Toby Tremlett🔹

• Toby Tremlett🔹 4 Jul 2026 6:33 UTC

  2 points

  0 ∶ 0

  on: The Learn­ing Trap: What Si­mu­lated Clue­less Agents Re­veal About the Unaware­ness Argument

  Hey Dan, just to clarify for you and other readers, this post wouldn’t be eligible for the competition, because a) we aren’t accepting pieces that have already been published and b) every entry must be submitted via the entry form.

  Sorry if this wasn’t clear in the announcement post! I’ll edit it next week to make this plainer.

• Toby Tremlett🔹 3 Jul 2026 14:47 UTC

  3 points

  0 ∶ 0

  in reply to: Owen Cotton-Barratt’s comment on: New EA Fo­rum LLM-use policy

  We updated the policy in a way that you guys might both like :) We’re using Pangram to assign a label, and there is no longer any need for manual disclosure.

  Thanks both for your thoughts! (This won’t be exactly what either of you had in mind, but I’m sure your feedback fed into it).

• Toby Tremlett🔹 3 Jul 2026 10:46 UTC

  2 points

  0 ∶ 0

  in reply to: Richard Möhn’s comment on: [Er­ror com­mu­ni­cat­ing with LW2 server]

  Hey Richard—this is as immediate as it gets (i.e, it was sent once we’d removed the accounts, deleted their messages etc...)

  Glad to hear not everyone was affected!

AMA: An­thony DiGio­vanni, au­thor of the ‘Challenge of Unaware­ness’ sequence

• Toby Tremlett🔹 3 Jul 2026 6:35 UTC

  30 points

  0 ∶ 0

  on: to­bytrem’s Quick takes

  It looks like we had a pretty massive bot attack last night, with over 600 accounts logging in and messaging users about an iPad scam. Please don’t click on any of their links. The scam looks like this:

  

  We’ll remove the accounts, and karma-gate messaging to 10 karma so that this cannot happen again. Sorry everyone—and thanks so much to everyone who reported this. I can’t respond to all of you, but thank you.

• Toby Tremlett🔹 2 Jul 2026 11:07 UTC

  5 points

  0 ∶ 0

  in reply to: BruceF’s comment on: Cul­ti­vat­ing hope: cal­ibrat­ing the ex­pec­ta­tions for cul­ti­vated meat to end fac­tory farming

  I agree, that’s where most of the conversation went, plus the critique of cultured meat was—as far as I recall—less developed than the rest in the post itself. I look forward to your post—let me know if you’re ever looking for feedback :)
  
  Curated means this post gets pinned to the top of the frontpage for a week, and it’ll have a star next to its name forever. In practice it means it’ll get more readership.

• Toby Tremlett🔹 2 Jul 2026 10:31 UTC

  7 points

  2 ∶ 0

  on: Cul­ti­vat­ing hope: cal­ibrat­ing the ex­pec­ta­tions for cul­ti­vated meat to end fac­tory farming

  I’m curating this post.

  I often say that not every post on the Forum has to look like an academic paper, but it’s awesome that we do get carefully researched and in depth content like this!

  I really appreciate this post, especially after animal welfare has an evidence problem highlighted some concerns with cultivated meat last month.
  
  Thanks @PabloAMC 🔸, and thanks @BruceF for some particularly valuable comments.

• Toby Tremlett🔹 2 Jul 2026 9:25 UTC

  0 points

  0 ∶ 0

  on: to­bytrem’s Quick takes

  Looks like the refactor led to a few days of analytics data getting a little messed up. If you’ve been checking your analytics and seeing zero reads on your post for the last few days, this is why. Should be fixed from now.
  

• Toby Tremlett🔹 2 Jul 2026 8:12 UTC

  2 points

  0 ∶ 0

  in reply to: calebp’s comment on: calebp’s Shortform

  Update- looks like this seemed like a particular issue this week because some part of the refactor broke the algo. Should be fixed now.

• Toby Tremlett🔹 2 Jul 2026 8:05 UTC

  2 points

  0 ∶ 0

  in reply to: abrahamrowe’s comment on: calebp’s Shortform

  I imagine you all could often whitelist certain users or just glance at a post and tell that it is worth pushing through without reading it

  Yep this is true. Though it’s perhaps not reassuring to new authors lol.

  “>15 karma + many curated new posts by the moderation team”

  My current plan is >100 karma, and loads of ‘featured’ posts by the moderation team for the featured page, and then keeping the frontpage as is. Once I’ve built this thing there will be a lot of room for experimentation with the algorithm though.

• Toby Tremlett🔹 1 Jul 2026 12:42 UTC

  2 points

  0 ∶ 0

  in reply to: calebp’s comment on: calebp’s Shortform

  We’ll definitely be doing (3), hopefully in the next few weeks.

  “It doesn’t appeal to me because I want higher info density”—fair!

  How about a normal list of posts but still on a ‘featured’ page. In other words, these would be the best posts, chosen by the Forum team.

  Sort of like the Digest, but all the time, and without the organisation updates and most of community, and with the addition of good recent discussion.

