Lorxus

• Lorxus 18 Apr 2026 3:12 UTC

  −5 points

  −3

  in reply to: habryka’s comment on: Let good­ness con­quer all that it can defend

  I think if the US had kept slavery around, that would have pretty solidly tipped things into the negative, though that feels like a bit of a dodge of your question. That feels more like “if the US had turned out less good”.

  I’ll remark here that the US did do that, to a very real extent. Sharecropping replaced chattel slavery and was barely better. And to this day, slavery is not illegal in the US—it’s just the only the government or its appointed representatives get to do it, and only as part of punishment for a crime. But this sets up the obvious bad incentives that we very much see—over-policing and a near-total lack of interest in rehabilitation, with the result that the US has the 5th largest proportion of its population imprisoned in the world—more by far than any other major country. (~5.4k/​million; the only fair-sized countries that even come close are Turkey (~4.2k/​million), Brazil (~4k/​million), and Russia (~3k/​million).)

  I think if the US had actively invaded Europe in return, or actively invaded other parts of the world that had stable governments and were more their own blooms of civilization, that I think would have also tipped it into the negative, but also feels a bit like dodging your question.

  What do you want to call the entire period of US colonialism and adventurism? I’m sure the Hawaiians would like to hear your thoughts on this; there was no reason to invade them but fresh fruit and a maybe a naval base. Or what about every time US forces destabilized a functioning democracy that was a touch too friendly with the Reds for their taste, only to replace them with an authoritarian strongman who brutalized and robbed his citizens and was a staunch regional ally of the US all the while?

  I think if the US had not accidentally brought smallpox, but had encountered the Aztec and Mayan empires and the native American population at its full peak, and had driven them down to the same levels with the same brutality, that, feels to me like it introduces very serious uncertainty, and I don’t know, and I feel confused, but I think I would say that it wouldn’t have been worth it.

  At the start, the introduction of smallpox was perhaps accidental. This state of affairs did not last. And also—do you really want to claim that carelessness at such a scale deserves a moral pass? Is your heart right with the incentives that that sets up?

  If I had known the synthesis of it all, I would not have felt the need to write either this post or the previous post, though I think they’ve both been pretty helpful in putting various intuitions on paper. I would be surprised if this was the last post on adjacent matters.

  I certainly hope so. I thought a lot of the intuitions here were worth using and thinking about, and I was excited to read about them, but getting jumpscared by American exceptionalism and a touting of the White Man’s Burden was not what I expected to see from someone respected on LessWrong today or indeed ever.

Lorxus Does Bud­get Inkhaven Again: 04/​08~04/​14

• Lorxus 11 Apr 2026 18:21 UTC

  3 points

  0

  in reply to: philh’s comment on: Does descal­ing a ket­tle help? The­ory and practice

  It very much does not! It comes as a powder, a little like dried lemon juice, and IME a solution of citric acid boils at somewhat more than 100 *C. I recommend it primarily because citrate is a much better solubilizing agent for metal ions than acetate and thus handles things like limescale and even rust vastly better. (Take as evidence the fact that “drink lots of lemonade and go on a roller coaster” is, while maybe the third-fakest-sounding piece of medical advice, an actual first-line treatment for kidney stones.) The one major caveat is, it usually comes from bioengineered Aspergillus mold, so if you object to GMO products or have a severe black mold allergy, it might not be suitable.

• Lorxus 11 Apr 2026 8:22 UTC

  1 point

  0

  on: Does descal­ing a ket­tle help? The­ory and practice

  Probably way late but: I strongly recommend citric acid over vinegar for descaling.

• Lorxus 7 Apr 2026 22:16 UTC

  4 points

  2

  in reply to: Mateusz Bagiński’s comment on: Rae­mon’s Short­form Feed

  For my part, I agree on object level with @Raemon but don’t share the same reason. My line of reasoning is: humans are (largely) moral agents. What a moral agent does to other entities says a lot about that moral agent. IMO LLMs probably aren’t yet moral patients, even with the complicated scaffolding and meta-level circuitbreaker LLM, but a lot of things about them sure are moral-patient-shaped. Accordingly, I claim that it’s still good to keep promises to LLMs and to refrain from lying to or cheating against them; at best you are practicing and making it easier for you to lie, cheat, and break deals with definite moral patients, and at worst you are in fact just plain doing that to a legitimate moral patient. Whether or not Claude’s a moral patient, I claim that you should still act largely as if it is (within reason).

