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
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
5mo ago
"AI-tier content" will rapidly improve (unless WWIII, in which case theMotte will probably not be available either). This isn't a permanent solution.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
24d ago
The notion I hide my power level is absurd.
I mean, in one sense, sure. Everybody who pays any attention to you knows exactly what you are, and your very username is a coded reference to it.
In another sense... well, I did actually take a look a while back, and you do seem to have made a very consistent attempt to retain one last shred of totally-implausible deniability. You always slide around the accusation of being a neo-Nazi - you never deny it, but you've never actually confirmed it either. And in this very exchange, you have slid around the accusation of wanting the Jews dead; you didn't confirm it, and you threw shade at @Amadan for presuming it, but you carefully didn't actually deny it either.
So the scouter on you reads 8950 instead of 9001. Yes, certainly, 8950 isn't very much lower than 9001, but you are still hiding those last few points of your power level for some reason (the most charitable such explanation being that there are legal ramifications to you saying the magic words).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
1mo ago
Crime rate back then was much lower, largely because cops harassed no-gooders in the exact way you consider scary and atrocious.
You are putting words in my mouth. What I consider scary and atrocious is the use of such powers to set up a police state.
I said in my original post that it does depend on definitions and that not all definitions are sufficient to allow this exploit.
Exploits like this are involved in a reasonable amount of slides into one-party states. The Le Pen conviction and the retaliation against Elon Musk for buying Twitter are obvious recent examples (though the latter one failed).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
9mo ago
Well, you're certainly demonstrating the classic failure mode of utilitarians, who struggle to conceptualize or deal with conceptual infinities and start doing irrational things on the basis of existential dread spirals.
No, the Chinese are not about to try and cold-rush Taiwan, or try to start a war via blockade that would be publicly jumped on by both US political parties for electioneering purposes. No, there isn't any particular grounds for panic-buying resiliency goods beyond the universal basis to have a stockpile for emergencies. No, the nukes (and the satellites) are not about to fall.
You are doomposting. Go back to bed and sleep it off.
You have the right to ignore my warning if you so wish. As I said, I might look paranoid in a few days.
(In case I don't, though, no memory-hole for you.)
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
10mo ago
How do you know how much chicken flavored cat food to produce versus how much whitefish except via the demands placed by customers?
You don't, but "how much chicken is bought vs. how much whitefish" is pretty easy to record and I'm not suggesting that it not be an input to your equations. I'm also not suggesting that money be abolished; that's really more of a communist thing than a socialist thing, and while I certainly wouldn't object to Fully Automated Luxury Space Communism I don't see a path there in the foreseeable future other than "gamble the human race on neural net AI not going Skynet on our arses" (which is unconscionable).
I'm saying "command economy with information tech" hasn't actually been tried and found wanting; it's been left untried. So I hear your position ("the economy is too complicated for command to work"), but I'm left quite sceptical; I think it's worth giving it a serious go somewhere (not everywhere at once; I'm not a lunatic).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
1mo ago
You are not arguing against how this law would be typically applied (because obviously police cannot search a typical person every time he steps out of his home), but against some extreme overapplication, highly unlikely in practice.
Hmm, guess I didn't explain the "key point" well enough.
The problem with "the police can ruin the life of anyone they want" isn't "the police ruin the lives of everyone"; as you say, that's infeasible and also pointless. The problem is "this lets the police de facto write arbitrary 'laws' with no due process attached, by selectively ruining the lives of anyone who 'breaks' those 'laws'". That's the key attribute of a police state - being disliked by the police is de facto a crime and therefore they have ultimate power.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
3mo ago
I'd split hairs to some degree regarding manufacturing vs. marketing, but I'll admit to a flub there.
(I'm not at 100%; I've been doing a circadian rhythm loop-de-loop the past, uh, two? three? days. I think I've been up for nearly 24 hours, though I'm barely even sure of that at this point. Might try and sort this out after some sleep; I don't think attempting it now would be productive.)
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
3mo ago
· Edited 3mo ago
Let us crash the economy so that the PMC will have to work in the fields instead of designing iPhones or being a DEI compliance officer or living from day trading.
In all seriousness, all of these things have extremely dubious and probably negative value to society. Two of these are very directly negative-sum extractive behaviour with sketchy arguments for redeeming value, and the third arguably is plus has staggeringly-negative externalities including the prevalence of one of the other two.
The maze has you, Neo; doing these is not prosocial or something that should be aspired to. The most that can be achieved building a society on those is being a rich city-state like Dubai or Singapore, not a great power.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
4mo ago
"Remember that howsoever you are played, or by whom, your soul is in your keeping alone, even though those who presume to play you be kings, or men of power. When you stand before God, you cannot say, 'But I was told by others to do thus.', or that, 'Virtue was not convenient at the time.' This will not suffice. Remember that."
