Could you talk more about how you think the norm might persist? 

Maybe there are other signals that aren’t vulnerable to the same kind of faking, like evidence of integrity and discretion paired with a track record of impact of the type you’re looking for. Seems harder to easily evaluate, but also seems like a more common way to evaluate alignment with a cause’s mission. 

Separately, while I’m vegan myself, I think using veganism as a signal is less reliable than it may appear because it appears not to meet the same effectiveness bar as other popular interventions, and there’s honest disagreement within EA on whether EAs should generally follow a vegan diet. I think I’d have the same concerns with using other marginally good but costly lifestyle choices, like donating a kidney, as a signal as well, though I haven’t thought about a lot of individual cases. 

Good luck! What outcomes are you planning to measure?

The best thing I’ve read on it so far is this article by Kelsey Piper. 

Thank you for posting this! I appreciated both learning how others feel about this and your thoughtful commentary. 

Your last paragraph especially reminded me of The Plague by Albert Camus. Beyond its political allegory (and topical subject matter), I read it as an absurdist case for persistent altruism and an approach to everyday life that centers present experiences over hope and despair that map onto the narrative of a broader cause. Both ideas resonated with me later when I read posts in the EA handbook on altruism and scope sensitivity like Nate Soares’ On Caring. 

I haven’t spent a lot of time thinking about this, but I suspect a couple reasons to continue pursuing this contract beyond the present revenue include (1) retaining relationships and a reputation that provides option value for (especially defense-related) future contracts and (2) increasing the likelihood that safer models are used in high-stakes settings, especially ones that could carry some non-negligible AI-related risks. While those are plausible (and plausibly right) lines of reasoning, I’m writing them without taking a stance on specific details that have central importance to their truth (e.g. are Anthropic’s models “safer” than competitors’? Seems quite likely based on reputation, but I’m not well-informed enough to make that claim confidently). If you’re right and the alternative is the use of another lab’s models for the same jobs, and if the article is right that Anthropic’s models are the only ones being used on classified networks, then I don’t think there are good reasons for Anthropic to intentionally cede that space to competitors. 

I thought this could be relevant to a few people interested or working in bioethics:

The Bioethics Interest Group is one of several dozen Special Interest Groups that operate out of the Office of Intramural Research at the NIH. Its monthly virtual seminars "provide a discussion forum, consider different views, and present research on complex ethical issues in medical research." If you are interested in or working in bioethics, I thought you might find it interesting to sign up for its newsletter so that you have the opportunity to read about and consider attending its seminars. 

Center on Long-Term Risk, Rethink Priorities, Fish Welfare Initiative, EA Infrastructure Fund

Thank you for the correction! I think what threw me off was the previous video’s impressive 6M views by comparison. I just spent some time looking at examples of how highly successful videos from smaller channels often perform, and I think my perception of the relationship between initial performance and total views was miscalibrated because of how I've observed popular videos from well-established channels performing.