Also this week: GPT-3 vs. pickup lines, strong opinions vs. weak effects, effective
altruism spite, and more.
Dan Bolnick on how his department rethought their graduate admissions essay. I agree with Dan that this seems like a good change on balance. Whether it will actually do much to increase the diversity of applicants on various dimensions remains to be seen. For thoughts from Meghan, Brian, and I on graduate admissions and how they might be reformed, see here and here.
New preprint from Aaron Clauset and colleagues on the socioeconomic backgrounds of 7000 US faculty from 8 disciplines. The unsurprising headline result: US profs tend to come from well-off, home-owning families, and are much more likely than other people to have a parent with a PhD. Not yet peer-reviewed, and I haven’t read it yet, but Clauset has done good work in the past on related topics. So I feel comfortable passing this on in case others want to read and evaluate it for themselves. One thing I’ll be curious about when I get around to reading it myself is whether there’s variation among disciplines. There’s substantial variation among disciplines in some other aspects of faculty hiring (e.g., where you got your PhD matters a lot in some disciplines, but very little in ecology).
How Europe and the US lost to Covid-19. Thank god we got like a 99th percentile outcome on vaccines.
Your strong opinions about what makes for a good paper title are out of all proportion to the tiny effects of paper titles on citation rates or retweets or whatever. I would broaden this point: there are lots of things that some scientists (including me!) have strong opinions on that just don’t matter much. For instance, if you’re an ecology faculty job applicant in the US or Canada, the nit-picky details of how you format your cv don’t really matter, even though I have strong opinions on the “right” format. Where you got your PhD doesn’t really matter, even though some people think it does. Whether you’re interviewed first or last or somewhere in the middle doesn’t really matter, even though some people think it does. Etc.
Gut microbiome startup uBiome charged with securities fraud. I confess: I linked to this mainly so that I could claim that the biggest conceptual advance in recent ecological history is also securities fraud. 😛
GPT-3 generates pickup lines. 🙂 It turns out that, just as AI is better than people at tasks people are good at, it’s worse than people at tasks people are terrible at. “I have exactly 4 stickers, I need you to be the 5th.” “Can I see your parts list?” “I will briefly summarize the plot of Back to the Future II for you.”
“But what do we do then, evil twin of Peter Singer?” 🙂 That sounds suspiciously like an AI-generated pickup line.
Dan Bolnick’s post on grad admission essays is interesting. It strikes me that the example list of questions given there would, with some modifications, also make a good tool for interviews of potential postdocs or faculty. In those interviews I often feel like I’m groping to find a good set of questions that will allow accurate comparison of candidates.
Yes, I had a similar thought–it makes graduate admissions a bit more like faculty interviews.