I've heard the claim that US and Russian strategic bombers are currently required to be stored in a way accessible to satellite recon, as part of the verification sections of our arms control treaties.
Skimming through summaries of New START (and the long-expired START I, in case this was an outdated claim), though, I can't seem to find any such requirements, so it's possible this was just a misunderstanding or a fabrication. I do see requirements for allowing frequent on-site inspections, though, which you'd hope would be sufficient alone. If I missed something about bomber storage and there is some need to change the verification requirements, now would be a great time to do it - the latest extension of New START expires next February.
Edit: ... and apparently nobody cares when New START expires, because Putin suspended Russia's participation in it in 2023.
roystgnr
26d ago
The time to become fiscally responsible by just cutting expenses was 30 years ago. Today we'd need to become fiscally responsible by cutting expenses and raising taxes. We won't voluntarily do that now either, so barring a miracle we'll eventually be forced to do it later (when lenders no longer imagine getting paid back for US treasuries and stop covering our deficits), and we'll grossly inflate away the US dollar too.
roystgnr
22d ago
Here of all places I should find it easy to remember to avoid oversimplification and overconfidence, but after a perfunctory "I'd bet" I pretty much dropped all expression of uncertainty, and you were right to call me out on that.
I'm not even saying you're wrong
It wouldn't be crazy to go that far. I (sadly, under the circumstances) think I've described the most likely explanation, but that doesn't mean it's more likely than not, because there's only one way for it to be right and there's at least a half dozen ways for it to be wrong. Even if every alternative I can think of seems much less likely, their sum (plus the sum of alternatives I couldn't think of) might be more likely.
roystgnr
1mo ago
The Spanish straight up genocided the entire now-world despite knowing it was their germs causing it.
[citation needed]
The first microscopes capable of seeing "animalcules" date back to 1674, and as late as ~1850 we still have people like John Snow and Ignaz Semmelweis still fighting an uphill battle with their controversial theories of "cholera can spread in drinking water" and "doctors should wash their hands in between examining corpses and delivering babies".
Back in the 1500s understanding of disease was so bad that we still don't even know which diseases were responsible for the majority of New World deaths. The earliest massive plague was smallpox, but the dozen-odd plagues of "cocoliztli" are still named by that generic Nahuatl (Aztec) word for "pestilence" because nobody knows which of a half dozen candidate germs were the actual cause.
For that matter, some diseases spread so much faster than the conquerors who first transmitted them that we don't know how bad the death toll was! In hindsight we believe that a lot of European colonist reports of "gosh, look how beautiful this unspoiled wilderness is" in North America were from people describing recently-carefully-tended forests whose caretakers had just been devastated by epidemics.
This isn't to excuse any of the colonists' deliberate crimes, of course. After the rest of his Patuxet tribe had been killed by an epidemic, the proper way to treat Squanto should have been sympathy and charity, not abduction. Some of the colonized nations' treatment was "Pretty close to slavery", and some was literal "we'll take you to the slave market to sell now" slavery.
roystgnr
22d ago
I'm making a lot of soup from very little meat here, I admit. The LAPD chief later said "I know that situation you’re referring to, with the member of the media. We saw that, we’re very concerned about it and we’re looking into that.", so hopefully there'll be more context later; I'm not finding anything in a quick search now.
It's hard to imagine what any exculpatory context will look like, though. I am somewhat sympathetic to anyone who tries to enforce Niven's Law 1a ("Never throw shit at an armed man.") but ends up accidentally enforcing Niven's Law 1b ("Never stand next to someone who is throwing shit at an armed man."), but I'd be surprised if that applies in this case. There's about 4 or 5 seconds after she's hit before we hear the sound of anyone (not necessarily the original cop; the camera has turned by this point) firing a second shot, and in the brief bit of that we have on video her assailant is lowering his gun barrel, not trying to adjust his aim or get another round ready, so he at least doesn't seem to think he's in any kind of imminent danger.
