Saturday, April 21, 2012
On a somewhat lighter note, the recent premier of 'Girls' on HBO has come under a fair amount of criticism for the lack of diversity in its main cast. I watched the premier, and thought it was pretty funny, but the criticism did get me thinking about the diversity of my own circle of friends. If I randomly chose 3 of my friends to start a sitcom, how diverse would they be? Since I'm in the middle of planning a wedding, I've been put in the awkward position of coming up with an "A" list of friends, so why not use this data to figure out how diverse my friends are? Of the 49 people on the list (I threw the bride to be in there as well, just for good measure), I scored each with a 1 or 0, based on my best guess on whether they self identified as white. Turns out 27% of the guests on my list are non-white. If I'm randomly choosing 3 friends for my sitcom (with replacement, and ignoring sex), this should follow a hypergeometric distribution. Randomly choosing from my guest list, I'll have an all white cast about 36% of the time.
Andrew Gelman has a couple of posts up on his blog about how to improve the peer review system. I'd lean to towards a system modeled on the arxiv, as Larry Wasserman suggests. We're coming to the point where we can move beyond academic gatekeepers, and more towards a system where everything gets published and evaluated openly. Search algorithms will only get better, and publishing on the web is relatively cheap. Of course this means a lot of crap will get published, but that's not much different than the current system. PLoS One gets a fair amount of criticism for the quality of papers published there, but at least people approach articles published there with a critical eye. I think many academics are all to ready to accept research published in journals like Nature or Science at face value.