Sarah Cheng 🔸

• Sarah Cheng 🔸 26 May 2026 14:21 UTC

  4 points

  0 ∶ 0

  in reply to: calebp’s comment on: Con­sider start­ing a for-profit com­pany instead

  Just to clarify, are you responding to the “Likely that more than 10% of your post was drafted by an LLM?” message? That’s now always shown at the top of the editor, to notify users about our new policy. So that’s not related to the body of your post. :) We do currently use Pangram but the results are only for internal use, and don’t affect the editor.

  I can see why that would be confusing, so I’ll update the wording to be more clearly a question.

• Sarah Cheng 🔸 6 Apr 2026 17:04 UTC

  10 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  Sharing an overview of the CEA Online Team’s Q2 OKRs (we run this Forum)

• Sarah Cheng 🔸 2 Apr 2026 14:43 UTC

  29 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  Just a reminder that you can customize your own Frontpage feed, so if you’d like to give serious posts a chance today you can hide April Fools’ Day posts.

  Click on “Customize feed”, add “April Fools’ Day” by clicking on the + to search for the relevant tag, then click “Hidden”.

  

• Sarah Cheng 🔸 23 Mar 2026 18:50 UTC

  2 points

  0 ∶ 0

  in reply to: Kestrel🔸’s comment on: AGI & An­i­mals: Dis­cus­sion Thread

  Oh no, thanks so much for flagging this! Toby was on holiday today unfortunately, so I’ve just updated it.

• Sarah Cheng 🔸 24 Feb 2026 16:58 UTC

  5 points

  1 ∶ 0

  in reply to: Midtermist12’s comment on: The Dis­cov­ery Prob­lem: Why EA Can’t Find What It Isn’t Look­ing For

  I do think an argument over funding allocation is importantly different from the one made in your post — for example, I think convincing EA groups to do more of this work could be an effective way to improve the situation that could also be essentially free. Although if you’re trying to make a point similar to Doing Prioritization Better (“Since cross-cause prioritization work (and to a lesser extent cause prioritization) is presently rare, and has considerable benefits, the EA community may well be radically misallocating its prioritization efforts.”) then I broadly agree. :)

  I’m also curious whether you think CEARCH matches the solution you’re thinking of?

• Sarah Cheng 🔸 23 Feb 2026 22:52 UTC

  5 points

  0 ∶ 1

  in reply to: NickLaing’s comment on: Nick­Laing’s Quick takes

  We’re actually currently working on updating our policy on AI-generated content, so this thread (and follow up poll) is helpful! :)

• Sarah Cheng 🔸 23 Feb 2026 22:34 UTC

  8 points

  2 ∶ 0

  on: The Dis­cov­ery Prob­lem: Why EA Can’t Find What It Isn’t Look­ing For

  Interestingly I’ve been thinking about something similar myself (in the context of what to do with the EA Forum this year), though I do feel skeptical that “shut down the EA Forum” is the best solution. I appreciate a lot of the points you bring up, and I am always happy to hear critical feedback about the Forum, but I do find your post too overconfident^[1] and I personally don’t think the existing data warrants that level of confidence.

  My best guess is that, although there is no perfect system for this right now, the Forum is close enough that it’s worth trying to make it serve this function better. There are many examples of Forum posts of this sort getting significant attention, such as Policy advocacy for eradicating screwworm looks remarkably cost-effective (which I think contributed to Launching Screwworm-Free Future – Funding and Support Request), Interstellar travel will probably doom the long-term future (which I believe led to the author getting his current position at Forethought), and Frog Welfare (one of the highest-karma posts from 2025). I would not characterize the comments in those posts as “aggressively red-teaming” — in fact they skew quite supportive, moreso than I would personally like (but I think I am unusually happy to receive criticism).

  You also mentioned EA Funds — I’m not that familiar with their grantmaking processes, but just looking at their “Featured grants” list I see that Alfredo Parra was given a “6-month salary to do prioritization research and community building focused on reducing extreme pain in humans”, plus he’s written extensively about this work on the Forum and recently founded ClusterFree. This seems to me to be a success story for current systems.

