@magic9mushroom's banner p

magic9mushroom

If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me

2 followers   follows 0 users  
joined 2022 September 10 11:26:14 UTC
Verified Email

				

User ID: 1103

magic9mushroom

If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me

2 followers   follows 0 users   joined 2022 September 10 11:26:14 UTC

					

No bio...


					

User ID: 1103

Verified Email

The PRC's doing another round of drills around Taiwan.

There's a real possibility that this one is the ruse of war that the others were meant to make believable; all the stars have basically aligned. The charm offensive failed in 2019-20 with the Darth Vader stunt on Hong Kong, there's a shitshow of a US election campaign in progress (two assassination attempts plus a disqualification attempt), the US President is significantly demented, William Lai got elected Taiwanese President earlier this year, and October's a good month in terms of weather conditions for amphibious assault, plus Beijing's adversaries are re-arming and Biden will be gone before April so that puts some degree of time pressure on them.

I wouldn't panic just yet; even if this is the big one (and it may not be), my guess is that they won't open WWIII with a nuclear first strike on CONUS/Europe/Australia (pre-emptive ASAT use to wipe out US satellites - and probably destroy all other low-earth-orbit satellites as collateral damage - is a possibility, though, so you may lose any communications dependent on those). But anything that you might have trouble doing later - beating the rush to buy bottled water/non-perishable food/aluminium foil/iodide tablets or whatever (not all of those are applicable to all of us), or maybe starting construction of a fallout shelter - this is your advance warning. As far as Guam/Japan/South Korea go, there may be pre-emptive missile attacks on US bases, but I still wouldn't expect cities nuked as part of the opening move so my advice is mostly the same. But if you're in Taiwan itself, I'd suggest getting out; if this goes hot there'll likely be a blockade attempt by the PLAN, so you may not be able to get out later.

To be clear, I'm more worried now than I've been since at least 2017 (the Trump-Kim yelling match) - and I was in Melbourne then, and thus personally at risk. I was mildly nervous back in April of this year, but you'll note that I didn't make a post like this then.

Remember that your life is worth a lot more than a few hundred bucks; it is rational to take action even if you rate the chance of nuclear war as "small but significant". Remember also that it is good to survive; while QoL might suck in the immediate aftermath of a nuclear war, we'll recover, and if you have any ideological goals you will in almost all cases help them more if you're still around to advocate and act for them (note that if you're in the military or can otherwise help win the war, that's a worthy cause; I'm not advising desertion). That said, good luck to us all and I hope I'm worried over nothing.

m9m out.

EDIT: The drills seem to have completed; we seem to be safe for now.

I mean, that's technically true, but somewhat misleading; it's less that he's making "mean tweets" himself and more that he abolished Twitter's censorship bureau to allow other people to make "mean tweets".

I think the reason that we are hardwired to not be psychopaths is that in most circumstances being a psychopath is just a poor strategy that a fitness maximizing algorithm will filter out in the longterm.

It was maladaptive in prehistory due to group selection. With low gene-flow between groups, the genes selected for were those that advantaged the group, and psychopathy's negative-sum.

It is entirely possible to just teach AI morality like children and then let the ecosystem help them to solidify that.

I doubt it. Humans are not blank slates; we have hardwiring built into us by millions of years of evolution that allows us to actually learn morality rather than mimic it (sometimes this hardwiring fails, resulting in psychopaths; you can't teach a psychopath to actually believe morality, only how to pretend more effectively). If we knew how to duplicate this hardwiring in arbitrary neural nets (or if we were uploading humans), I would be significantly more optimistic, but we don't (and aren't).

There's a speech from Order of the Stick which I'm just going to quote in its entirety because it's easier than re-inventing it from scratch.

Shojo: I mean The Game, the big one. The one that each of us plays every day when we get out of bed, put on our face, and go out into the world. Some of us play to get ahead, some of us just want to get through the day without breaking character. It's called "Civilization" No, wait, there's already a game called that... OK, it's called "Society." Your problem is that you don't want to play the game at all, you want to sit on the couch and eat Cheetos while everyone else is playing.

Belkar: Well, why shouldn't I? What's the point of their Society, anyway? It never did anything for me.

Shojo: The point is that if you laugh and spit in their faces enough times, they'll kick you out of the house—which in this extended metaphor means killing you.

Belkar: So, what, you're saying that the only alternative is to show up and play by everyone else's stupid rules??