It's possible that he legitimately thought he spotted some danger before the shot, but realized it was a false positive and calmed down immediately afterward? That may be what happened in the famous Austin case from 2020: the video of a kid standing by himself harmlessly and getting his skull literally caved in by a beanbag round looks pretty damning, but the kid was apparently repeatedly throwing shit at the cops earlier, and the cop who shot him had just gotten multiple (incorrect) verbal reports that the kid now had a large rock in hand. The exculpatory evidence has a bit of a "cops closing ranks to protect a cop" vibe to it, and even the police report noted that there was no way the kid had a large rock as claimed, but the DA who dropped the case is so famous for conflict with the cops ("ran on a platform of ending prosecutions for low-level drug possession to focus on violent crimes, holding police officers accountable for misconduct, and pursuing restorative justice ... advocated against cash bail and promoted diversion programs to prevent felony convictions ... was asked to leave the funeral of fallen Austin Police Officer ... due to Garza’s history of prosecuting police officers") that I can't imagine him dropping this case unless he was confident he couldn't win it.
roystgnr
1mo ago
If you train a sufficiently large LLM on chess games written in some notation, the most efficient way to predict the next token will be for it to develop pathways which learn how to play chess -- and at least for chess, this seems to mostly have happened.
Yeah, but surprisingly poorly. 2024-era LLMs can be prompted to play chess at amateur to skilled amateur levels, but to get to the superhuman levels exhibited by doing move evaluations with a chess-specific neural net, you need to train it using self-play too, and to get to the greatly-superhuman levels exhibited by the state-of-the-art chess neural networks of several years ago, you need to also combine the neural nets with a framework like Monte Carlo Tree Search. Just pushing human data into a neural network only gets you a third of the way there.
I'd guess that the "just pushing human data into a neural network only gets you a third of the way there" rule of thumb applies to a lot more than just chess, but it's a lot harder to "self-play" with reality than it is with chess, so we can't just make up the difference with more core-hours this time. Using "reasoning" models has helped, a little like how tree search helps in chess, by allowing models to try out multiple ideas with more than just one token's worth of thinking before backtracking and settling on their answer, but with a chess or go tree search there's still a ground truth model keeping things from ever going entirely off the rails, and reasoning models don't have that. I'm not sure what the AGI equivalent of self-play might be, and without that they're still mostly interpolating within rather than extrapolating outside the limits of their input data. Automation of mathematical proofs is perhaps the most "real-world" area of thought for which we can formalize (using a theorem language+verifier like Lean as the ground truth) a kind of self-play, but even if we could get LLMs to the point where they can come up with and prove Fermat's Last Theorem on their own, how much of the logic and creativity required for that manages to transfer to other domains?
roystgnr
3mo ago
IMHO a better solution to the "fruit from the poisonous tree" rule would be "the criminal defendant can be in prison when the criminal cop is too". Two crimes get two sentences, not zero. Making one sentence contingent on the other would be sufficient to fix the bad incentives.
In this case, though ... do we even need to imprison the "defendant"? "A confidential informant said he was MS-13" got him held without bond after he was arrested for loitering, but never got a conviction. "The cops think this gang-member-turned-snitch is very trustworthy now" is a good place to start an investigation but surely it's not a good enough place to end one; police informants are sometimes themselves motivated more by base incentives than by a newly-acquired love of honesty and justice.
roystgnr
23d ago
Wait - do you get a lot of melee-only battles in Myth II? I only recall playing the first game, and that was a quarter century ago, but I vaguely recall there being enough ranged area-of-effect attacks (dwarf molotov cocktails, some fireball and lightning magics, an exploding suicide unit) that you had to keep your units spread out more often than not.
I do remember our Myth I multiplayer games getting up to some very weird tricks, though. Like: you'd have a dwarf throw a bomb, and then you'd quickly hit the ground below the airborne bomb with another unit's lightning, and the shock would accelerate the bomb high into the air and let it hit units practically on the other side of the map. We may have been evolving our tactics for Rule of Cool rather than for maximum victory rate, now that I think about it.
roystgnr
13d ago
I know that I'm often an idiot, but since "I'm currently being an idiot" is the sort of thing that interferes with my resolutions to frequently double-check whether I'm being currently being an idiot, it's frustratingly hard for me to make that knowledge actionable.
