TL;DR
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
Astral Codex Ten published Open Thread 453, its weekly reader discussion thread. The post highlights up to $10 million in grants for work on middle powers and transformative AI, an exchange with California House candidate Jamie Joyce, and new effective altruism campus meetups.
The blog Astral Codex Ten published Open Thread 453, its weekly visible open thread where readers can post questions and discuss any topic, following Open Thread 452. Beyond the discussion function itself, the post carried several concrete announcements: Astralis Foundation and Macroscopic Ventures will distribute up to $10 million in grants for work on how countries other than the US and China should respond to transformative AI, and the blog’s author revisited his earlier California primary endorsement after the losing candidate, Jamie Joyce, asked him to reconsider.
The largest concrete item in the thread is the grant program. According to the post, Astralis Foundation and Macroscopic Ventures — which the author described as parts of the blog-affiliated network of organizations, while admitting he “can’t keep track any more” — will distribute up to $10 million for work on “middle powers and transformative AI.” The program covers “research, training, public writing and new organizations focused on what countries beyond the US and China can do to help navigate” the AI transition. Interested applicants can read more and apply through the linked application process.
The thread also revisits a political endorsement from earlier in the year. In his spring California primary ballot recommendations, the author wrote that he was voting for the “boring incumbent” in the 12th District House race because the challenger seemed eccentric and had no chance of winning. That challenger, Jamie Joyce, contacted him after the primary and asked him to change his mind — and because both candidates will appear on the ballot again in the November general election, the question of a renewed endorsement is live. The author declined to relitigate his May comments but said he “may reconsider” if he writes a post on the November election, and at Joyce’s request he linked to her response, which he recommended to interested 12th District voters.
The thread additionally lists updates to the blog’s network of college effective altruism meetup groups. According to the post, Oxford, University College Dublin, and the University of the Philippines now have meetups planned, with details available in the blog’s list. The post also rounds up “comments of the week,” all drawn from the recent “Mysteries of AI Generalization” post, including contributions from Eliezer Yudkowsky, Ezra Newman, Elias Schmied, and Stuart Armstrong — a roundup tradition that also appeared in Open Thread 451.
Why the $10 Million Grant Program Stands Out
The grant announcement is the most newsworthy item because of its scale and its framing. Up to $10 million directed at “middle powers and transformative AI” signals that funders in the rationalist and effective altruism-adjacent ecosystem are broadening their focus beyond the US–China axis that dominates most AI policy discussion. The stated goal — equipping countries outside the two leading AI powers to “navigate” the AI transition — spans research, training, public writing, and the creation of new organizations, suggesting an effort to build institutional capacity rather than fund a single line of academic inquiry.
The Joyce exchange also carries modest political weight. A public endorsement from a widely read blog can influence a small number of voters in a safely Democratic district, and by linking Joyce’s response, the author has given her campaign a visible platform ahead of the November rematch. His stated openness to reconsidering keeps the endorsement question active through the fall.
How the Thread Fits ACX’s Weekly Rhythm
: “Open threads are a recurring feature of Astral Codex Ten: weekly posts where readers can raise any topic, ask questions, and share links. The blog also maintains an unofficial subreddit, Discord, and bulletin board, plus in-person meetups around the world. Most content is free, with some posts reserved for subscribers.
The blog’s spring California primary recommendations post is the origin of the Joyce item. In that post, the author described his choice as a vote for the incumbent specifically because the challenger appeared eccentric and unelectable. The incumbent and Joyce will face each other again in the November general election for the 12th District House seat.
The “comments of the week” selections all come from the blog’s recent post on AI generalization, reflecting an ongoing community debate about how and why frontier AI systems fail to generalize — and what those failures imply for AI safety. The featured commenters include Eliezer Yudkowsky on why recent generalization failures are “bad news,” Ezra Newman on distinctions between pretraining, midtraining, and RLVR, Elias Schmied linking Evan Hubinger’s argument that “Hacker Opus” did not show emergent misalignment, and Stuart Armstrong on how emergent misalignment arises.
“I don’t want to relitigate what I said in May, but I may reconsider if I make a post on the November election.”
— ACX author, Open Thread 453
What the Thread Leaves Open
The grant program’s details as presented in the post are limited to the funding ceiling, the thematic focus, and the application link. It is not yet clear how the up to $10 million will be divided among grantees, what the deadlines or evaluation criteria are, or whether the two named organizations are the sole funders. The author’s own aside that he can no longer keep track of which organizations belong to the network adds ambiguity about the institutional relationships involved.
On the political item, the author made no commitment: he said only that he “may reconsider” if he writes a November election post, and there is no confirmed timeline for such a post. His May assessment of the race remains his stated position unless he revisits it.
Watch For a November Endorsement Post
The clearest upcoming milestone is the potential November general election recommendations post, which would determine whether the author revises or maintains his 12th District endorsement. Joyce’s response, now linked from the thread, will likely be part of that discussion if it happens.
For the grant program, the next development will be the application process itself: prospective applicants can apply through the link in the post, and grant awards or recipient announcements would follow on the funders’ schedule, which has not been specified.
The weekly open thread series continues, with the next numbered installment expected in roughly a week.
Key Questions
What is Open Thread 453?
It is the 453rd weekly visible open thread on the blog Astral Codex Ten, where readers can post about anything, ask questions, and share links. This installment also included announcements about grants, an election endorsement follow-up, and new campus meetups.
Who is eligible for the middle powers and AI grants?
The post does not specify eligibility criteria. It says the up to $10 million will fund “research, training, public writing and new organizations” focused on what countries beyond the US and China can do about transformative AI, and directs interested readers to an application link for details.
Why did the ACX author discuss Jamie Joyce?
He had recommended voting against her in the spring California primary, backing the incumbent in the 12th District House race. Joyce contacted him asking him to reconsider ahead of their November rematch, and he linked to her response while saying he may reconsider if he writes a post on the November election.
Which universities gained new effective altruism meetups?
According to the post, Oxford, University College Dublin, and the University of the Philippines now have meetups planned, with details available in the blog’s meetup list.
What were the featured comments of the week about?
All were drawn from the “Mysteries of AI Generalization” post and addressed AI safety and generalization topics, including why generalization failures matter, distinctions between training methods, and debates over emergent misalignment.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
