magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
No bio...
User ID: 1103

I know what he said, but I was deviating slightly; I think that given the Chinese IADS and given a committed-to-AGI CPC (note that this latter is a condition that I do not think is necessarily true IRL), there is probably no way to actually destroy the Chinese capacity to pursue AGI without nuclear attacks (against the IADS, but also against the datacentres themselves; it's not like the CPC doesn't have the resources to put them inside conventional-proof bunkers if it fears an attack, after all, and actually invading China to put an end to things that way is roughly in the realm of "either you drop a couple of hundred nukes on them first to soften them up, or this is as much of a non-starter as fucking Sealion") and even if there were a way to do it conventionally, this almost certainly exceeds the threshold of damage that would get the Chinese deterrent launched. Thus, it is a lot more pragmatic to simply open up with a nuclear alpha strike; you know that this ends in a nuclear exchange anyway, so it's best to have it on your terms. I would agree that it's best to keep to conventional weapons if e.g. Panama were to try to build Skynet.
I'm not advocating Nuclear War Now IRL, because the situation posited is not the real situation; the USA has not made the offer of a mutual halt to AI, and I find it fairly likely that such an offer would actually be accepted (it's not like the CPC wants to end the world, after all; they're way up the other end of "keep things under control and stable, no matter the cost"). To the extent I'm less opposed to nuclear war than I'd otherwise be, it's because I suspect that the gameboard might be in an unwinnable state - and mostly on the US side, because of too much of US discourse being held on platforms controlled by AI companies (YouTube, Facebook, Twitter are all owned by companies/people that also do AI, and devices themselves are mostly Microsoft/Apple/Google OSes which also do AI; the latter is relevant because e.g. the Apple Vision Pro is designed to function as a brainwashing helmet) and Andreessen Horowitz having potentially captured/bribed the Trump admin on AI policy - making a mulligan seem like it would probably lower P(Doom). I'm not going to go out and start one for that reason, though, even if I knew how; Pride is my sin, and it's not even close, but I still don't have that much of it.
AI safety is myopic, obsessed solely with the dangers of race dynamics above all else. Besides the danger of decentralization, there's also the danger of losing the race. Who is to say that the US can afford to slow down with the Chinese breathing down their neck? They've done pretty well with the resources available to them and there's a lot more they could do - mobilizing vast highly educated populations to provide high-quality data for a start.
Eliezer Yudkowsky has explicitly noted* the alternative solution to this problem:
If intelligence says that a country outside the agreement is building a GPU cluster, be less scared of a shooting conflict between nations than of the moratorium being violated; be willing to destroy a rogue datacenter by airstrike.
Frame nothing as a conflict between national interests, have it clear that anyone talking of arms races is a fool. That we all live or die as one, in this, is not a policy but a fact of nature. Make it explicit in international diplomacy that preventing AI extinction scenarios is considered a priority above preventing a full nuclear exchange, and that allied nuclear countries are willing to run some risk of nuclear exchange if that’s what it takes to reduce the risk of large AI training runs.
That’s the kind of policy change that would cause my partner and I to hold each other, and say to each other that a miracle happened, and now there’s a chance that maybe Nina will live.
If you think China is going to destroy the world, the correct solution is not to destroy the world yourself as if RL is a game of DOTA; it's to stop China from destroying the world. Tell them that doing this will end the world. If they keep doing it, tell them that if they don't stop, you'll nuke them, and that their retaliation against this is irrelevant because it can't kill more Americans than the "all of them" that will be killed if they continue. If they don't stop after that, nuke them, and pray that there's some more sanity the next time around.
*To be clear, I was nearly done writing a similar essay myself, because I didn't think he had the guts to spit it out (certainly most top Rats don't). Apparently he did.
There was a tagline on an old rationalist blog - was it Ozy's? - that I felt summed up this religion well: "The gradual replacement of the natural with the good".
It was Ozy's, and it's "The gradual supplanting of the natural by the just". I think your (mis)quote actually represents Rats as a whole better, though, because, well, Ozy's word choice hits a point that Scott's explicitly decried.
That's still incredibly flawed to assume reading through my limited postings on a single website make you meaningfully less of a stranger who does not know me or what I believe/have previously done.
100 posts would be enough to get a decent start on a picture of me. Not a complete one, certainly - even a hypothetical person who'd read all 35,000+ posts I've made on the Internet still wouldn't know everything, and I'm the sort to randomly admit to crimes if they're relevant - but it'd be meaningful.