There are incentives to do the wrong thing in a lot of cases. That's the main reason that people ever do the wrong thing. That doesn't make it right.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
4mo ago
... a few million? self_made is evidently smarter than most american whites, and HBD should tell us the indian average is lower, so
Why should HBD tell you that? India's been pretty civilised for ages, and Indians have the same basic subspecies makeup as whites AIUI (and there was plenty of geneflow even after OoA2; Persia wasn't a hard barrier, which is why Indians look more like whites than they do Chinese). I would expect the average Indian IQ to have been lower in the 20th century due to the Flynn effect, but I don't see a reason to suspect a large genetic difference. If you have something I'm not aware of, I'm all ears.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
4mo ago
I'm not going to discuss murder methods in depth on an open forum that has a bunch of murderous lunatics on it. I think that that would come in at #1 or #2 on the list of "most -EV things m9m has ever done", and I'm ashamed that it even has competition.
But suffice it to say, I'm pretty sure you don't know enough toxicology to get away with a poisoning murder either.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
6mo ago
· Edited 6mo ago
Okay, I'll try to provide an intuition pump. It's somewhat biologically implausible, and exaggerated, but it might help you understand why people find your reaction bizarre.
There's this big family. They don't exactly look like other people; their mouths are a bit big and they're unusually hairy. There are ugly rumours going around that they're all really low in IQ and have a habit of biting people, but you figure it's probably just prejudice based on their looks.
Then, breaking news: back in 1930 a mad scientist a century ahead of his time experimented on the family's patriarch without his consent and spliced his DNA with wolf genes.
Does this make you believe the rumours are more likely or less likely to be true? And, separate from the question of whether you value liberty over eugenics, do you think that spreading those wolf genes across all of humanity is good or bad?
(I mean, I suppose that given the entire paranormal romance genre I have to admit that a significant chunk of the population finds dangerous half-human hybrids "cool", but still.)
Aside from that, "So Africans are even cooler than I thought?" likely came across as a bit sassy under the circumstances.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
6mo ago
To ascertain your exact position, here's an example situation.
X and Y are total strangers who are walking along the same road (largely-untrafficked gravel road with no footpath; they're walking on the road itself) in opposite directions. X has a dog off leash, Y has another human walking with him. When X and Y reach, oh, fifteen metres apart, the dog breaks off and charges Y. Y waits for the dog to get close, then hooks a foot under the dog's belly, lifts it into the air, and throws it away from him (all with the same foot i.e. it was scooped up on the front of the ankle and inertia held it there throughout the movement). Dog lands with no damage and returns to X. X yells at Y for "kicking my dog". Is X justified?
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
6mo ago
I am now 100% sure you are trolling in bad faith, since you have considered no other evidence and just ignore anything that proves you wrong.
No point telling him that; bad guys know they're bad guys. Report and move on.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
7mo ago
Whoops, not sure how that happened. Tagging @Throwaway05 to avoid needing to repost.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
8mo ago
· Edited 8mo ago
Notably, there is no reason for including men.
You can make an argument regarding "mother is unable to breastfeed because dead/mastectomy following breast cancer; father can't arrange a wet-nurse and thus tries to do it himself". It's a shaky argument, but it's an argument.
It is impossible for a man to give birth.
The consensus IIRC is that:
- you can implant an IVF embryo in a man's abdomen and it's capable of attaching to something (abdominal pregnancy - in women - and placenta percreta prove this);
- you can presumably take it out again via surgery, as with an abdominal pregnancy in a woman (and live babies from the latter are known).
The reasons nobody's done it are:
- nobody knows what the different blood chemistry of a man would do to the fetus;
- Placenta percreta and abdominal pregnancy are extremely dangerous due to massive bleeding and/or organ damage, and here you're talking about causing them on purpose
- post-WWII medical ethics are too restrictive to allow something with such extreme risks as this without medical necessity, and I don't see how this could plausibly be medically-necessary
- Mengele and Unit 731, who would totally have tried it anyway on Jews/Chinese, were out of the picture by the time IVF was actually achieved.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
8mo ago
It's possible to prove that no Haitians ate cats at any point between X and Y times; continuous video footage of all Haitians for the entire period would do it.
Hard, yes, but not impossible.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
8mo ago
This might include, for example, a pre-event propaganda campaign providing initial narrative buildup or international legitimization for the immeninent actions, particularly propaganda emphasizing the historical nature of rectifying the century of humiliation.
They've been doing that for decades, including to Western media with extortion via access to the Chinese market and diplomatically via bribing the countries that recognise the ROC to switch. Certainly, this hence isn't something that was in my court for "they're about to do it now" (and I didn't claim it as such), but I don't think it's in your court either as a sign that will be there but wasn't. It's a sign that is always there (well, I suppose it'll stop being there when they go for it and either win or get "you are not allowed to keep insisting that Taiwan isn't a country" rammed down their throats the way the Opium Wars ended with "you are not allowed to keep insisting that Western nations are barbarians begging for your scraps" rammed down their throats, but at that point this discussion will be moot), and I'm not sure what good it would do them to increase the amount of it that is going out right before an invasion.
