Links for Winter 2024
🔭 Large teams develop and small teams disrupt science and technology : The authors use an enormous dataset of academic articles to discover that team size changes the probability that a paper will disrupt citation patterns in future articles.
Solo authors are just as likely to produce high-impact papers as teams with five members, but solo authors are 72% more likely to be highly disruptive. […] We find that solo authors and small teams much more often build on older, less popular work. Larger teams more often target recent, high-impact work as their primary source of inspiration, and this tendency increases monotonically with team size.
They speculate that this could in part be a function of the greater need to insure a steady funding stream for larger teams.
🪬 Ian Leslie's review of "Rational Ritual" : Social coordination depends on having an accurate model of what other people think. Since the legitimacy of our institutions requires people to believe in them and believe that others believe in them, rituals allow beliefs to be publicized and our models of other people’s minds to be updated. Interesting discussion of how targeted advertising and algorithmic content presentation disrupts this process. This is the first explanation of the role of ritual that’s made sense to me.
📝 X thread on 2023 PISA results from Crémieux. Be sure to scroll down to see the corrected science and aggregate scores. This has recalibrated my understanding of the quality of secondary education in the US. Related thoughts on whether the alphabetic Korean writing system boosts Korean writing scores versus its East Asian peers. I’m a bit skeptical because most (all?) languages in the Indian subcontinent are alphabetic too and they have a high math:verbal ratio in their scores. This could be an effect of the diversity of languages in India? Regardless, I’m a sucker for Hangul fanboying. Also, amidst general dismay about downward trends in PISA scores and even concern whether they’re just noise, Crémieux came back with a demonstration of how the rank order is preserved across years.
🤔 Possible effect of pro-Palestine marches on 2023 Dutch Prime Minister election polls This has been under-considered. In early October, Hamas attacks Israel from Gaza. In early November, large-scale pro-Palestine protests began around the world. In the Netherlands, this coincided with an enormous swing towards the anti-immigrant Partij voor de Vrijheid (PVV), support for which doubled between the initial attack and the election won by Geert Wilders. I suspect there’s a similar effect in the US, but it’s harder to detect here because the party platforms are less granular.
🙀 I’ve been having a lot of fun with Instagram’s background generator, which works by merging selected objects in your photos with backgrounds created by your text prompts. The results are occasionally brilliant and occasionally kind of horrifying. In the first image, my cat Nina is performing at the Super Bowl halftime show (and looking absolutely thrilled to be there). In the second, my other cat Pierre is supposed to be lying on a beach under an umbrella, but has actually been transformed into a piece of driftwood, with his tail forming part of the umbrella and his ear part of a lounge chair. I suspect that his upside-down pose confused the algorithm a bit.
📖 Saved by Infinite Jest I first read Infinite Jest a couple years out of high school, and loved the world-building and style of it. But when I read it again a decade later, I realized I hadn’t understood anything at all about it. I’d totally glossed over its rawness and vulnerability. I suspect when people criticize it, they’re reading it the way I initially did: as just a stylistic exercise about characters that wouldn’t be out of place in a Wes Anderson film. It seems like there’s a slow reevaluation of it, and if I had to guess, it’ll be a novel that people still talk about in a hundred years.
🎭 Toward a shallower future This piece by Noah Smith starts with an analysis of a Twitter troll, and ends up being incredibly moving. A reminder that all the good art in the world isn’t worth the adversity that brought it into being:
Some romanticists feel the urge to knock over the edifice of industrial society intentionally, in order to kick against the seeming shallowness of modern life — to return humanity to a world of toil and struggle, in order to ennoble us. But these dark romantics are rightfully recognized in fiction and public discourse as villains. The heroes of our stories are the people like [developer of modern anti-HIV drug regimens] David Ho — the ones who fought to hoist humanity up from the muck so that future generations could be a little more childlike, the ones who studied politics and war so that our grandchildren may study statuary, tapestry, and porcelain.
I have a pet theory that a kind of behavioral neoteny is necessary for civilization and that the degree of neoteny and the degree of civilization go hand in hand. This is very much in the same vein.
✍️ Literary theory for LLMs David Gunkel talks about how LLMs interact with the idea of “the death of the author.” This idea doesn’t say that authors are irrelevant, only that their intentions aren’t authoritative once readers get involved and start creating their own meaning. It’s easy to deny this when most texts still have an author’s name on them, but LLMs bring it to a kind of crisis. Texts created by LLMs have no authorship to which we can assign an intention (“something to say”) outside of the text itself. This isn’t a criticism of LLMs so much as it is a way of reflecting on our preconceptions of how meaning functions in writing.