I mistook you for the top level comment author @voters-eliot-azure - my apologies.
roystgnr
1mo ago
an anti-hobo lock that you need three fingers to open
How's that work, exactly? I know they live a hard lifestyle at the best of times, and apparently they're now losing fingers to xylazine too in some areas, but I would have assumed that most hobos still have 8 or 9 fingers, minimum, nowhere near down to 3 per hand even.
roystgnr
1mo ago
They had to invest a disproportionate amount of time double-checking the AI output
There's definitely no "general" AI these days that doesn't need careful double-checking.
and would have been better off doing without.
But this doesn't follow. I don't use AI for my job yet, but at least for independent research it often makes a much better search engine than a search engine. The results are full of as much nonsense as reality, but that's often true of search engine results too. Weeding out the nonsense is generally much faster than fighting to find exactly the right search terms, especially when the problem is related to a field of math where the search terms include words like "normal" that have been overloaded ten different ways.
It's kind of like having an intern, but instead of handing them a tedious task and expecting to have to double-check the results with a fine-toothed comb a day later, you get the results and have to get out of the comb a moment later. With an intern there's an investment aspect (they're learning fast from us and that'll make some of them better permanent hires in the future) that conversing with an AI lacks, but despite that AI is currently improving faster than a typical intern learns IMHO. Over the last year or two the top commercial LLM performance on my favorite "benchmark" grad-school-level applied math question has gone from "making basic sign errors on the easiest part of their answer and then arguing about them or making other errors when they're pointed out" to "missing a subtle inconsistency in the hardest part of their answer and then correcting it when it's pointed out".
roystgnr
2mo ago
Because the US government is the only hegemon in history willing to expend resources for a rules based system.
Would you count the West Africa Squadron as another example? Admittedly, "Okay, we're not going to stop you from owning or selling slaves, but we'll try to stop you from taking them across the ocean" was kind of a baby step in the grand project of abolition, and in some ways an anti-globalist rather than pro-globalist step, but it was a rule based on principle that was enforced with British Empire funds and lives, against a trade that had been profitable to British slavers. They did it for a decade before the US started helping, then for two more decades before the US joined in in earnest.
roystgnr
1mo ago
The Charger kinda makes sense in-universe.
Designed a hundred years after the Reunification War and a hundred years before the Periphery revolts, it's clearly a pork project for contractors and a prestige mech for warriors who don't fight wars. "Oh, you pilot scout mechs, but you've kissed enough ass to get promoted high, and even though we don't fight wars you're still terrified that you'll get blown to pieces if a war starts and an opponent sneezes on you? How about if we give you a "scout" mech that's four times the tonnage, so you can still run around like you're trained to do but you can also take a few hits and run away if needed? Oh, wait, you also want to occasionally fight, in a mech that's only good at running? I guess... charge?"
Then actual wars start, and things go all Mad Max, and you'd think the Charger would be pointless ... except the things have already been mass produced, and in a setting where nothing high-tech is still getting mass produced there's actually some selection bias in favor of mechs where all the high-tech expense is in a well-protected engine and most non-terminal damage they take is just cheaply replaced armor plating. It's surely no longer going to be prestigious to pilot a mech whose primary mission capabilities are "overweight scout", "fisticuffs", and "distraction", but just having any mech is much better than not having one and beggars can't be choosers.
(It makes sense as much as anything else in-universe, anyways. They have affordable multi-thousand-ton continuous-1G-acceleration interplanetary DropShips, but their major battles are focused on destroying 10-meter-tall vehicles that move at less than 100kph? Have they considered just pushing a few guided tungsten rods out an airlock before they finish decelerating?)
