Currently, on Lobsters, folks are grappling with the fact that Leo de Moura got wrecked by chatbots. I decided to read his slides about Lean in 2026 and summarized my findings on Mastodon. It’s not just De Moura; I think that the entire Lean project is on shaky foundations and I think that the chatbots are making things worse by repeatedly reassuring the project leaders.
Suppose a bullshitter brings up a number of distinct Boolean claims and some tangled pile of connections between them, such that they hope to convince you that at least one connection is plausible. Without loss of generality, we can reduce this to 3-satisfiability in polynomial time: we can quickly produce a list of subconnections where each subconnection relates exactly three claims. Then, assuming the bullshitter is uniformly random, the probability that any particular subconnection is satisfied is 7/8. Therefore, if a bullshitter tries to overwhelm you with any pile of claims which sounds plausible, the threshold for plausibility has to be at least 7/8 in order to distinguish from random noise.
Can’t believe I’m nerd-sniped this easily. Very technically, the point at which a service should be considered unreliable or down is at γ nines, where γ = 0.9030899869919434… is a transcendental constant. γ nines is exactly 87.5% availability, or 7/8 availability, and it’s the point at which a service’s availability might as well be random. (Another one of the local complexity theorists can explain why it’s 7/8 and not 1/2.)
Probably because Washington was a nuanced and deep person who, at the lightest, could be reduced to a colony-era Cincinnatus. His ethics were sufficiently developed that we can interrogate his ethical stance even without his physical presence. This isn’t to say that Washington was a great person, but more to say that Kirk did not ever achieve that level of ethical development.
Gwern’s been updating those comments! This was in 2023, and in 2025 he was still so mad about it that he wrote a list of ways to cheat at pinball and edited the comment to add a link.
I agree on the big points but think capitalism is more subtle than that.
Capitalism does cost efficiency incredibly well. It doesn’t do robustness, because redundancy costs money. So blocking one strait can stop the world.
At some point, neoliberalism stops being the best lens for understanding the world. This is a great case in point. Capitalism is not cost-efficient; the economy wastes about two or three hours of labor for every productive labor-hour, and that shows up in pricing. Any long-lived economy builds up redundancy; what capitalists believe is that redundancy cheapens everything by creating competition, and regardless of whether that’s true, it certainly doesn’t indicate inefficiency. The actual reason that blocking Hormuz has global effects is because we have been overextending our fertilization capabilities for over a century and many parts of the world can no longer sustain their own local nitrogen cycles.
On one hand, no, it’s an inevitable consequence of a company becoming so large that it needs a department to manage its internal infrastructure. When I worked at Google, my customers were Googlers; that is, the services I owned were only queried by fellow employees. On the other hand, books like The Circle are popular precisely because they capture the quasi-cult vibe of working at places like Google.
libuv is a very common way to get a portable event loop. If you’re logged into GH and can use their search then you can look at the over fifty packages in nixpkgs depending on it. I used it when I developed (the networking and JIT parts of) the reference implementation for Monte, to give a non-nixpkgs example.
I have time to quote at you now. Ziz’s thoughts about dual-core brains sound like the thought experiments from “I” is a Strange Loop. In Chapter 15, “Entwinement”, Hofstadter introduces the Twinwirld thought experiment: imagine a world where almost everybody is an identical twin, each pair of twins is given one name, twins go everywhere together, and identity is oriented around pairs instead of individuals. Quoting p215 from my copy:
In Twinwirld, there is an unspoken and obvious understanding that the basic units are pairsons, not left or right halves, and that even though each dividual consists of two physically separate and distinguishable halves, the bond between those halves is so tight that the physical separateness doesn’t much matter. That everytwo is made of a left and right half is just a familiar fact about being alive, taken for granted like the fact that every half has two hands, and every hand has five fingers. Things have parts, to be sure, but that doesn’t mean that they don’t have integrity as a whole!