Lorxus Does Bud­get Inkhaven Again: 04/​01~04/​07

• Lorxus 20 Mar 2026 17:11 UTC

  2 points

  0

  on: Videogames for Rationalists

  I will never get tired of shilling 4D Golf, especially given the section on games with weird space and time mechanics. Achron is another deep cut here. I think Stephen Lavelle is an unsung legend—Stephen’s Sausage Roll and English Country Tune deserve more press.

  Among games already listed in your post or in comments, +1 on Tunic, Baba is You, Opus Magnum and its literally days-new DLC (and also SpaceChem and Infinifactory, as well as Molek-Syntez, The Codex of Alchemical Engineering, and KOHCTPYKTOP: Engineer of the People for deeper cuts), Patrick’s Parabox, Case/​Rise of the Golden Idol (and its DLCs), Return of the Obra Dinn, The Witness, and Blue Prince. Of Paradox games I think I recommend Vicky 3 most strongly.

• Lorxus 11 Mar 2026 6:18 UTC

  1 point

  0

  in reply to: TsviBT’s comment on: Pro­mot­ing en­mity and bad vibes around AI safety

  Chewing it over more, I think you may have neglected to consider Newcomblike self-deception as a possible factor in Sinclair’s razor. It’s not necessary for the person to be lying about what they believe, or for them to have consciously convinced themselves of that lie. They can just have a big convenient cognitive blind spot.

• Lorxus 9 Mar 2026 21:47 UTC

  5 points

  2

  in reply to: TsviBT’s comment on: Pro­mot­ing en­mity and bad vibes around AI safety

  While it seems true that promoting symmetric enmity pretty much always comes with a big cost (though it could in theory come with benefits too), which is approximately what I think you were saying, it doesn’t seem true that promoting one-way enmity is necessarily coming with a big cost compared to the alternative.

  On my model, the large additional cost that it imposes is half the point. If you’re in a situation of one-sided enmity, and you would rather not be, then an obvious thing to do is destructively seize leverage and better your BATNA by making that particular tragedy symmetrical. Then, your counterparty which was previously treating you as an enemy with impunity will now have a reason to come to the table (and you win) or be forced to engage in open public warfare when previously they may have been deniably catty (and you win by means of everyone losing).

  On this model, it is if anything much worse when serious one-sided enmity gets started, or when one of the parties behaves so as to be indistinguishable from an enemy, or acts rashly and then tries to avoid paying repair costs. It’s also especially bad when battle lines are unclear, such that from one reasonable point of view, person P is obviously on team T, and from another reasonable point of view, they’re equally obviously not.

• Lorxus 9 Mar 2026 21:40 UTC

  3 points

  0

  in reply to: TsviBT’s comment on: Pro­mot­ing en­mity and bad vibes around AI safety

  Also, for the record, I’d volunteer my time to talk with anyone who is currently doing capabilities research at an AI research group, or who is seriously considering doing that, and try to explain why they shouldn’t but in a way that is kind and open and understanding. (I don’t think I have a legible track record of doing this, but I would unaccountably claim that there’s a significant chance I’d be good at this for a substantial subset of such people.)

  There’s this one Upton Sinclair quote I think about a lot in this context. I imagine you’ve seen it?

• Lorxus 7 Mar 2026 7:00 UTC

  4 points

  0

  in reply to: Garrett Baker’s comment on: Maybe there’s a pat­tern here?

  I just finished writing a short essay today about, basically, how so many civilian-use technologies end up getting used for war in ways not expected at invention. Bell-casting. Compasses. Coal-tar dye and the resulting chemical industry. Photography. Steam engines and train travel.

  If it’s good for anything, it’s good for getting advantage of some kind over other humans possibly by doing things they straight-up can’t, and if it’s good for gaining advantage over other humans, that advantage can always be somehow parlayed into their subjugation and death.

• Lorxus 2 Mar 2026 23:02 UTC

  17 points

  13

  on: OpenAI em­ploy­ees: Now is the time to stop do­ing good work.

  ITT: People who have never heard of a pen-down or work-to-rule strike in their lives.

• Lorxus 27 Feb 2026 23:01 UTC

  4 points

  0

  in reply to: johnswentworth’s comment on: What‘s in your list of un­solved prob­lems in AI al­ign­ment?