• Toby Tremlett🔹 1 Jul 2026 9:11 UTC

  4 points

  0 ∶ 0

  on: The EA Superstructure

  
  For example, it seems like CEA’s Forum team is focused on labor-intensive activities like active moderation, running public events, or generally contributing to the Forum, rather than high-level algorithm or mechanism design changes.

  This seems accurate but isn’t set in stone, I’m very open to suggestions here. I’m putting together a post about our plans for the EA Forum at the moment, so more discussion of that soon.

Open thread: 2026 Q3 (July—Septem­ber)

• Toby Tremlett🔹 30 Jun 2026 10:53 UTC

  2 points

  0 ∶ 0

  in reply to: calebp’s comment on: calebp’s Shortform

  Thanks! That’s great to hear because it’s a problem I’m thinking about right now.
  Obviously we can’t just not show low karma posts to people because every post was once low karma (and every author too).

  One solution, which I think I mentioned to you before, is to have a more curated and streamlined frontpage, for those who want to only see the best of the (recent) Forum. Below is a mock-up:
  

  

  Would something like this solve your problem do you think? Posts which appeared on this page would be hand chosen, with a similar bar to the Digest, but updated daily.
  
  PS- I know it is bad product discovery to just ask lol. I still think you get some evidence from a strong no or a strong yes.

• Toby Tremlett🔹 30 Jun 2026 8:25 UTC

  2 points

  0 ∶ 0

  in reply to: Matthew C. Halteman’s comment on: An­nounc­ing: Clue­less­ness Cri­tiques Competition

  Hey Matthew!
  
  We’d be happy with externally constructive critiques. For example, there are responses we would value which argue that we should not aim to be impartial altruists or that we should let go of even the broad conception of ‘expected value’ reasoning from Premise 1.
  
  However, eligible posts for the competition will respond to one of the premises, defend a counterargument, or pose a constructive resolution. If your argument doesn’t fall into one of these, I’d urge you to post it during the event week anyway (August 17th) and take part in the discussion, but it wouldn’t be eligible for the competition.
  
  A judge also said to me: “I’d think that ‘externally constructive’ responses are OK, but good ones will answer this problem: It seems like any plausible moral view is going to require some limited form of consequentialism (e.g., you should save two drowning children rather than one ceteris paribus) and we would like some account of how that consequentialism is supposed to work without bringing cluelessness back in.”

• Toby Tremlett🔹 30 Jun 2026 7:20 UTC

  2 points

  0 ∶ 0

  in reply to: calebp’s comment on: calebp’s Shortform

  I don’t think so. Best bet is to use the all posts page, weekly, sort by karma. Then close your eyes when you scroll down to posts below 15 karma.
  
  What was the use case here?

• Toby Tremlett🔹 26 Jun 2026 13:29 UTC

  11 points

  2 ∶ 0

  on: Was Par­ti­san­ship Good for the En­vi­ron­men­tal Move­ment?

  I’m curating this post from a couple of years ago because a) it makes a noteworthy point in itself, and b) I’m interested to hear people’s takes on whether AI is becoming more of a partisan issue.

  For my part, I feel like we are in a strange middle-ground where groups are forming along battle-lines, but most of these groups are cross-party. For example, anti data-centre populism, acceleration/​ techno-optimism, AI existential risk, these have not settled as being the domain of one party or another (especially in the US, where it likely matters most).
  
  Does this match your impression, and does it matter?

• Toby Tremlett🔹 22 Jun 2026 11:30 UTC

  4 points

  0 ∶ 0

  in reply to: Robi Rahman🔸’s comment on: Clue­less­ness: Sum­mary of the ar­gu­ment, why it mat­ters, and counterarguments

  Yep, and I think I disagree that you’re justified in doing that, but I’d be interested in reading the case for it.

• Toby Tremlett🔹 22 Jun 2026 11:02 UTC

  4 points

  0 ∶ 0

  in reply to: Robi Rahman🔸’s comment on: Clue­less­ness: Sum­mary of the ar­gu­ment, why it mat­ters, and counterarguments

  I think we’ll end up going in circles—my point is that cluelessness is a problem with the reasoning that gets you to your preferred action, so if it is true, then the reasoning is faulty and the preference is unjustified.

  Also, since you can still have preferences if you believe in cluelessness, I don’t quite know how you’d compare between your preference to act deontologically in cluelessness world and your preference to maximise impartial welfare in non-clueless world. Both would be good preferences in their respective worlds. IDK how the meta-normative comparison should go here.

  Edit: to clarify, I think our crux might be at this metanormative level, i.e. at comparing cluelessness world preferences with non-cluelessness world preferences. I realise I’m pretty agnostic on how to do that, but a good argument here could cause me to find the wager response to cluelessness more convincing.

• Toby Tremlett🔹 22 Jun 2026 10:34 UTC

  5 points

  2 ∶ 0

  in reply to: Robi Rahman🔸’s comment on: Clue­less­ness: Sum­mary of the ar­gu­ment, why it mat­ters, and counterarguments

  Are you assuming that the previously preferred action would still have some normative force behind it? I think that’s the bit that’s confusing me. The cluelessness argument (if it goes through—idk if it does or not, hence competition) attacks your reasons for preferring that action, so you’re back to the drawing board. It doesn’t make much sense to still insist on the action when you find out that your reasons for preferring it are flawed.