2026-03-15
fear evil like fire
Approved podcast on the history of Geron Corporation (included in Evan DeTurk biotech linkthread), which is quite interesting as a description of how little has actually changed as far as life extension goes in the past thirty years. Also, I wonder what the relationship is between the idealistic desire to live forever and the mercenary desire to control information and monopolize the gains. Perhaps related to an intuitive understanding that once senescence is solved, the limiting factor of one’s lifespan is likely to be a result of Malthusian dynamics.
Natasha Bita with an interesting story on a custom mRNA cancer vaccine, which is currently being described as all of unlikely, trivial, or plausible. (Edit: Ruxandra on how the current system of clinical trials is particularly unsuited to small-n experiments, which is particularly important given the trend towards personalized medicine involving increasingly inexpensive technologies like genetic sequencing and AlphaFold for personalized treatments like CAR-T, mRNA, or gene-editing. On the other hand, if the problem right now is that this is something which is only available to billionaires, then that’s actually a promising sign that the market alone should be able to figure out a viable path forward. It’s possible that even after the FDA finally figures out a workable platform approval process, the de-facto standard of care for “incurable” cancer is nevertheless a process where the first step is always a flight to a charter city).
Sabrdance has an interesting piece on LLMs which seems to be stating something like that LLM’s don’t actually produce “true knowledge”. I don’t quite understand this point, because it seems to me that it’s impossible to obtain full knowledge about anything, since that would require also knowing everything about its interactions, recursing endlessly into infinity. In which case, all knowledge must be just a compression of full understanding, making it an empirical question as to whether any of tacit, book, or LLM-recall are more effective ways of knowing within any particular domain. Somewhat related, Defender has an interesting description of his algorithm for eliciting understanding from would-be-persuaders in adversarial information environments. I have an alternative algorithm which is less comprehensive but converges faster, which is to ask the question: “if I adopt your worldview, then what will I be able to achieve with it?” There are a notable lack of useful discoveries which were developed using the flat-earth view of the world.
Both Sides Brigade has a piece on evolutionary debunking arguments against moral realism which I also don’t understand; it seems to be saying something like “natural selection is sufficient to produce humans, but that doesn’t mean that God wasn’t involved in the process”, which of course is accurate, but violates Occam’s razor. It seems quite clear to me that morality is a product of natural and cultural selection as a coordination mechanism for adaptive reproductive fitness, which is why people and different cultures disagree on moral matters1, which is the underlying justification for the statement that “natural selection is not a truth-tracking process”. Within this explanation, morality is fully explained by evolution, in which case there doesn’t seem to be any justification for getting this surplus element of moral truth involved here.
Tangentially related, Jack Thompson on jealously, which functions as a pretty good example of how closely tied claims about morality are to individual preferences.