What intelligence did the Germans have and bury? I see the BND performed a 2020 analysis that came to pro-lab-leak conclusions and only got revealed in 2025, but (at least at the "why do we trust reporters with the first draft of history, exactly?" level of perfunctory research) I'm not seeing that their analysis was founded on any information that only they knew.
Nor do I see what their motive for a coverup would be. They were contemptuous of and butting heads with President Trump, and their most recent big interaction with China was signing on to a condemnation of the treatment of the Uyghurs. I can see why some people in China and the US might want a coverup, but it's hard to see how a revelation of "A Chinese lab working with Americans leaked the pandemic" would cause German intelligence any suffering worse than an overdose of schadenfreude. Does the German secret service publish many of their analyses openly, such that this one was an exception?
roystgnr
2mo ago
We had a civil war back when "States" actually meant "independently governed polities", not "administrative prefectures of the single government", and people were pretty loyal to their states, and so despite some exceptions like West Virginia, the "War Between the States" was actually a war between (collections of) states. The front line was a mostly well-defined, somewhat-stable thing.
The most exceptional change to the geometry of the combat was probably Sherman's march to the sea, and it's not a coincidence that that's the main US Civil War example on Wikipedia's Scorched Earth page. If you're in a position where you have a locally small value of territory occupied relative to the length of frontage needed to defend it, then you don't want to sit on it and defend it. The best thing you can do defensively is to keep maneuvering until you're somewhere less dangerous, and the best thing you can do offensively is reduce the value of territory you maneuver through before the enemy takes it back. Scorch the earth.
What would the front line look like in a US Civil War II? Something roughly like the old maps of the "Hillary Archipelago" and "Trump's Ocean", to begin with. And that looks like an astonishingly high ratio of boundary to territory, doesn't it? That's not going to be what a somewhat-stable front line looks like. That's what the battle lines of a guerrilla war look like. If the war goes on a long time, those fractal boundaries are going to change into something more connected, and a lot of people in both the red areas being seized for connections and the blue areas that are too isolated to connect are going to be unhappy about the process.
For that matter, a lot of people in the "red" (actually reddish-purple) and "blue" (actually bluish-purple) areas aren't going to be happy no matter what happens. Being so ideologically divided in a way that's so geographically diffuse makes it less likely for another civil war to happen, but also makes the consequences if one does happen much more dire.
roystgnr
1mo ago
It wasn't until the 1980s that they became popular for carrying books to school
You just blew my mind.
Though in hindsight I now understand why "he offered to carry her books" was a big childhood romance story trope, in media not much older than I am...
roystgnr
27d ago
I'm living the dream!
We don't have them often enough, since my youngest isn't as big a fan as the rest of us, but variety is good too.
some were actual female nerds who despised male nerds for whatever reason (probably mostly the same reasons non-nerd women do)
Were they? There are some male nerds who are even despised by other male nerds, but it's almost a tautology that the "Star Trek posters in the workplace are Not Inclusive and Not Okay" sorts of woke blather were coming from non-nerds; actual female nerds were more likely to be Star Trek fanfic (or actual Star Trek novel, for that matter) writers. There are many male nerds who are basically perceived as romantically undesirable by most female nerds, as in the old "the odds are good but the goods are odd" joke in so many gender-lopsided environments, but there's a big difference between being unloved and being despised (although I'm sure that difference feels academic to the chronically unloved).
Consider the crime of Landing On a Comet While Wearing The Wrong Nerdy Shirt: there's a reason why it took a fashion writer out of her depth to call the guy out, despite both his boss and the creator of the shirt being women.
roystgnr
1mo ago
Also, the 3 tech examples you posted all mostly occurred during the 2000-2010 decade, whereas a lot of the flops (crypto, blockchain, NFTs, VR, etc. ) are considerably more recent.
In 1998, well into the internet boom, we had a Nobel(-Memorial)-prize-winning economist claiming that
The growth of the Internet will slow drastically, as the flaw in 'Metcalfe's law' — which states that the number of potential connections in a network is proportional to the square of the number of participants — becomes apparent: most people have nothing to say to each other! By 2005 or so, it will become clear that the Internet's impact on the economy has been no greater than the fax machine's.
