2026-05-11
dark eyes (alt. tk.)
Alex Kesin has an interesting piece comparing the business model of the pharmaceutical industry with that of Hollywood, possibly related to the recent post by Owl Posting on financial engineering in biotech, which has a (paywalled) followup specifically on the Chinese drug development financial architecture. If the analogy to big box office productions is that sequels are what allows the studios to keep running and fund novel IP, then the Chinese production of fast-followers on known endpoints is effectively a means of taking away the most profitable component of the market, that is, the sequel machine. Moreover, even though Australia can provide a cheaper and faster location to perform phase I clinical trials, the most expensive phase II and III trials are still bottlenecked in the United States, which is presumably one reason why there are so many abandoned assets for companies like Roivant to pick up, despite the adverse selection problems which would normally make such attempts unadvisable. On that note, relevant Infinite Loops interview with Saloni Dattani on the need for clinical trials reform, and Adam Kroetsch on the real-time clinical trials pilot.
Afra Wang impressions of a recent trip (also described by Nathan Lambert) to China visiting their AI and robotics labs, noting the distinction between the attitudes of AI engineers between the two countries, where Chinese AI is produced for and within their existing system, while American AI is both the favored child and the heir apparent. Somewhat related, Buck Shlegeris has a piece considering the potential benefits of becoming an AI insider1. It’s notable that there do not seem to be any major benefits to being an AI insider in China at the moment, as seen by the relative openness of their labs to external observers.
John Psmith review of 50 Years of Text Games, as a discussion on the limitations of designed interactive media like video games and choose your own adventure stories. The fundamental limitation of responsiveness within these genres has always been one of the primary creative constraint to fully implementing the creative vision of game designers, creating two divergent artistic traditions: either more and more advanced cellular automata in the vein of Dwarf Fortress; or metagames like The Stanley Parable, played with the self-awareness that it’s actually the game designer who’s playing the player, not the player playing the game. As a response to whether LLMs have allowed us to finally overcome this fundamental limitation, obviously the answer is no, because any world which is created and understood through the human corpus is still just a map, and not an actual representation of true physical reality. And yet this equally applies to other games like financial markets, politics, and relationships, built under specific limited rules and constraints, which show that a game doesn’t need to be larger than life to be larger than the life of any individual or even group of people that it might subsume.
Kevin Kodama review of The Wandering Inn, as a representative of the excessive larger-than-life extended universe story which was enabled by the webnovel format.
Father Karine story adjacent to escapism.
Julia Wise review of the Girl Scout Handbook 1956.
Besides what he mentions about potential time-sensitivity of information, it seems to me that there’s also a structural aspect, in that in order for the environment where outsiders are able to get information to exist, there necessarily must be insiders which are disseminating it; also, while outsiders are able to obtain more or less all the information of insiders, the distinction as to how real any piece of information is something which benefits greatly from the verification of direct experience.