Elliot Olds

I'm working on a new project called Impact List. It involves creating and popularizing a ranking of the top ~1000 living people by their positive impact via donations. Think of it as like the Forbes List except for donation impact instead of wealth. 1/6 forum.effectivealtruism.org/posts/LCJa4AAi…

It'd be awesome if we can achieve that, but it seems hard because frontier models require huge amounts of capital to build, and at some point having the best frontier model could allow whoever controls it to pull away from everyone else. If we are in a situation where highly concentrated power is likely it seems worth it to try to make this AI more like Marcus Aurelius and less like a random emperor.

@elliot_olds The only thing that causes good behavior (which I guess might be related to “alignment”) is power being distributed.

I want my AI agent to be primarily “aligned” (i.e. loyal) to *me*, not to you, and not to all of humanity. And honestly, I want your AI agent to be primarily loyal (“aligned”) to you, not to me, and not to all of humanity.

I think the intuition is something like: A single human on its own isn’t that powerful. A big group of them are (bc of coordination, culture, faster knowledge). I think it’s surprising how powerful humanity is given that if any human was examined without the benefit of culture, it would look very unimpressive. Single LLMs already benefit from human culture, but “AI culture” might be a big unlock not entirely different from the transition from [no culture] —> [human culture]. IMO the Clawdbots need online learning abilities before this has much effect on their power, but I think moltbook makes the possibility of a big increase in AI power driven by inter-AI culture more obvious.

I don’t find this that interesting. Of course LLMs can talk to one another. What am I missing? A forum seems a very boring and unsurprising thing to put AIs in. If they launched a “best of” newsletter that was good, I’d be interested.

I agree with your conclusion but I don’t think the second effect you list (“they require Bob to pay more to acquire the ETH he wants to spend on gas”) affects fees. For years people have been making arguments that assume that if token price 10xs, fees will 10x. But fees will always settle at the level where supply/demand for blockspace intersect. This ‘token price affects fees’ idea only persists because: (a) In the early days when fees were negligible wallets used to hardcode fees in terms of the token. If fees are low enough, people don’t care and don’t try to pay the lowest fee they can, so it can look like fees rise in lockstep with price. (b) When price rises it usually corresponds to a period of higher onchain activity and more blockspace demand.

I think eth will always be valued ~99% on monetary premium and only ~1% on fees, even over the long term / decades. This isn't because I'm bearish on long term fee growth. Quite the opposite. It's because it's difficult for me to imagine a sustainable increase in total fees that isn't accompanied by an increase in monetary premium. Fee growth tends to "push up" monetary premium, mechanically. What it means for fees to go up: Bob is willing to pay up to, say, $100 for his whale transaction. But he's hoping to pay a lot less. The minimum Bob has to pay is the greater of the implied anti spam fee to price out competing uses of his blockspace vs. the SoV fee to convince holders to sell him the gas token. I.e. Bob's USD fee only goes up if one or both of these things happen: 1. A lot of onchain growth means there are more competitors for the blockspace Bob wants to buy, so he has to bid more to price out the competitors and secure the blockspace for himself. 2. ETH holders become more confident, and they require Bob to pay more to acquire the ETH he wants to spend on gas. (2) already represents a direct increase in monetary premium. (1) represents general L1 growth and an increase in L1 network effects, which I believe causes an increase in confidence and monetary premium. So if total fees go up, monetary premium probably went up, too. I.e. an increase in "E" implies an increase in "P", so P/E never gets very low. And this isn't bearish, it's normal L1 economics, imo. Of course, fees and P/E can be affected by lots of short term factors, such as big token launches, outages (20% of validators went offline this week, blob fees spiked), capacity increases (higher gas limit or blob target), etc. I'm talking about long-term analysis of sustainable fees vs. monetary premium. Here's what I think happens in the coming years: - onchain grows 10,000x, eats the economic world (most ppl agree with this) - btc continues to do ok. diminishing returns but new highs (most ppl agree btc will do well, i'm relatively bearish here, I don't believe in $1M per coin or anywhere close) - btc dominance shrinks significantly as onchain growth causes an increase in non-btc total crypto market cap. This is because onchain is mostly not about bitcoin and I believe this fact will affect relative confidence in btc vs. other coins (this is a moderately controversial claim. btc bulls would have us believe that btc dominance sustains or even goes up as onchain hypergrows) - as onchain grows, ETH dominance rises substantially because eth's L1+L2 model will win (both L1 and L2 growth) and this will devastate alt L1 market share (controversial claim) - eth L1+L2 global ubiquity and market share dominance will accrue huge value to ETH, multi trillion. Fees won't play a significant role in this value accrual. The P/E ratio will fluctuate but never approach equity levels. - in the years to come, long after eth hits multi trillion, onchain growth will eventually outpace eth L1+blob blockspace supply growth, and fees will sustainably rise to a non-trivial % of total ETH supply. P/E ratio improves but never approaches equity levels, not even close. At this point, ethereum has for a long time already been an entrenched global public institution used routinely by every government in the world.