Shojo: Of course not, my woolly friend. You can cheat. Twelve Gods know that I always did. Nudge die rolls, palm cards, "forget" penalties... but you have to sit down to play first. As long as the people at the table see a fellow player across from them, they'll tolerate you. A crooked player is a pain in the ass, but someone who refuses to play at all makes them start questioning their own lives—and people hate to think. They'd rather lose to a cheater, than dwell too long on why they're playing in the first place.

The expectation up until Trump was that everyone serious in US politics would at least pretend to stay within the bounds of Polite Society as defined by the Cathedral or whatever else you want to call it. Trump didn't. He didn't cheat; he refused to play the game at all, and spat in the faces of those that demanded he do so. That was a fundamental challenge to the legitimacy of Polite Society/the Cathedral/etc., the same way that the Comics Code Authority was fundamentally undermined when Marvel ran a story in defiance of the Code and got away with it - if people can openly defy you without getting immediately punished and forced to repent, you aren't a consensus authority anymore, just another guy on one side of a controversy.

Now, one can certainly say that somebody like Trump was overdetermined to appear once SJ started drastically curtailing what counted as "acceptable for Polite Society". But that's not quite obvious even to me, much less to someone who thinks SJ is "just common fucking decency", and so he gets blamed for putting a bunch of propositions that had previously seemed like bedrock up for debate.

We shouldn't let the apocalyptic scenario of Skynet make us downplay that, or accept it as a lesser problem.

To be clear, a dickhead with a singleton is still plausibly worse than Hitler. The "lesser problem" is still very big. But it is both somewhat less bad and somewhat easier to avoid.

Hitler and Mao were ruthless and power-hungry. But it's beyond any serious doubt that both of them wanted a future with lots of happy people in it; they were merely willing to wade through oceans of blood to get there.

To be clear, I utterly loathe Sam Altman. But that's because I think he's taking unacceptable risks of Skynet killing all humans, not because if he somehow does wind up in charge of a singleton he'd decide of his own accord to kill all humans.

Saying they "sample" goals makes it sound like you're saying they're plucked at random from a distribution. Maybe what you mean is that AI can be engineered to have a set of goals outside of what you would expect from any human?

Nobody has a very good idea of what neural nets actually want (remember, Gul Dukat might be a genocidal lunatic, but Marc Alaimo isn't), and stochastic gradient descent is indeed random, so yes, I do mean the first one.

But I wouldn't expect generality seeking systems to become Skynet.

There are lots of humans who've tried to take over the world, and lots more who only didn't because they didn't see a plausible path to do so.

Yes, if you have a good lawyer

The problem there is that lawyers have been known to get fired for taking RW clients, so it's not so easy to get one. Part of why Trump's legal team sucks so much is that it's common knowledge that taking Trump as a client will result in being blacklisted, and better lawyers have more to lose from that.

Now, the Trump claims regarding 2020 were basically ludicrous AFAIK, and that much is on him, but his lawyers sucking can legitimately be blamed on SJ/cancellation.

"AI safety" needs to focus less on what AI could do to us and more on what people can use AI to do to each other.

Skynet is still the greater problem, both because even an AI-enabled human tyrant would still be pushing against entropy to remain in charge and because the vast majority of humans want a future with lots of happy people in it, while AI samples a much wider distribution of goals.

It's specifically hostile annexation that's banned, where you take territory/people by force over their wishes. There's some degree of grandfather clause for existing state boundaries, but supporting rebels to get what the rebels want (as opposed to what you want) is generally OK (at least as far as the norms go; the state being rebelled against can retaliate).

The Donbass rebels were fine as far as the norms went; other states were free to back the Ukrainian government, and the Ukrainian government had some degree of cause of action against Russia (not that Russia cared), but Russia wasn't breaking the norms. Russia coming into Ukraine under its own auspices to chop off bits of it and annex them to Russia, that's breaking the norms.

As for Israel, as long as the US, or nations in general, maintain border and immigration controls, the State of Israel must continue to exist as a haven for Jewish people persecuted in other countries.

I look at this, and then I look at the Kurds. The exact same argument applies, except far more so because the Kurds are currently persecuted and the Jews aren't. You could also say this about the Uyghurs, or the Rohingyas, or any other nation that does not have a state. Am I missing some reason that the Jews are a priority here?

I assume it's fine to report a spotted bot (e.g. "As a large language model, I can't...") as "bot", as long as it's only in the report function rather than a public accusation?

