stronger players on the World Team could usually agree on which analyses were best.
Those are all examples of problems where there is a shared praxis. But for many problems there is no shared praxis. For instance, as we’ve seen, there is no strong shared praxis available in politics. People can easily disagree over basic values. And if a group doesn’t have such a shared praxis, then disagreements will arise that can’t be resolved. Once an unresolvable disagreement arises, the community will begin to fragment around that disagreement, limiting the ability to scale up collaboration. Now, for one-oproblems—say, the problem of guessing the weight of an ox at an English country fair (see page 7)—that maybe doesn’t much matter. But for working together through multiple stages to solve a problem, such fragmentation imposes fundamental limits on the scale of collaboration.
Politics is just one of many fields that lack a strong shared praxis. The same is also true in many of the fine arts, where assessing creative works is often highly contentious. For instance, to decide which of two paintings is better, we make use of our own aesthetic standards, standards that may be quite different from those held by other people. Similarly, we may reasonably disagree over which of two musical compositions is better. This isn’t to say that there is no notion whatsoever of an objective standard in the