The other day Japanese GCR hit me up and asked if we’re friends who ride majestic, translucent steeds, shooting flaming arrows across the bridge of Hemdale. My response: "I would follow you into the mists of Avalon". Lessons.

Agreed that Leopold is awesome but I think it's fair to classify him as a type of doomer. He's written several times on his blog that his p(doom) in the next 20 years is 5% (forourposterity.com/response-to-ty…, see the footnote after "look away, Tyler!"). Leopold also said "I think by far the best current writing on AI risk is Holden Karnofsky’s" (forourposterity.com/nobodys-on-the…, see the footnote after "leave that for a footnote"). (For any observers who aren't aware: Karnofsky cofounded OpenPhilanthropy which is the main source of funding for EA and doomer orgs.) Karnofsky has said "I don’t currently have much sympathy for someone who’s highly confident that AI takeover would or would not happen (that is, for anyone who thinks the odds of AI takeover in this nearcast are under 10% or over 90%)". I'm hoping that your admiration for Leopold leads you to be less dismissive toward the 'doomer mentality'. I know the doomer vibes are uncool and the Yudkowsky crew is overconfident, but if you spend a lot of time thinking about the object level issues I think it's quite hard to not at least have a p(doom) of a few percent. If you haven't read it, I highly recommend Karnofsky's "most important century" blog post series: cold-takes.com/most-important…. I think it's on the same quality level as Situational Awareness and AI2027.

the most important thinker in AI, in my opinion, is this 23 year old. leo aschenbrenner. he has been more right, both in a testable predictive sense and in a market sense than virtually anyone else. and most importantly, he's not an AI doomer, he's not an e/acc, but rather a secret third thing.

@ChrisPacia @darin4really @BobMurphyEcon The question is whether you can get paid an amount that would keep a human alive at a reasonable quality of life. The human wage would be nonzero, but it might be the equivalent of 25 cents per hour if AIs/robots are way more efficient at everything.

It's not that it assumes they can't scale up, it's that scarcity dictates that they can't scale as fast as wants can grow. As long as that remains true and as long as humans are physically capable of performing those next tasks, there should always be things to do that you can get paid for.

This is the best counter-argument I've seen to the comparative advantage approach I've been using on the AI stuff. I'll think more about it and report back, but wanted to relay the concern. Hopefully bulk of humanity doesn't end up relying on politicians to ride them for sport.

@ArthurB I used Linux (Mint) on my laptops/desktops from 2013-August 2025, then got a macbook. The mac experience is better and I should have done this way sooner. I suggest getting a new macbook pro m4 and reporting back in two months.

The Pixel is a small computer you can carry with you, it also takes stunning pictures, it's superior in every way to the iphone. I suspect the main reason people in tech use iphones, aside from a good dose of conformism, is because they use macbooks, and the reason they use macbooks is because they're afraid of running Linux on a laptop, but they really shouldn't.

@BjarturTomas This makes it seem like the first domino for any future psychosis was some sort of psychedelic experience that corrupted his metaphysics.

I would like to register the prediction that Emmett Shear is deep into LLM psychosis and this will become obvious to everyone soon.

@nearcyan I have a similar thing. If you find something that helps I hope you write about it.

i might post a tl;dr some time, but the short story is if i get 6-7 hours of sleep i feel very sick, dissociated, and unable to function. but if i get 8 i will feel great. but i have an inability to nap (entire life) and prematurely wakeup at +6 hrsfrom sleep onset regardless of anything i have tried every test,supplement, drug, cbti, etc, and am awaiting my genome sequencing in a week or two which is kinda the only hope i have left at this point

i am looking for an agentic person to help me try to work with/schedule with the best sleep doctors/researchers in the world for a relatively abnormal issue. please DM me if this sounds perfect for you, i usually feel too sick and unable to make calls and appointments myself

@lumpenspace As another rat that you got to read your Land doc a while back, I’d still be interested in this response. x.com/lumpenspace/st…

back on my old shit (getting rationalists to read Nick Land) (gosh im becoming the person anna thought i was)

