TiB 220: AI-imagined histories; the geopolitical consequences of age; institutional rot in science; and more...

Welcome new readers! Thoughts in Between is a newsletter (mainly) about how technology is changing politics, culture and society and how we might do it better.

It goes out every Tuesday to thousands of entrepreneurs, investors, policy makers, politicians and others. It’s free.

Forwarded this email? Subscribe here. Enjoy this email? Forward it to a friend.

Public investment in science: great value for money

If you're interested in the (enormous) benefits of public science funding and the challenge of science stagnation, this Twitter conversation between Alexander Berger (Co-CEO of Open Philanthopy) and Pierre Azoulay (Professor at MIT's Sloan School) is well worth a read. Berger tweets this 2018 paper by Azoulay and his co-authors, which tries to quantify the impact of grant funding from the US National Institutes of Health (NIH) on private sector patent activity. Using a novel dataset and methodology they find that on average every $1.6m of NIH funding is associated with one additional biopharmaceutical patent.

It's hard to put a value on this, but the authors estimate that the private and social returns are worth many times the size of the grant - not least because grants lead to about as many patents in areas outside the stated focus areas of the grant as in them! Do read the full paper for more, but Azoulay's tweet is a good summary: "our numbers imply that NIH research offers a phenomenal return, and that's before we talk about social returns". So far, so good... but Azoulay adds: "My conjecture: when we repeat the analysis 20 years from now, the results will be much less impressive". Why? "Institutional rot". He links to these slides from a talk he gave last month.

The story will be a familiar one to long-time readers: incentives that encourage incrementalism, risk aversion and conformity in scientific careers. Azoulay links to this piece, which I'd not seen before, that tells the story of the last 30 years of Alzheimer's research and argues that the dominance of a single thesis (the amyloid hypothesis) and funding bodies' failure to back alternatives has slowed progress towards a cure. Azoulay is a fan of randomised controlled trials for science funding itself (Michael Nielsen and Kanjun Qiu have a must-read essay on this, which we discussed in TiB 202). I doubt there are any silver bullets, but it seems clear that far more experimentation in funding approaches is essential.

The geopolitical consequences of generational change

Tanner Greer has an interesting short post (which I'm late to) on generational churn in the Chinese Communist Party. Greer points out that the current leadership of the CCP has an extremely narrow spread of age relative to the West (see Figure 1 in this Macro Polo piece). The entire leadership is effectively from a single generation born in the 1950s and early 1960s. This matters for a number of reasons, but above all, Greer argues, because generational turnover is a major driver of historical change (see also his essay Culture Wars are Long Wars, which we discussed in TiB 172).

It's particularly important in the context of China because generation is a predictor of pro-market and liberal attitudes (see Figure 2 in the same Macro Polo piece, which is really worth a read). Greer suggests that this may open up a window for a more accommodative China-US relationship in the 2030s. If that's right (and I'm not fully convinced of the magnitude of the effect), it makes the case for doing whatever it takes to delay any escalation of conflict over Taiwan (see TiB 136 and 161) even stronger (I thought this was an intriguing and provocative piece on that topic)

It's also interesting to think about generational turnover in the context of emerging technologies that might delay or attenuate its impact. We discussed before (see TiB 93 and 104) the idea that life extension might be a pro-authoritarian technology, but that was largely from the perspective of extending the rule of a single totalitarian dictator. But there's a softer, and perhaps more realistic, version: when people can (as the US demonstrates!) hold powerful political positions well into their 80s, the sort of generational turnover that Greer argues undermined the Soviet Union takes longer to come about - and is perhaps more destabilising when it does.

AI-imagined histories as a tool for thought

I love it when two apparently disparate interests collide, so I was excited to come across this podcast episode with Sam Arbesman (who was on the TiB podcast back in February on GPT-3 and alternative histories. We've obviously talked about GPT-3 and other AI language models a lot here, but long-time readers may remember that I'm also a fan of "counterfactual history" - i.e. asking what might have happened had certain historical contingencies played out differently (see TiB 132 and 157).

Sam asks, can GPT-3 “imagine” compelling alternative histories for us? And the answer appears to be yes. In this post, he gives some examples. He asks it what might have happened had Jimmy Carter won re-election in 1980 and then (more exotically and inspired by Harry Turtledove) what might have come about had… aliens invaded the Confederacy in 1862. The results are a bit patchy, but certainly interesting and, as Sam says, feel like they "rhyme" with reality.

It turns out Sam has written on this theme before, so if you're interested in counterfactual history, I recommend his piece on "Alternate Histories and Societal Complexities" and especially this piece on maps from alternative histories. The second piece has enough excellent links to take you down the counterfactual history rabbit hole for at least an afternoon. But, even if imagined histories are not your thing, as Sam says, they do point to a way that AI will change the human thought process: as a sort of "imagination assistant" (like Elicit) that can generate, very cheaply, large numbers of plausible ideas and connections for humans to play and engage with.

Quick links

Nobel lords. Wonderful short piece interviewing Nobel Laureates on what they’ve never understood about their own fields, among other topics.
Kids as a luxury good? The relationship between fertility and income. (What other “goods” have this pattern?)
Natural borders and geopolitics. Speculative but interesting on Russia’s motivations.
“Is Reduced Visual Processing the Price of Language?” Fascinating paper on the emergence of language in humans. Semi-related: ancient art and volcanoes.
Tourism and ripoffs. I’m on holiday this week, so this is top of mind…

Thank you, please come again

Thanks for reading Thoughts in Between. As a strictly non-commercial endeavour, TiB survives owing to the vanity of seeing the subscriber count creep up, so shares and forwards are always appreciated.

I'm always happy to hear from readers. If you have comments, questions or recommendations, just hit reply.

Until next week,

Matt Clifford

PS: Lots of newsletters get stuck in Gmail’s Promotions tab. If you find it in there, please help train the algorithm by dragging it to Primary. It makes a big difference.