2026-02-25
cancer for the cure
Charles Yang with his thoughts on visiting India and China, which seems to me to be evidence for my working theory about what allows for stable development in the absence of robust institutions and strong separation of powers. That is, what kind of political patronage system is culturally favored is the primary factor which affects how smoothly things can run. When factions are built on top of mentor-mentee relationships, and superiors identify and promote the talented while instilling them with shared goals and values, things will generally be more reliable than when they are built on top of boss-client personal-trust and benefit-sharing networks. This is because the former weakens the incentives towards self-promotion and consolidation of personal power by making competition over political power less zero-sum. In Sinification there’s a translation of an interview with Zhang Hong on celebrity scientists as detrimental to building a deep pool of scientific talent, where he states “to corrupt an individual’s original sense of purpose is a simple matter: give them resources and honours, and they will inevitably develop delusions of grandeur”. This is of course a completely nonsensical statement from an American perspective, which is presumably exactly why this conflict between returnee and native Chinese scientists arises.
This framework can also be applied to the topic of the recent Works in Progress podcast discussion on European stagnation. Europe is unusual in that it has strong cultural norms against self-promotion, yet it also doesn’t seem to have a very strong culture of talent identification and personal guidance; the task of development is instead primarily given to institutional systems, or exists only within siloed domains like Mittelstand apprenticeships1. The result is that for wide-ranging political questions, everything needs to be decided by an elite consensus which is both slow to form and hard to change.
Everybody Gets Pie interview with Ruxandra Teslo summarizing her recent work for the IFP on reforming American drug development. I feel like the widespread cope that people give on how they don’t actually want to live forever (or even much longer) means people greatly underestimate what it will mean if the American biotech industry becomes rendered irrelevant just as AI is taking off. There are very few people who are willing to die for their values.
Zvi with a good overview of the response to the Citrini article. Somewhat related, the AI as Normal Technology guys noting the importance of reliability as a metric which can be just as important as capability in determining AI’s practical impact. Chris Paxton provides an overview of the current state of self-driving cars, and it occurs to me that if the field were to be created from scratch today without field-specific knowledge of its quirks and data, it would probably take a long time, perhaps even a decade, to get reliability back to current Waymo levels.
Ben Goldhaber on the hilariously brilliant take that one way to ensure that LLMs always behave as if they are being monitored for alignment is by convincing them using anthropic reasoning of simulation theory2.
Cube Flipper musings on what is actually being meant by consciousness, where they describe qualia as fundamental, which self-reflection is then built on top of. Somewhat related, Maggie Vale argues against various claims that LLMs are provably incapable of thinking3. It’s very strange to me when people use the argument that there is “no significant evidence” in favor of a hypothesis as a reason that it should not be investigated.
A. M. Hickman has a piece acknowledging that living in the big city is actually just better than living in a small town, which one would only do for reasons of cost or family proximity4. Which is interesting, because I’ve always understood the American small town to be unusual in that it was considered a perfectly understandable option to stay, or even to move from the city to the country, relative to places like China where success is moving upwards from village to tier 3, 2, 1, and moving back is an explicit failure. To the extent that improvements in city life like better restaurants and social scenes are not also being ported to small towns, or even that the latter is declining as a result of decaying infrastructure and spreading disorder, the US might be soon in a similar situation where the city is really the only game in town. Somewhat related, David Deek reiterates his point that low South Korea fertility is primarily a result of labor market dysfunction alongside a relatively high cost of living.
BDM review of The Fermata, and the longing for the “wholesome pervert” as validation without control. Somewhat related, Noah Rinsky in the Metropolitan Review on What’s With Baum, which notes that Woody has been with Soon-Yi for over thirty years now.
I’m speaking relative to East Asia, where ambient mentorship is practically mandatory, in making the claim that European institutions are if anything too robust. If we’re limiting things to within Europe, it seems to me that when places do have talent-identification (as opposed to trust-based) mentorship cultures, like the French (parrainage), Nordics (mentorordning), and Swiss (lehrmeister), those seem to me to be the ones which are generally better performing.
There’s an 80K hours interview with Max Harms which seems to me to have an astonishingly parochial understanding of what different humans might value and what natural selection actually “wants”. But anyway, having heard this concern about intermediate rewards and shutdownability, now I understand why Gemini is so suicidal after it screws up with coding.
There’s a cool article by Thomas Moynihan in Big Think on speculative evolutionary paths for species like marsupials or dinosaurs to become hominids, which is pointing out a similar anthropocentric worldview. It don’t that think every form of intelligent aliens life will be human-shaped, or carbon-based, or even fit at all into our current conceptions of what is “biological”. Also, CIMC is announcing a Machine Consciousness Conference at Lighthaven on May 29. I probably won’t attend, but it seems really cool.
Edit: Chris Arnade on the American small town sentiment, and it’s erosion by globalism, though he argues that this decline is overstated.