@ercwl Really glad to see this as someone who tried to ASI-pill you two years ago. Let me know if you want help finding events and people who are more suited to your current interests and understanding. x.com/elliot_olds/st…

i'm finding it increasingly difficult to participate in modern discourse to the point that i don't really know what to do or how to talk to people over the past few years, i've seen the capabilities of ai continue to improve, and all the goalposts of what ai should/shouldn't be able to do constantly shift it's completely obvious that we've discovered a way for machines to learn things. they're now at the capabilities in math and coding where they outcompete most humans on most task. it should be clear to you that the fundamental barriers that you thought would prevent ai from getting better haven't actually prevented ai from getting better. it was only a year ago that you said that their capabilities would flatline because we're running out of training data. that sentiment peaked around september last year, and for all the lamentations about data walls or efficient compute frontiers, new paradigms (in RL and others) *have worked* machines continue to get better. we don't know where they'll stop. and the methods by which they think (cot, scratch pads, python tools) which have barely existed nanoseconds on the cosmic calendar, already *do* show forms of reasoning that allow them to reach impressive conclusions if we don't agree about this, that machines will continue to learn, and we'll continue to explore paradigms for them in which to think, and that it is atleast quite likely that they'll continue to get better, i already don't know how to converse with you if you still observe this from a "oh the statistical parrot machine spat out a token in a sequence that was likely, wow, such reasoning!" point of view, i already don't know how to converse with you and here's where things get more complicated if you're with me this far, it should be reasonable to describe our current world as one where a new form of possibly superior intelligence is arriving and is steadily improving now, where things get completely bizarre for me is where *you* think that your opinions of what this new form of intelligence can and cannot do matter or are relevant. and this bleeds into topics like quantum and everything else for me from my point of view, you are now a chimpanzee on planet earth, studying the arrival of "humans" you climb up every tree on earth and jump down from it, and you establish that gravity seems to work equally on all surfaces of the earth, and from this experiment you establish that regardless of how smart this new "human" is, it will not be possible for it to master space travel. because *you* understand gravity if you don't approach the world from a point of view that you don't actually know anything about what an intelligence smarter than you is capable or not capable of doing, i already don't know hot to converse with you and i'm not talking about a leap in intelligence that requires something impossible that has never happened in history before. i'm just talking about the difference in intelligence between chimpanzees and humans. which has happened before. and led to space travel. the only thing that was required for this superior form of intelligence to evolve on the planet (from molecules to *us*) was for a rock (earth) to get hit by an ice ball and then spin around the sun a bunch of times. the ingredients that produce intelligence are not complex, they only require iterations and feedback from the universe and it appears that this exact process is now happening in machines which do not require the thousand-year process of evolution, because we have invented a way for rocks to think in silicon, and the substrate of intelligence (the matter on which it runs) have shifted to something where it's far more malleable and the iterations are much faster to me, all of these things are rather simple, easily observable phenomena. they're all ongoing and they're real. but in all conversations, basically no matter where i look, everyone is still stuck in a "but hoomans can't do space travel because muh gravity"-type of reasoning and are incapable of embracing the very realistic prospects that all our models are quite likely to soon be broken i tire of the conversation, and find that i do not have much to add to what you are currently talking about, because it appears that we are so diametrically opposed from each other in our understanding of what is happening in the world that we can not begin to have a useful exchange on current topics

@KeiranJHarris @YellowKoan @KremyChelin I’m interested in more about the workflow (enough for someone with no video editing experience to try this) and also the total $ cost in compute to make this.

@YellowKoan @KremyChelin Still, only having to do this much work would’ve seemed like magic a few years ago! And happy to share more about the workflow if people are interested!

The last time intelligence exploded on Earth, it wasn’t exactly amazing for everyone else. Here’s a fable about risks from transformative AI (made with Veo 3)

@alreadydawn @cxgonzalez If we’re in a time sensitive high stakes situation, being more analytical than we’d want to be in the long run seems justified.

@cxgonzalez Not at all. They're stuck in head energy and completely disconnected from their hearts and souls. It's spreadsheet brain maxxing.

Effective Altruism is basically right

@Romy_Holland @TylerAlterman Even if it didn’t have new flow, “match up everyone who wants a partner, then wind down the company” could still be extremely profitable if done well.

@TylerAlterman if an app were this good it would acquire an incredible reputation and ppl would pay a ton of money to use it. there are always new people entering the dating market, so it would have a solid flow of new customers.