  ---

  In any case, I think a lot of people around EA have been successfully impactful because they do things rather than just post on the Forum. Since you think this is a problem, I would encourage you to take action to try to improve the situation, for example:

  1. You could post a thread on the Forum asking people to pitch potential Cause X candidates, to get a better sense of “what might we be missing”. You’re welcome to contact the Forum team if you’d like us to consider pinning it or promoting it via other channels. I would be excited for more Forum users to proactively try to make the Forum a more valuable community space.

  2. We’re very open to Forum users running events, since our capacity is stretched pretty thin. If you want a green-teaming contest to happen on the Forum, especially if you’re willing to put in the effort to help organize it, contact us!

  3. Pitch this as an activity for EA groups — I think this could both be valuable for the community and a valuable exercise for groups to work on together, to practice EA thinking. (And I’d love to see them post their results on the Forum! 😊)

  4. I would guess that, if you attended some EA events or conferences and talked to some people about this, you could find at least one other person who would be interested in starting a project to address this problem (especially if you had a list of “potential Cause X candidates” that they thought looked promising). I bet if you could show positive results from a small side-project, you’d have a significantly easier time getting funding to work on this more or convincing someone else to work on it.

  5. If you have specific suggestions for the EA Forum (including if you want to expand on your “shut it down” idea), I’d be happy to hear it! :)

  1. ^

     Which may just be an artifact of the AI-assistance

• Sarah Cheng 🔸 18 Feb 2026 23:16 UTC

  27 points

  4 ∶ 0

  on: Why I will not be par­ti­ci­pat­ing in the draft amnesty.

  Hi 98373, I run the EA Forum team so I thought it would be helpful to respond.

  First off, I’m always happy to hear feedback about the Forum, especially critical feedback, so thank you!

  I’ll also say that I think not participating in any specific Forum event or even not using the Forum at all is very reasonable, so I’m not here to try to convince you otherwise. Things are not always a good fit for all people.

  But I would like to correct the places where I think your post is misleading:

  1. “you cannot delete your post”

     1. You can archive any draft post, which effectively deletes it from the site. That doesn’t delete the text from our database, but you can always just contact our team and ask us to do so.

  2. “When you make a comment, after a period of time, you cannot delete your comment at all.”

     1. I think what you’re referring to is that users can’t delete their comment when it has a reply. There is no time component.

        1. This was a decision made before my time, but my guess is that this is just a complicated case due to needing to weigh the interests of many users (including the respondents) — as above, you’re welcome to contact our team if you want to delete such a comment and we can discuss on a case-by-case basis.

     2. It’s also very possible that you’ve run into a bug. We have a small team and there are plenty of issues to fix. Our codebase is open source so you’re welcome to take a look at how this is implemented.

  TL;DR if you’d like to do something with your Forum data that you can’t do via the site, you can just contact us and ask us to help.

  It is irresponsible for the EA Forum not to provide the easiest and best opportunities for users to preserve their anonymity while participating in the community… The EA forum does not have that limitation because it is not aiming to maximise profits by on selling data to advertisers or government agencies.

  I think we do pretty well at this given our very limited resources. To make a Forum account, you only need to give us an email address and come up with a username, neither of which need to be closely associated with your real identity. So I’m not quite clear on what threats you are thinking of that can’t be addressed by current systems — in all your examples there are pretty straightforward ways to disassociate your Forum account with your real identity. (As above, if you want help changing your account username or email address you can just contact us.)

  We have orders of magnitude less resources than most comparable apps (for example, last year we spent less than 2 FTE on the Forum). We are a nonprofit, and have a responsibility to use our charitable dollars thoughtfully. As a very small team, we certainly won’t have state-of-the-art secure software. I hope anyone who is concerned their software usage putting them in physical danger will default to not providing identifiable personal information to websites, especially in a case like the Forum when it’s clear that your username will be publicly visible and discoverable by search engines. For example, don’t use an identifiable username or email address when creating a Forum account.

  For anyone who has uncertainties, I think it’s very reasonable to not even create a Forum account. There are also many ways to engage with the EA community outside of the EA Forum, for example via in-person groups, conferences, or other programs.