  To what extent are these questions you still consider important? To what extent are they still badly understood? What progress, if any, have you seen on them? What would that progress even look like?

• Lorxus 27 Feb 2026 3:55 UTC

  3 points

  1

  in reply to: Kabir Kumar’s comment on: AI #157: Burn the Boats

  This doesn’t answer my question at all. We seek leverage, right? Ways to make a given dollar go much further towards (e.g.) doom reduction? Then why isn’t there already an initiative for taking a shockingly large, surprisingly bipartisan, not actually that unsuccessful movement with good (received) AI takes, and funding and activating it some?

• Lorxus 27 Feb 2026 1:33 UTC

  3 points

  2

  on: AI #157: Burn the Boats

  Neil Rathi: at a bernie town hall and he just mentioned the metr plot?

  OK, dumb question: are we (and by we I mean, like, EA/​AI safety/​longtermism) trying to get Bernie and associated politicos elected, like, at all? And if not, why not? Probably it’s easier to run PR/​get out the vote type things for politicos with good AI takes (and, ah, alignment) already than to try to talk the median politico into being definitely well-aligned on AI?
  
  Edit (2026-03-09): https://​​x.com/​​SenSanders/​​status/​​2029301587647046034
  I’ll accept those Bayes points now please.

• Lorxus 5 Jan 2026 19:26 UTC

  6 points

  2

  on: The Plan − 2025 Update

  Thanks for the yearly update! I have some thoughts on why we care about string diagrams and commutative diagrams so much. (It’s not even just “category theory”.) I’ll poke you later to talk about them in greater depth but for quick commentary:

  For string diagrams it’s something like “string diagrams are a minimal way to represent both timelike propagation of information and spacelike separation of causal influence”. If you want to sketch out some causal graph, string diagrams are the natural best way to do that. From there you start caring about monoidal structure and you’re off to the races.

  For commutative diagrams the story is different but related, though admittedly I understand what’s going on with [commutative diagrams]+[sparse activations] way way less. I’d say it’s something like “the existence of a satisfied commutative diagram puts strong constraints on other aspects of the neural net, like what form the latent space(s?) and maps to and from them have to look like and what they have to do and what information has to get preserved or discarded”.

  For one last observation, a friend’s been poking me about the sense that constraints and equipartition/​environment are dual to each other, and that there’s a correspondence (for bounded systems at least) between phase volume size-change and the sign of something like an informational analogue to thermodynamic temperature. (And also that your approach is importantly incomplete in currently only dealing in theory and not engineering, but for my part I think that that’s priced in to how you talk about your plan.)

  Bother me on Discord?

• Lorxus 28 Dec 2025 9:33 UTC

  1 point

  0

  in reply to: JennaS’s comment on: Glu­cose Sup­ple­men­ta­tion for Sus­tained Stim­u­lant Cognition

  +1 this. Wouldn’t it work pretty much just as well if not better to just, like… eat a spoonful of honey under those circumstances?

• Lorxus 15 Dec 2025 17:33 UTC

  3 points

  0

  on: Towards a Less Bul­lshit Model of Semantics

  An intriguing attempt to cash out semantics in a gearsy way, of a kind I haven’t seen anywhere else but personal notes and a single much less gearsy research paper. This ended up seizing my attention during a summer research program and made me believe in the explainability of nearly all things and the tractability of all simply-explainable open questions once again. “What followup work would you like to see building on this post?” Yes. My answer to that is “yes”.

• Lorxus 15 Dec 2025 17:30 UTC

  3 points

  0

  on: Anvil Problems

  A concise breakdown of what I’d call a type of development trap—a way in which you can end up pushed back down into a dire bootstrapping problem. I might want to see a larger taxonomy of those, but that’s not what this was for.

  What links here?

  • Deeper Re­views for the top 15 (of the 2024 Re­view) by Raemon (14 Jan 2026 23:59 UTC; 45 points)

• Lorxus 15 Dec 2025 17:28 UTC

  1 point

  0

  on: White­board Pen Magaz­ines are Useful

  True, useful, and inspiring. It led me to make my own variation on them, made me realize a way that my brief teaching career might have been more comfortable, and ultimately led me to write up a brief post. I still carry one around. I’d hoped other people might also start making these for themselves, and improving on the technology, but that largely hasn’t happened.