I recently saw Habryka lamenting[1] that the 2026 International AI Safety Report didn’t cite any LW or Alignment Forum posts, even though it claims to be “comprehensive”. It seems worth spending some time thinking about solutions to this problem.
Like, it’s apparent that what’s going on is the authorities do not consider LW/AF Credible™ or Respectable™. But they will still cite academic papers that are copying years-old ideas from LW. The situation is better than it was a decade ago—at least now people are paying attention to LW’s ideas after they’ve been laundered through academia, rather than not paying attention to them at all—but there’s a lot of room for improvement.
Here’s a random idea I just thought of that probably wouldn’t work: create a pipeline to convert (some subset of) LW posts into PDFs on arXiv, or some other Respectable™ site that doesn’t require peer review. The people behind things like the International AI Safety Report probably feel more comfortable citing non-peer-reviewed arXiv papers than they do citing LW posts. Maybe we could also have some facility to make it easier to add the trappings of an academic paper (like abstracts and signposting sections).
That’s not the best idea, but I imagine there are ways to make LW posts more Respectable™—either by converting them into a more Respectable™ format, or by shifting people’s perception of LW.
Why not just make an academic journal, identify some alignment researchers who frequently use LW and have PhDs, then aggregate their karma over month-long time slices, call that the “review”, and then publish an edition every month? Maybe the authors would need to clean up their work a bit.
This seems easier to implement than (the good version of) my idea.
I wouldn’t want to require PhDs. I think most of the best contributors to LW don’t have PhDs—I can think of at least one who doesn’t even have a high school diploma, although he doesn’t post too often these days. Although having a “LW journal” with a PhD requirement seems better than not having one at all.
I recently saw Habryka lamenting[1] that the 2026 International AI Safety Report didn’t cite any LW or Alignment Forum posts, even though it claims to be “comprehensive”. It seems worth spending some time thinking about solutions to this problem.
Like, it’s apparent that what’s going on is the authorities do not consider LW/AF Credible™ or Respectable™. But they will still cite academic papers that are copying years-old ideas from LW. The situation is better than it was a decade ago—at least now people are paying attention to LW’s ideas after they’ve been laundered through academia, rather than not paying attention to them at all—but there’s a lot of room for improvement.
Here’s a random idea I just thought of that probably wouldn’t work: create a pipeline to convert (some subset of) LW posts into PDFs on arXiv, or some other Respectable™ site that doesn’t require peer review. The people behind things like the International AI Safety Report probably feel more comfortable citing non-peer-reviewed arXiv papers than they do citing LW posts. Maybe we could also have some facility to make it easier to add the trappings of an academic paper (like abstracts and signposting sections).
That’s not the best idea, but I imagine there are ways to make LW posts more Respectable™—either by converting them into a more Respectable™ format, or by shifting people’s perception of LW.
[1] https://x.com/ohabryka/status/2062196198539501686
Why not just make an academic journal, identify some alignment researchers who frequently use LW and have PhDs, then aggregate their karma over month-long time slices, call that the “review”, and then publish an edition every month? Maybe the authors would need to clean up their work a bit.
This seems easier to implement than (the good version of) my idea.
I wouldn’t want to require PhDs. I think most of the best contributors to LW don’t have PhDs—I can think of at least one who doesn’t even have a high school diploma, although he doesn’t post too often these days. Although having a “LW journal” with a PhD requirement seems better than not having one at all.
The PhDs are the reviewers, not the authors.