• Sarah Cheng 🔸 30 Jan 2026 16:26 UTC

  4 points

  0 ∶ 0

  in reply to: NickLaing’s comment on: Sarah Cheng’s Quick takes

  The surprise is what kind of sweet! ^^

• Sarah Cheng 🔸 29 Jan 2026 23:45 UTC

  11 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  @Toby Tremlett🔹 and I will be repping the EA Forum Team at EAG SF in mid-Feb — stop by our office hours to ask questions, give us your hottest Forum takes, or just say hi and come get a surprise sweet! :)

  Reminder: applications for EAG SF close soon (this Sunday!)

• Sarah Cheng 🔸 16 Jan 2026 0:44 UTC

  2 points

  0 ∶ 0

  in reply to: Ben_West🔸’s comment on: Sarah Cheng’s Quick takes

  Thanks for the question Ben! The main reason that this is a priority is to help EA Funds (which is now part of CEA) grow and diversify their donations, by making it easier to gather info from donors^[1] and build relationships with them, and giving us more freedom to optimize the UX of the donation flow. AWF in particular has ambitious 2026 plans and a significant funding gap, and we’d be excited to help them reach their donation goal for this year! :)

  1. ^

     GWWC, the primary platform EA Funds has used historically, defaults to opt-out for donor data sharing. As far as I understand, this prevents us from being able to contact the majority of donors. We recently added the option of donating via every.org as well, which is opt-in by default so that’s improved the situation.

• Sarah Cheng 🔸 15 Jan 2026 15:33 UTC

  2 points

  0 ∶ 0

  in reply to: JP Addison🔸’s comment on: New Fo­rum Li­cense: Creative Commons

  Thanks JP — that’s right that we originally showed it at the top of the post editor for first-time authors (and it wouldn’t appear anymore after the user checked the box). Now I believe, because we moved the checkbox to the new user onboarding flow, it doesn’t appear on the post page anymore.

• Sarah Cheng 🔸 9 Jan 2026 20:48 UTC

  10 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  For those who are interested in what the CEA Online Team^[1] is up to, I’ve set up a new OKRs doc for 2026 and summarized our Q1 2026 plans.

  1. ^

     Our team runs this Forum

What do you want to see from the Fo­rum team this year?

• Sarah Cheng 🔸 23 Dec 2025 16:46 UTC

  6 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  Quick note: Forum support coverage will be lighter than usual this week and next as people are off on holiday, so you may encounter some process delays (for example, things may take longer than usual to get approved). We expect to be back to normal on Monday Jan 5.

  Hope you all get to enjoy some relaxing time off as well! :)

• Sarah Cheng 🔸 26 Nov 2025 20:33 UTC

  4 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  I’ve updated the public doc that summarizes the CEA Online Team’s OKRs to add Q4.2, which takes us to the end of the year.

  Posting a bit late since I just got back from a long holiday — here’s a bonus photo from my trip! :)

  

• Sarah Cheng 🔸 8 Oct 2025 16:24 UTC

  3 points

  0 ∶ 0

  on: EA Hand­book Au­dio Read­ing?

  Hey, thanks for asking! I’m not sure if there’s a specific reason — it’s likely that simply no one prioritized doing it. Note that I think many of the posts in the EA Handbook besides the first one do have audio, it’s just not consistent.

  The Forum Team has manually enabled audio on all the posts, but some of them will require additional work from Type III Audio (who provides the AI narration), so it may take some time for them to appear on the site.

  In the meantime, you’re welcome to make your own audio or video versions of the posts! I think it would be great for them to reach a wider audience. :)

EA Fo­rum mod­er­a­tor con­flict of in­ter­est policy

• Sarah Cheng 🔸 1 Oct 2025 19:01 UTC

  4 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  I’ve updated the public doc that summarizes the CEA Online Team’s OKRs to add Q4.1 (the next six weeks).

• Sarah Cheng 🔸 23 Sep 2025 15:51 UTC

  8 points

  0 ∶ 0

  on: Sarah Cheng’s Quick takes

  Quick warning: crossposting from here to LessWrong is currently buggy, due to some ongoing infrastructure changes on their side. If you run into any issues saving or publishing your post with “Crosspost to LessWrong” checked, that may be the reason.

  We’re working with LW to resolve any bugs, but they have higher priority issues right now so I’m unsure how long it will take to fix. In the meantime, you can try unchecking that box and manually posting in both places. You can help us diagnose the issue by reporting any bugs you see.