Just for the record, going to someone's userpage by clicking on their handle does, in fact, allow you to see everything that person's posted on theMotte. Also, theMotte is small enough that some people (though not me) do read basically everything.
TBQH, I was actually one of the volunteers rating some of your posts in this thread, and one of them (four upstream from the one I'm replying to now) I did actually rate as "deserves a warning". Not so much because of the amount of heat (which was quite high), but because you went cross-thread to keep after CBNOS. There's this bit explicitly in the rules:
Please remember that you can always drop out of a conversation, ideally (though not necessarily) with an explanation; if a user follows you and harasses you, report them.
...which is intended to reduce yelling matches via giving an escape valve when somebody can't be civil anymore. I use that valve now and then, which is part of why I have zero warnings here. When you go cross-thread to keep after someone, you're jamming that valve.
Time in this community will thoroughly disabuse you of that notion, presuming you can avoid the traditional leftwinger meltdown and flounce-out when you realize that other people are going to continue to be allowed to argue back.
No, I think people with that view do indeed tend toward liberalism.
The mistake is in assigning the word "liberal" incorrectly; social justice isn't liberal, it just (in the USA) has been wearing the word (and Officially Designated Intellectualism) like a skin-suit.
Some further thoughts:
-
theMotte seems like it's kind of intended as a place for people to play on simulacrum level 1, not level 2 where you're telling lies to children to make them behave in the way you've explicitly reasoned about without (yet or ever) being able to do that explicit reasoning themselves. @Amadan am I barking up the wrong tree here?
-
I think "prison conditions" is much less susceptible to Ozy's argument than torture, because prison conditions up to a point have at least one tradeoff that is always in play i.e. expense. Like, at one end of the Pareto frontier you are basically Auschwitz, enslaving everyone who can/will work and incinerating those who can't/won't; this is maximally cheap (indeed, potentially cheaper than free) and also horrifying (though not maximally horrifying; you're not doing "rape them to death, eat their flesh, and sew their skins into your clothing, and not in that order" because holy shit that's a lot of extra work). At the other end, jail is basically a hotel, with maintenance; this is maximally nice, but also horrifically expensive due to all the stuff that gets broken or stolen (raising the cost well above the already-substantial cost of a hotel that can actually kick people out). Nobody can currently afford the good end; if you look back in history a lot of societies couldn't afford better than the horror end (though as you look back further, you don't have incinerators or scalable oversight, so this starts to tend more toward "summary execution, (mass) unmarked grave" which AIUI was nearly omnipresent in prehistory). Thus, any "though the heavens fall" seems like an obviously contingent principle which for most values of "okay" that modern Westerners accept would, if applied to a pre-Black-Death society, not work; you'd be overthrown if you tried to implement them by a) peasants starving from your taxation and/or b) other elites trying to avoid a) in order to save their own skins, and if you somehow weren't overthrown then the law-abiding populace dies in plague from undernutrition and the criminals either escape or starve. You are thus, in a sense, always talking price, in a way that doesn't play nice with injunctions; you can argue that 2025 El Salvador is wealthy enough to comfortably pull off better conditions, or you can argue that it's importantly not at the Pareto frontier ("these conditions are worse than death; summarily executing them all would be more humane" is an example of such an argument, because summary executions are cheaper than any prison where the inmates can just sit around) - and the reason I'd not have commented there is that I literally don't know enough about the particulars to participate in either discussion about this case - but if you think there's a one-size-fits-all correct answer you're committing the just-world fallacy.
-
You will note that I haven't contradicted the claim that there are things worth letting the heavens (literally or metaphorically) fall to avoid, because obviously there are such things even for a first-order utilitarian; the heavens literally falling is not maximally bad, and the heavens metaphorically falling (e.g. WWIII) is generally less bad still, so it's worth WWIII to prevent the Earth being destroyed and it's worth the Earth being destroyed to prevent Allied Mastercomputer.
This is a better argument, and if you'd been plain about this rather than engaging in hyperbole I'd not have chimed in.
You said these:
[I reject] the assumption those arguing against human-rights violations are somehow responsible for anything that can be attributed to not committing them.
There are lines that one should not cross though the heavens fall
I responded to those, because they sketch out a policy which I think to be insane (i.e. "one should let the state fail rather than take the gloves off").
This is The Motte, where you're supposed to "always attempt to remain inside your defensible territory, even if you are not being pressed". Either defend your claim or retract it; don't deflect and yell at me for responding to what you plainly said.
