[Note: this post was published prematurely, then deleted, and is now back.]

Michael Nielsen gave a great talk at TEDxWaterloo about the idea of “open science”:

There’s a great deal of buzz about “openness” in certain sectors of the science community; largely this has passed physics and astronomy by, because we’re already pretty darn open. It’s hard to image something more open than arxiv, where everyone puts their papers up for free even before they’re published in a journal.

But Michael’s talking about something much more ambitious: opening the process of creating science, not just publishing it. For experimentalists this would be difficult, for obvious reasons. (You think people who sweat to build an experiment are going to invite the public in to take a whirl?) For theory it is also hard, but the reasons are more subtle.

The point is that credit in science is given out on the basis of getting your name on published papers. In the arxiv era, the papers don’t necessarily have to appear in a traditional journal — but that’s a topic for another day. The model is set in stone: you have an idea, you work out its consequences to the point where it’s publishable, and you write a paper. Without that last step, you’re not going to get any credit. (Very occasionally you will see references to “unpublished work” or “private communication,” but it’s rare and not really for big-ticket ideas.)

So if I had an idea, I would either work it out myself or start working with students or collaborators. I certainly would not go around publicizing an undeveloped idea; I wouldn’t get any credit for it, and someone else could take it and develop it themselves. I might give seminars in which I mention the idea, but that’s only recommended once it’s to the point where a paper is on the horizon.

Michael and others want to overthrow that model. Their shining example is this blog post by Tim Gowers. Gowers is a mathematician who proposed attacking an open math problem right there on his blog, by asking for comments from the crowd. If they succeeded, they could publish a paper under a collective pseudonym. He next chose a problem — developing a combinatorial approach to the Hales-Jewett theorem — and, several hundred comments later, announced that they had succeeded. Here’s the paper. Buoyed by this success, people have set up a Polymath Wiki to expedite tackling other problems in this way.

Could this work for theoretical physics? I don’t see why not. But note that Michael spends a lot of his time in the talk pointing out the obvious — crowdsourcing doesn’t always work. I could easily imagine ways in which such a project could fail; too much noise and not enough signal, everyone with good ideas deciding they would rather work on them by themselves rather than sharing openly, etc.

Might be worth a shot, though. I’m thinking of suggesting some ideas here on this blog and seeing whether we get any useful input. Let me sleep on it.