And as @Goodguy and @MaiqTheTrue noted, there are understandable and possibly-valid reasons for that.

But I felt that @Gillitrut might have needed the pointer on the "is this actually happening Y/N" question.

I think I may have encountered a Russian troll. Specifically, this guy. He went into a bunch of WP articles about US surveillance, ruining them, and when I noticed the pattern and alerted WP he made a few ominous-but-vague threats and then vanished.

At the time I thought he was simply an NSA/CIA agent, but in retrospect I think that's unlikely. He was very sloppy, copypasting entire sections of NSA propaganda into Wikipedia without even changing the "we"s to "they"s, and my read on the Five Eyes is that they're usually slicker than that; a real NSA/CIA agent would also have no motivation to make vague public threats and then disappear, rather than simply ghosting straight away or picking up the phone to threaten someone for real. And if he wasn't a Five Eyes spook, he was somebody pretending to be one, presumably someone intending to get caught in order to frame them for vandalising Wikipedia. Could be a random lunatic, I suppose, but the people with a logical motive to do that are strategic adversaries of the USA, and my read based on PRC external propaganda and the Sam Dastyari fiasco is that 4D-chess shenanigans like this aren't their style. I suppose I'll never know, particularly since I've left Wikipedia.

Depends on your definition of "caught". He can kill identified criminals before they're punished (though this is very much a minority of his kills), but not unidentified criminals (unless he happens to witness it personally; I know his second kill is one of these, and the "associates" L brings in for the Yotsuba investigation might count, but I don't recall any others). He has no special powers of investigation, which is the real bottleneck in the criminal justice system.

I find it interesting, reading the first article, that it decries "right-wing influencers" saying that FEMA is denying other rescue teams access, but the article does not actually say that this isn't true.

The steelman is that institutional DIE focus causes both uselessness and detectable changes in racial/sex ratios, which creates a correlation between those ratios and uselessness - valid Bayesian evidence - even in the absence of significant causation.

The left's "empathy" project has never been unlimited and all-encompassing.

This isn't true; the original hippies did actually buy into this (LSD and MDMA likely had something to do with this). Less so in the 90s-00s, but the window of "you're okay" was much wider than it has been since SJ congealed. And even SJers very rarely intend massacre as an end*, though that's a very low bar to clear.

*As in, given a sufficient stranglehold on power, the vast, vast majority of SJers do not want to massacre their enemies. Most are willing to fight a civil war (and many are willing to commit terrorism) if that's the only way to get that stranglehold, but that's a means, not an end in itself. Disenfranchisement, re-education, and institutionalised kidnapping to prevent enemy culture transmission all have significant (though in the latter two cases I'm not sure about majority) support as means to ensure permanent victory, but not massacre.

I mean, false confessions are a problem with the Japanese system, yes. Was thinking about editing that in but you pre-empted me.

Note that while they do push pretty hard, it's still AIUI hardly peine forte et dure (either in the original "food and water on alternate days" sense or the later "lol we crush you with weights" sense).

I'd point to the absurd conviction rate

The Japanese conviction rate being so high is mostly a result of two factors:

  1. The Japanese take confession cases to trial (the confession is presented as evidence), so they show up as "conviction" in Japanese statistics whereas they show up as "not a trial" in almost every other country's. Because confessions are very common, this drastically inflates Japan's conviction rate.

  2. Japanese prosecutors are actually quite reticent about pressing charges without a confession, so cases that might show up as "acquittal" in another country tend to show up as "not a trial" in Japan.

Japanese culture is indeed hilariously disgusted with criminals (the best example is probably the manga/anime Death Note, in which a vigilante who decides to kill all the criminals - but who can only kill the ones who the justice system has already caught - is presented as morally ambiguous rather than an utter lunatic), but AIUI their justice system isn't actually as vicious as you'd expect from that (note that Japan does not have jury trials, which is probably a good thing).

They lost their culture to the extent that precious little is known of pre-Roman Gallic culture and the language spoken in France is classified as Romance language, heavily descended from Latin.

S_S did also mention the Suebi, who have plenty of descendants.

I think we are functionally in agreement.

This challenge is important and demonstrates many good qualities.

Under current sexual harassment norms, you're wrong. It currently reduces to an arsehole filter; current norms are that asking out a woman who doesn't want it is evil, so asking out a woman not known to want it is, probabilistically, also evil, and only arseholes do it.

Set fire to those norms, and you'd be right.

Small-scale questions allows CW.