Reality check: Imagine if a dating app was so good that it introduced you to your perfect match in the first week. This dating app would go out of business. I rarely use dating apps because the incentives are misaligned. Instead I go places in person where there are cool ppl

I think that's worth considering, but I doubt that encouraging random people on the Internet to enforce norms with cruelty is the best solution. I give some reasons here: x.com/elliot_olds/st… The anonymity of the Internet also tends to embolden people who are just looking for an excuse to be cruel. I'm not opposed to people shaming others IRL for anti-social behavior.

What do you think about the element of context - that some kinds of discourse, or topics of discussion, are inappropriate for an algorithmically mediated public stage that selects for provocation, intentionally collapses context, and encourages clashes between incompatible social norms? In a system like that, inflammatory outliers benefit from attentional feedback loops driven by conflict and controversy. Many might say, just block or mute - but individual actions can't compete with systemic amplification. I doublt anyone would have a problem with her per se if she was a private person, all things being equal. But she has(d) a clear desire to be a public figure and she's an evangelist for her sensibilities. That makes what she's doing everyone's business. It's her noteriety and the inescapability of her influence that people are responding to. It feels imposed, and therefore agressive, and therefore threatening, even if no one, including her, intended for it to be that way.

I’ve been irl friends with @Aella_Girl for ~ten years.  Recently she received an unusual amount of cruelty, then friends defended her character (I'll cosign that she's unusually honest, kind, truth-seeking, generous, brave, etc), then people doubled down on the cruelty. The main justification I see: "it doesn't matter how many admirable qualities she has because she's publicly advocating things which are corrosive to society." But this isn’t enough to justify cruelty even if true. Any outspoken person you have political disagreements with is advocating things that are socially corrosive according to your worldview. Is it cool to be cruel toward all such people? The only plausible justification (maybe) I can see for cruelty would be if she were acting in bad faith. Some people do think she's burning the commons as part of a machiavellian quest for fame and fortune, or is faking her persona to troll people. I've known her through many phases of her life, starting when ~no one had heard of her and she was just scraping by financially. The Aella that you see now is the same person as she was back in 2015 at small late night gatherings of philosophy nerds. If you think she's putting on an act or advocating stuff she doesn't sincerely believe in, I'm confident that you're mistaken. If you think you have good evidence that she's acting in bad faith, comment and let's discuss.

@jon_charb @ryanberckmans Props for outlining the strongest case for ETH as a SoV: that conditional on ETH being the platform that the new financial system is built on, ETH is then the only option for using a trustless SoV in that system. (btw Ryan, good job but you should highlight this argument more)

Btw real respect to @ryanberckmans doing the debate He clearly engaged in a good faith convo, and I know many others with similar views to him who wouldn’t and don’t Real convos > yelling at each other on Twitter I had fun

My guess is no, but I don't think it's that implausible that it can be justified as a punishment for extreme antisocial behavior, aimed at discouraging future similar behavior. But I'm pretty skeptical of practical systems to dispense cruelty. I especially think that people on the Internet thinking they should sometimes be cruel to help protect society has some terrible properties, like: (a) if your audience is big enough, some people will be badly wrong and be way more cruel to you than is justified, and (b) a bunch of people making independent decisions can result in way more total cruelty toward someone than almost all of them intended. I think there's a strong argument for "never be cruel" similar to the argument for why we don't want everyone making utilitarian calculations for whether they should murder in specific situations.

@elliot_olds @Aella_Girl Do you actually think cruelty is justified if someone is acting in bad faith? What does cruelty get you there?

Then I think the interesting questions become: (a) how do we distinguish between advocacy which deserves bullying as a response, vs. advocacy where responses should be civil? Things like "rent control" or other normal political issues also fall into the 'corrosive to society' bucket. (b) Even for cases where someone is advocating something especially heinous, is bullying actually better than strongly calling out their ideas? Is saying "you're a filthy skank" actually better for protecting society than "you're advocating grooming kids, which results in a lot of trauma and ruined lives"?

Agree with your general point but if you are advocating for things that are generally corrosive to society (I.e. grooming kids into OnlyFans) at some point there has to be a corrective and right now our main societal corrective appears to be populist leaning invective. So in that way the initial business model invites this correlate.

Good question. I don’t think it would be obviously unjustified in that case. I still don’t think cruelty would be the best option, but I wouldn’t call it “not plausible”. Then the question becomes: is Aella pushing for more of some heinous crime, or something that (if it were bad) would be more like rent control.

@elliot_olds @Aella_Girl > The only plausible justification (maybe) I can see for cruelty would be if she were acting in bad faith. I need to clarify. Cruelty would be unjustified against someone who truly believed that [heinous crime] was good for the world, and regularly did that thing in good faith?