Education has an important task to teach where to find reliable information (written by experts, or near-experts). In the context of collective intelligence, I would like to venture the following opinion. It seems to be a commonly held truth, or belief, that since knowledge in various fields has expanded so much, no one person can know everything and therefore we can be good at only one specific thing, expert on that, and rely on other experts or collective intelligence for the rest. I think it is a too defensive ambition. Certain essential things to know are not so many, and therefore not so hard to have an overview of a few things. (This overview must be essential in finding knowledge.) In this regard the American college system, the idea of liberal arts education in the modern sense, is great. The trend world-wide seems to be the opposite, toward more specialized university degrees.

Here is a striking example. Some years ago, one prominent medical school started requiring its students to take a course in art history:

(The link is supposed to be nytimes.com/2002/05/19/nyregion/yale-s-life-or-death-course-in-art-criticism)

This has now spread to most top American medical schools and there have also appeared studies indicating that these art classes make medical doctors better at diagnosing patients.

These efforts have had some successes and perhaps some set-backs and has received scepticism from some other leading mathematicians. One trouble that I think arise is that sometimes the main progress is made by a single person providing the crucial idea which leads to a quantum leap in progress. (Progress is not always continuous, maybe rarely so? comments?) It will then be less natural to incorporate it into the collaboration, also the individual person may want more credit for herself (this can be a matter of getting a good job, or an award giving money — so it is not a superficial concern).

Some polymath projects have been successful, this is not surprising since it involved a large effort, so it does not necessarily mean that polymath projects is a good approach. One should, like an economist, when evaluating its benefits, ask about the alternatives? Maybe other approaches are more efficient, maybe it depends on the problem, can one identify suitable problems in advance?

Are there other examples of such collaborations in other domains?