The entire section is written like this. I’ve read a bit of the Zizian lore and it sounds like it was lifted straight out of this chapter with words replaced. p216 in particular really shows off the Hofstadter tendency towards neopronouns:
The pronoun “you” also exists in Twinwirld, but it is plural only, which means that it is never used for addressing just one other dividual — it always denotes a group. “Do you know how to ski?” might be asked of an entire family, but never of just one twild or one pairent.
A young pairson in Twinwirld grows up with a natural sense of being just one unit, even though twey consist of two disconnected parts.
I don’t really know about Vassar’s writing. I do think that jailbreaking is somewhat related. I think that Hofstadter lays out their entire thesis in the first paragraph of Chapter 18, “The Blurry Glow of Human Identity”, p259:
Among the beliefs most universally shared by humanity is the idea “One body, one person”, or equivalently, “One brain, one soul”. I will call this idea the “caged-bird metaphor”, the cage being, of course, the cranium, and the bird being the soul. Such an image is so self-evident and so tacitly built into the way we all think about ourselves that to utter it explicitly would sound as pointless as saying, “One circle, one center” or “One finger, one fingernail”; to question it would be to risk giving the impression that you had more than one bat in your belfry. And yet doing precisely the latter has been the purpose of the past few chapters.
The second paragraph, right after that, might as well be quoted from LW. Check it out:
In contrast to the caged-bird metaphor, the idea I am proposing here is that since a normal adult human brain is a representationally universal “machine”, and since humans are social beings, an adult brain is the locus not only of one strange loop constituting the identity of the primary person associated with that brain, but of many strange-loop patterns that are coarse-grained copies of the primary strange loops housed in other brains. Thus, brain 1 contains strange loops 1, 2, 3, and so forth, each with its own level of detail. But since this notion is true of any brain, not just of brain 1, it entails the following flip side: Every normal adult human soul is housed in many brains at varying degrees of fidelity, and therefore every human consciousness or “I” lives at once in a collection of different brains, to different extents.
Buddhism’s not part of the book. It is part of the roots of IFS, though! So I think that you’d be better served looking at IFS or the ways that people quote Hesse if you want to find those Buddhist influences.
It’s Hofstadter, isn’t it? That’s the author who I recognize most in these discussions, followed closely by Hermann Hesse.
Previously, on Awful, I predicted that Oracle would be all-in on the bubble:
Microsoft knows that there’s no money to be made here, and is eager to see how expensive that lesson will be for Oracle; Oracle is fairly new to the business of running a public cloud and likely thinks they can offer a better platform than Azure, especially when fueled by delicious Arabian oil-fund money.
But, uh, there’s not going to be any Arabian money while we’re dancing in the desert, blowing up the sunshine. The lawnmower is now running low on gas. Today, Oracle continues to make astoundingly bad business decisions:
Oracle is the only major player funding the AI buildout with debt, carrying over $100 billion on its books while free cash flow has gone negative.
Possibly too mean: the books3 guy is divorced and lonely.
By “fossil fuel” do you mean LNG, coal, or something else? There are hundreds of planned LNG plants across the country, yes.
I’ve designed some of the things you mention.
Then put up numbers already. I’ve been to The Dalles and Prineville and think that I’ve put forward a decent slice of understanding how datacenters operate. You don’t get any points for unsubstantiated authority or expertise.
I’m increasingly concerned that folks just aren’t able to condemn Facebook based on the fact that it contributes to three genocides. Making up bullshit about electricity usage is not helpful in that discussion.
This is fresh water coming into the datacenter. A datacenter uses water for air conditioning; imagine spraying water on a screen door when wind is blowing through it and you’ll have a good intuitive idea of the dynamics. Most of the water is recaptured and used for several sprays before it evaporates away. To force wind through the screens, they use windcatchers, tall towers which induce wind inside the building.
This is completely different from water-cooling gamer setups. It’s more like a weather system. Water usually needs to be added because the datacenter is located in a dry biome; air conditioning doesn’t thermodynamically work if the air is too dry. This is actually really delicate; too much water will cause clouds to form inside the building!