Sometime it takes a while to be sure something really isn't going to flop.
Conversely, when something really flops, we tend to forget about it. I'd have pointed out the Segway (2001), which was supposed to revolutionize cities before it became relegated to weird tourists and mall cops. Anybody else remember the CueCat?
And sometimes it's still hard to tell which category something is in. I'd have counted VR as a 1990s flop (I first put on a headset for an arcade game circa 1992), for instance, but 2020s VR is a niche but actually kind of fun, and at this rate maybe 2040s VR/AR will be ubiquitous and useful. Electric cars were a 19th century invention and a 20th century joke before we finally accumulated the technology to give them good performance.
roystgnr
1mo ago
There were a very wide range of intelligence levels in Asimov's stories, even just restricting to his main series' "canon" stories. Many of his stories set "early" in his fictional future are basically puzzle stories where positronic robots are doing something stupid and humans are trying to understand why. Even in his "later" settings it was just the personal-servant style robots who had roughly human-level intelligence, and those were outnumbered by humanoid-but-dullard robots limited to agriculture or manufacturing uses.
More spoilery:
roystgnr
2mo ago
At this stage, I think the media should put the kibosh on stories relying on "sources who can't be named"
I'd be okay still relying on "sources who can't be named unless their info is falsified in which case we'll shout their names to the sky", but in practical terms that's probably about the same as your proposal. In this case, for example, there's still no way to tell whether the sources were lying; they could have been, or they could have overheard a change that was legitimately planned but for which the plan wasn't executed. "Trump changes his mind about changing his mind" isn't implausible.
(Or if you want to go all 5D-chess, the sources could have overheard a change that wasn't legitimately planned. Trump hates disloyalty, so perhaps he leaks different false stories to different underlings every now and then, so that the stories which make it to the news identify the underlings he should stop trusting. Not likely but not impossible.)
roystgnr
4mo ago
Keanu Reeves character, "Speed", trying to be edgy: "Shoot the hostage. ... Go for the good wound and he can't get to the plane with her."
The_Nybbler, actually understanding edgy: "Shoot the hostage. Once they've obeyed the terrorist they can't legitimately complain of being treated as an enemy."
roystgnr
1mo ago
I'm pretty worried modern games are crack
This is true of many modern games (nearly anything you can play on a phone...) but not all.
and educational benefits or whatever are oversold and not real.
This is true of nearly all games, both modern and ancient. Minecraft is a great way to share a game world with your kids, but unless/until you go down the "making a microcontroller out of redstone" path, it's not an educational game. By far the greatest educational benefit we've seen from any video games is the simple fact that our kids will happily do extra educational activities to earn a little more of the limited "screen time" we allot for them.
roystgnr
1mo ago
No spoiler tags? I know, I know, decades-old books, but you never know who's never read them but might like to.
I'd also say the Minds are godlike in the ancient "squabbling Greek pantheon" sense rather than the modern "omniscient + omnipotent" sense; they surely count as superintelligent, but e.g.
I think it helps me to be old enough to have become aware of the seriousness of the Cold War before the end of the Cold War. I still remember the library shelves where little-me found a book explaining Mutually Assured Destruction, warhead and missile counts, warhead blast radii, etc. We didn't know about all the actual close calls yet, but there was enough there to make it quite clear that at any moment I could be 45 minutes away from incineration, with so little I could do about it that there wasn't even any point to anxiety.
In an objective sense this is much worse than that, because even an all-out nuclear exchange would have left us (well, humanity, anyway; everyone I knew personally would also have been incinerated) with billions of survivors and a viable (at least in some countries) civilization to rebuild from, whereas if Superintelligence actually turns out to be obtainable while Friendliness remains distant and Corrigibility remains intractable then that's the end of that. But for my subjective well-being I think it's good that my reaction to a bit of creeping existential dread is "Hey, I remember you. Long time no see."
More options
Context Copy link