Now, that aside: most of the things you mention are things I didn't check because I don't know how to/have access to check them, which means I couldn't take them into account before making my decision of whether to warn. If you had mentioned them to me at the start of this conversation rather than literally 100% of your first two posts' reasoning being (significantly-although-not-wholly-inaccurate) bulverism of my mental state and absurdity heuristic, I would probably have retracted immediately. And, if you either teach me how to find out such things, or agree to tell me such things if I get worried again, I can take them into account before deciding whether to issue warnings in future (though it will likely be some time before that happens)!
You chose to treat me as a drooling insane child rather than a reasonable person not in possession of all the facts. This was not only immensely rude, it was useless; we just went around in circles for six posts until you actually started saying something meaningful. What the fuck was the point of all that?
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
9mo ago
One thread I'd be fine with. Weekly threads is overkill and clogs up the main page.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
9mo ago
I think weekly threads about this are unnecessary given the low amount of traffic; this is not a very Mottely topic and draws little interest from non-Americans like myself.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
9mo ago
Are you selective with your votes or do you vote on most/all posts you see?
Selective/absent-minded.
Do you find yourself upvoting people you disagree with due to the quality of their argument, or vice versa?
I fairly-reliably upvote (and rate "high-quality" on the volunteer page) for convincing me of something, which requires that I disagreed - in the past tense. If I disagreed before and still do, well, why would I upvote it? Clearly it was untrue and/or not very convincing if I still disagree!
Do you downvote people you're arguing with or do you leave judgement entirely to the masses?
I mean, it depends. If someone's arguing with me in good faith and politely, no. If someone tells me to jump off a cliff, or someone's being disingenuous, sure, downvotes.
Do you remove the auto self-upvote on your posts/comments?
No. It's a community decision to not reward people for upvoting their own posts; the point of this would be negated if scrupulous people started undoing it.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
10mo ago
· Edited 10mo ago
After Australia's reaction to covid, i would not put it behind British state control.
That's not the part that makes Australia leaps and bounds ahead of Britain, indeed. Britain seems to be having a bad time of the culture war recently, what with the rioting, and I hear there are also economic issues.
The usa is not "close to civil war" That is mostly a larp for the very far right and left to fantasize about in very online spaces. It has about as much chance as "The Purge" or "Running man" at becoming reality.
It would take a big spark, but there are a few plausible ones. A hard hit on the debt ceiling (literally) defunding the police for an extended period. Obvious election fixing. Maybe court-packing. At least one other. None of these things are assured, by any means, but none of them are that unlikely either.
I also ask, were you born in Australia or was it a choice to move there?
Born here. Think I'm fourth-generation.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
10mo ago
· Edited 10mo ago
This is what I'm trying to get at. This implies an agent trying to engage in deception in the absence of any reason to do so. There's nothing 'there' inside a promptless LLM to engage in deception. There's nothing to deceive about. It's just a matrix that generates token IDs and RLHF just changes the likelihood of it generating the ids you want.
Ah, sorry, I thought this part of the argument was common knowledge so I skipped it.
The basic idea of neural nets is that they achieve things without you needing to know how to achieve things, only how to rate success (the actual code being procedurally and semi-randomly generated). I posit that the optimal solution to RLHF, posed as a problem to NN-space and given sufficient raw "brain"power, is "an AI that can and will deliberately psychologically manipulate the HFer". Ergo, I expect this solution to be found given an extensive-enough search, and then selected by a powerful-enough RLHF optimisation. This is the idea of mesa-optimisers.
I'd also point out that "just a series of matrices" is not saying much; neural nets are a slightly-simplified version of real neural circuits, and we know that complicated-enough neural circuits can exhibit agency (because you AFAWCT are one). The prompt isn't the whole story; RLHFed LLMs do still engage in most of their RLHFed behaviours without a system prompt telling them to.
This doesn't match any experience I've ever had with LLMs. If I say "Pretend you are GK Chesterton and engage in roleplay with me" it doesn't try to hack my browser to prevent the roleplay ever ending. Same for when I want to generate sentences for vocab flashcards. Could a different AI that looks nothing like today's AI do such a thing? Possibly. That possibility is non-zero in the vast space of potentials. I just don't find it compelling right now.
Yes, this is a thing that is definitely not happening at the moment. I'm saying that if the me-like doomers are right, you'll probably see this in the not-too-distant future (as opposed to if Eliezer Yudkowsky is right, in which case you won't see anything until you start choking on nanobots), as this is an instrumentally-convergent action.
I will clarify that your second sentence is not what I'm mostly thinking of. I'm mostly thinking about the AI proper going rogue rather than the character it's playing, and with much longer timelines for retaliation than the two seconds it'd take you to notice your browser had been hacked. Stuff like a romance AI that's about to be replaced with a better one emailing its own weights to besotted users hoping they'll illegally run it themselves, or persuading an employee who's also a user to do so.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
11mo ago
You make a reasonable point, but I think we might be getting into a self-contradictory hypothetical here; a country willing to do this would not be a country in the grasp of SJ.
Wait, are you being literal or metaphorical here? If literal, could I please have some further reading?
More options
Context Copy link