Thoughts in Between
TiB 173: A brief history of the future; Chinese AI; talent and invention; 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.
Does geography determine the future?
I mentioned a few weeks ago Astral Codex Ten’s book review competition, which has some outstanding entrants. One of the winners is this superb essay by Misha Saul, who compares and contrasts Peter Zeihan’s Disunited Nations and Bruno Maçães’ The Dawn of Eurasia (We’ve discussed Maçães’ work several times before). Saul focuses on the authors’ competing theories of the future of geopolitics - Zeihan’s geographical determinism vs Maçães’ manipulable tapestry. It’s a long but highly recommended read, particularly if you’re new to the two authors’ work.
The question of how much agency states and individuals have in shaping the future has been a frequent theme in Thoughts in Between. See, for example, our discussions of Peter Thiel’s “definite optimism” (also cited by Saul); Joe Henrich and Walter Scheidel on the very long run impact of institutions; and last week’s piece, featuring essays by Adam Tooze and Tanner Greer, on “long wars”.
My own position is that Marx was (uncharacteristically) right when he said, roughly, that people make history, but not in circumstances of their own choosing. I believe that we are both moulded and constrained by the institutions we inherit - but that technological change enables institutional change, and so determined actors do have levers to shape the very long run. It’s one of the reasons that I see intentional institution building as one of the most important tasks of our era (and why I’d like to see many more of our most talented people attempt it).
Companies don't innovate, people do
I’ve been thinking a lot about talent clustering and particularly how it helps and hinders startups’ ability to compete with well funded incumbents. I was therefore interested to come across this paper from last year. It looks at whether “inventiveness” is primarily a characteristic of firms (i.e. certain organisations have embedded innovation capabilities that make their employees more inventive) or of individuals (i.e. certain individuals are particularly inventive, irrespective of where they work)
The authors use a large patent dataset and a clever statistical strategy to disaggregate these two effects and come down firmly in favour of the latter explanation: inventor quality explains more than five times the variance in output between firms than differences in organisational capability does*. Moreover, the authors are able to use the same dataset to ask what predicts the movement of high quality inventors between firms. They find that the biggest determinant is the quality of the other inventors there, rather than firm-level capabilities.
This is good news for research-heavy startups. It suggests that if a new company can attract a critical mass of star researchers, its relative lack of organisational capital is no barrier to success. This seems particularly applicable in AI research. It’s one reason why new entrants like Anthropic (see TiB 167) have a good shot, despite the apparently formidable leads of DeepMind and OpenAI. Talent wins.
*There’s the obvious caveat that patenting might not capture all of what we care about when thinking about innovation, but this is still striking
Podcast: Jeffrey Ding on AI in China
This week's Thoughts in Between podcast episode is a conversation with Jeffrey Ding. I consider the development of artificial intelligence in China one of the most important topics in the world today - and if you're serious about understanding it, Jeff's work is indispensable. He works at Oxford University's Centre for the Governance of AI and is the author of the newsletter ChinAI, probably the best English language resource on Chinese AI. We've discussed his work in the newsletter many times.
One of the most interesting points Jeff makes in the conversation is about the relative levels of civil-military integration in China and the US. We've talked before about Silicon Valley's social liberalism as a potential strategic weakness for the US (see TiB 80). But, around ten minutes in, Jeff argues that you could make the opposite point: there's no Chinese equivalent of Palantir or In-Q-Tel - and the Chinese government on balance probably envies the US military-industrial complex more than the other way around.
We cover a lot of ground in our conversation, including:
- Where the gaps remain between in the US and China's AI capabilities
- Why it's important to avoid thinking of "China" as a monolithic entity when it comes to AI
- The limitations of "arms race" framings for thinking about national AI strategies
- Why there's no Chinese DeepMind or OpenAI (yet)
- AI safety in China
- ... and much more
I learned a huge amount from Jeff. I hope you enjoy the conversation.
- Too soon? Great (and sadly prescient) summary of what the data tells us about penalty shootouts.
- Life, but not as we know it. Extraordinary and/or creepy demo of using AI to design and create new biological organisms.
- Life imitates (the suppression of) art. Not a quick read, but a fascinating write up of a German military research programme to use literature to predict geopolitics.
- Building back better? Americans' self-reported well being is at all time highs. And debt service is at record lows.
- The attention budget. Intriguing study on how sports referees make decisions, which suggests attention is a resource that gets depleted (though it's hard to read stuff like this these days and not think, "replication crisis!")
Thanks for reading
I'm grateful to have such a great readership for Thoughts in Between. I'd be even more grateful if you'd forward it to a friend or share the link on social media.
And do feel free to get in touch by hitting reply.
Until next week,
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.