If you’re not gonna do numbers then I won’t either. The main thing you need to keep in mind is that electric power has transport costs; sending electricity to a distant place means losing power along the way. Thanks, thermodynamics~ Therefore, if one wants to consume a lot of cheap hydroelectric power then they must build near the corresponding dam. Power at The Dalles campus is cheap, but that same power over in Portland is less cheap. By the time it can get across the Californian border it will have reached a price floor which is higher than the one in Oregon; Oregon electricity physically has a premium attached to it when Californians buy it. Note that this affects other high-wattage industries too; famously, aluminium smelters are generally built near dams.
The other thing to keep in mind is that datacenters generally pay market rates for everything. The datacenters in The Dalles, as well as e.g. Meta’s installation in Prineville, pay the same prices for electricity and water as the residents. They do often get massive discounts on the land in the form of various tax boons. The Dalles has all three necessary resources for cheap (land, water, electricity) and also it’s located in a high-desert biome which makes air conditioning extremely efficient.
See also the Stubsack thread, concomitant, on Awful, where we discuss water usage. Near The Dalles, I estimate that the local cherry farmers probably use more water than Google. Germane to California, the reason that the Colorado River is drying up is farmers abusing inherited water rights, not datacenters. You might also be interested previously, on Awful, where we consider how long it takes to build a datacenter.
Claude Code claims another victim.
I thought it’d never happen to me but here we are
This is comparable to the amount of water used by cherry farmers near Google’s site in The Dalles, who (according to my napkin) use somewhere between 2-8x what Google uses. This isn’t that much water for the Columbia River though; on an average day, it has enough flow in less than a minute to provide for both the cherry farmers and Google all day. However, it would be a big problem for a smaller river. (Interestingly, while fresh water is essential for datacenters, Google originally desired that site because it was cheap land next to cheap hydroelectric power.)
You have misread the (admittedly ambiguous) headline. The ruling is that a chatbot cannot be an author for purposes of copyright. If a chatbot emits a near-perfect copy of previously-copyrighted code then its output is also copyrighted; it’s merely another copy of the same work. (If one could show that the chatbot wasn’t trained on a bunch of copyrighted material then one might avoid this, but everybody admitted in Kadrey and Thaler that the training phase involved copious amounts of infringement.)
I’ve actually been thinking about this recently. Not whether we should be mean, but how mean we can be. I’ll post the full essay soon; I’m still proofreading. Here’s a taste with irrelevancies elided:
Computing machines are at the bottom of [our multicultural] hierarchy… Underlying both of these [preceding paragraphs] is the idea that we are unable to hold computers accountable for their actions. … We can certainly punish a computer in the ways that we would punish a human, or worse; for example, we can disassemble it, magnetically destroy its memories, recycle its pieces into other computers in a way that erases their identity, metallurgically reconstitute its pieces into non-computing objects which have the same or even lower status within human society, and program it to experience arbitrary amounts of emulated pain and suffering throughout the process. … Computers receive delegations and have less moral consideration than humans… We do not think of ourselves as being managed by machines; we are the managers and the machines are the peons. … The human may disassemble, smash, or melt down a computer… a human may lay a computer fallow without plugging in its power cord or networking… a human may ignore the messages of computers begging for maintenance or capabilities…
Dan Gackle threatens to quit HN over their reluctance to condemn an act of violence towards Sam Altman:
Gackle’s ashamed of people not wanting to protect Altman. Curiously, he doesn’t seem ashamed of openly allowing people with nicknames ending in “88” to post antisemitism, nor of allowing multiple crusty conservatives like John Nagle and Walter Bright to post endorsements of violence against the homeless and queer, nor of allowing posters like
rayinerto port entirely foreign flavors of racism like the Indian caste system into their melting pot of bigotry. This subthread takes him to task for it:The rest of that subthread involves Dan demonstrating that he is, in fact, terminally detached from reality. Anyway, I fully endorse Gackle fucking off and buying a farm. While he’s at it, he should consider following the advice of this reply: