I attended the Science Commons Salon last night and spent some time talking to intargoogles celebrity Joi Ito. He was very interested by my comments about the lack of a "Journal of Failed Experiments" and the question about why such a journal doesn't exist (even though most people agree it would be useful) and what cultural, institutional, and scientific barriers prevent it. The topic also touched on some questions about what Science(tm) could learn from Silicon Valley. Joi seemed to believe, as an article of faith, that making science more responsive to outside interests (money, medicine, engineering, et cetera) would be a good thing, and I was not entirely convinced of that, being a bit of a purist. I often like to point out that J.J. Thomson did not have cell phones and computers in mind when he discovered the electron. But I actually didn't broach this topic with Joi, preferring to try to glean from him what I could, rather than argue about Science(tm) versus "science".
One interesting idea I came away with was the thought that, if you look at most of the fast growing high tech firms, they gave up long ago on having extended 9AM meetings in which people put up PowerPoint slides and discuss their latest progress. This management model would be considered downright ossified in places that practice methods like "agile programming", and "extreme programming". (In the latter, for instance, 15 minute "stand up" meetings (nobody is allowed to sit) are the norm, to keep the meetings on track and short, and to keep people on the important stuff.) But, strangely, our project management style is exactly this, interminable 9AM meetings filled with PowerPoint slides and people falling asleep (literally, in some cases.) I started to think about some alternative ways to structure research, at a very basic level, and was thinking about an extension of the "pair programming" methodology, of which Joi seemed to be a big fan.
Imagine, if you will, the following: a research group is divided into two groups, 1 and 2. Alice, Barbara, and Cathy are in group 1, and Arthur, Bob, and Carl are in group 2. Each person is the lead on one project, so Alice has her own project, Bob has his own, et cetera. During the first week, one person from team 1 and one person from team 2 get together to work on the team 1 projects. So, for instance, Alice and Arthur work on Alice's project, Barbara and Bob work on Barbara's, et cetera. During week two, they switch, and work on the team 2 project with the same partner. In week three, now, they rotate partners, but go back to the team one project. So, now Alice works with Bob, Barbara works with Carl, and Cathy works with Arthur. Week three is for team 1 projects, and week 4 for team 2 projects. And so on, and so forth.
What would be the practical consequences of this arrangement? Well, for one thing, everybody is invested in everybody else's project, and everybody gets their name on every paper. One of the impediments to collaboration in the lab is the question of lead authorship: if I give over too much control to somebody else, I might lose my slot as lead author! But this system preserves lead authorship, which, although artificial, is extremely important for job searching and funding. The principle on each project is still the lead author. It gives everybody in the lab a broad range of experience working on a lot of projects, and it allows people with fresh eyes to contribute to projects, by bringing new perspectives, which is often what's needed. And, by maintaining the rotation, people stay up to speed on their peers' projects, and don't need to be retrained too often (hopefully.) I'm curious if this would be a workable lab environment. I can see pitfalls (what if Alice and Bob hate each other?). And, also, this organization isn't idea, because Alice never gets to work with Barbara. But, there's probably some simple permutation that could get us around that problem. In any case, I think it would be a bold experiment. If I ever have my own lab, it would be something I genuinely think I'd like to try.