Bonitas non est pessimis esse meliorem. (Being better than the worst is not goodness.)
I think you're misinterpreting me here. My point here was that if the only way to stop MS-13 from imprisoning sex slaves in abominable conditions is to imprison MS-13 in slightly-less-abominable conditions (which also stops a bunch of other crime), the latter option strictly dominates the former.
I can now draw a 75-lb. compound bow, although I can only manage a couple of draws right-handed (left-handed is easier). So that's what, 6 months from when I started out with noodle-arms? Practice makes progress, as my Mum would say.
There are lines that one should not cross though the heavens fall, and those arguing against crossing those lines do not thereby assume culpability for the actions of others.
Zvi Mowshowitz:
Have democracy and civil rights been dramatically violated? Oh yes, no one denies that. But you know what else prevents you from having a functional democracy, or from being able to enjoy civil rights? Criminal gangs that are effectively another government or faction fighting for control and that directly destroy 15% of GDP alongside a murder rate of one person in a thousand each year. I do not think the people who support Bukele are being swindled or fooled, and I do not think they are making a stupid mistake. I think no alternatives were presented, and if you are going to be governed by a gang no matter what and you have these three choices, then the official police gang sounds like the very clear first pick.
MS-13 literally has a motto of "kill, steal, rape, control". Do you think they treat their sex slaves better than Bukele is treating them? When your choice is "do X, or state failure and warlords do X anyway", you need to be exceedingly-invested in not personally sinning, to a degree that I'd argue is selfish, to pick the latter. This is not to say one should not look for third options, or try to create them, but no, do not actually let the heavens fall.
You missed #4: the journalist is a lying piece of shit and the sources do not exist. This wouldn't be my first pick, but it is a possibility.
To be clear; I'm anorexic (in the proper sense); I don't get hungry*. Obviously, this largely negates the "ate too much" side of the coin.
*I recently discovered that I can get cravings for specific foods; when I started training with my bow, I started getting meat cravings, presumably because I needed protein to add muscle.
This is certainly what I do - weigh myself every couple of weeks, if my weight's gone up stop eating lunch for a few days, if it's gone down start eating dessert for a few days. Hadn't heard the name "Hacker's Diet", though; it seems kind of too obvious to need a name and I kind of thought anyone who's actually at target weight would be doing it.
That seems greatly distinct from what he said; "Trump is running on vibes and obsessions rather than means-ends reasoning" is a theory of mind, and an analogous statement is true of many people (at least in a lot of situations) regardless of whether it's true or not in this particular case - this is the simulacrum levels 3 and especially 4.
Until Trump climbed down today the slide was showing no signs of stopping whatsoever. We're barely more than a week removed from the original announcement.
I will say, having my net worth in term deposits (because I'm a pessimist, if largely for other reasons) has served me well this week.
The cockpit security doors are less obviously insane than most of the anti-Twin-Towers measures. There's a drawback in the whole "pilot suicide" issue, but pilot suicides are a lot less bad than ramming attacks and are in some ways easier to stop.
Yes, the Flight 93 scenario is the norm now which makes it far harder to pull off a lookalike, but some defence in depth isn't crazy.
Far bigger economy and military (aside from nuke count, which is rapidly rising) = far more credible threat to depose the US as hegemon.
By landmass, Canada is the second-largest country in the world, after Russia.
No, it's not. The USA and China have more land area than Canada (although not by much).
It comes out second on total-area calculations because of its very large territorial waters due to its many lakes and vast coastlines.
I figured I'd better ask because, well, it's not like Trump can't order a nuclear test, and it probably wouldn't even be the most shocking thing he's done this year (though ordering a nuke used in anger would). Hell, I'm not even sure it'd be a bad idea, if only to check that they still work.
Not always. Yudkowsky's example of "if the world will end in ten years, and you know this, this won't help you make money" holds water. Things shaped similarly to this also tend to be hard to make money from, due to difficulty collecting winnings and/or spending them; being a nuclear doomer might mean that I'm unusually well-equipped to survive a nuclear war, but I haven't figured out a way to actively profit from it.
There's also the famous (or perhaps infamous) saying, "the market can remain irrational longer than you can remain solvent". Frauds tend to go up and up and up before they come down; knowing they're fraudulent without knowing the exact timing of said fraud's discovery means you might not be able to hold out against the margin calls.
- Prev
- Next
From where I sit, hoping for neural net alignment is itself a strategy with ~0 chance of success. Reality is under no obligation to give you a "reasonable" solution.
More options
Context Copy link