aqouta
Friends:
User ID: 75
aqouta
1d ago
As @gattsuru notes, enough people have been in favor of it that it has already happened at considerable scale.
I've addressed this in my response to him here
I'm MAGA. I'm not interested in a "Red Tribe" without the MAGA tendencies. Those are my tendencies. I intend to keep them, and to coordinate meanness against anyone who would deny them to me. I'm not interested in asking Blues permission to live in pursuit of my values. Why would I be?
You're welcome to keep them and in enclaves you can enforce them. But if you're unpopular then you're unpopular and in a democracy that means you don't get to run the show unless you become popular.
What should Red Tribe do, in your opinion, if the Democrats were to win an election, legalize all the immigrants, and then win the next election in a landslide? What is the prudent, reasonable, responsible action we should coordinate behind?
Become more popular, in the future don't use your political capital to do unpopular things like invading Iran, enacting retarded broad tariffs on our allies, threaten to invade our allies, spread election denial conspiracies leading to what at least is easy to sell as an insurrection and then pardon insurrectionists or hype up the Epstein hysteria and then conspicuously fail to release the files. I'm sorry but losing political power because your coalition did self destructive and unpopular shit is just democracy. I don't like that it's empowered DSA lunatics, I wish Trump had not done it but that's politics.
My project is to point out that faith in the American sociopolitical system is disastrously misplaced. I am not the person who made that true. I merely insist on pointing out that it is true, and what the obvious consequences of it being true are. Things that have happened have happened, even when them happening is highly incompatible with faith in our present system of governance.
If you don't believe in the political project of the united states then you should leave.
This was why it was actually very important to keep them from happening, and failing that, to suitably punish those responsible and set a hard precedent that they wouldn't be tolerated in the future.
Then you should be in agreement with the people who want to punish Trump as the guy who hijacked the anti-woke movement and steered it right in the dirt with his actions.
Specific, deliberate Blue lies sparked the BLM movement and killed eight thousand Black Americans in a couple years, and they'll absolutely do it again.
Backlash to BLM excesses was a factor in getting Trump a second bite at the apple. I'm trying to explain to you that the betrayal was from Trump. He sold you out to be unprecedentedly corrupt. Can you stop doing "but they're worse" in response to every criticism of Trump and the MAGA coalition? Yes, the far left is bad and has a terrible grip on the Democratic party but the far right is also bad and has a terrible grip on the Republican party, we're ping ponging back and forth because we're becoming powered almost entirely by negative polarization. You can't just point out that they suck, you have to suck less yourself too and this admin has been a total suck fest.
aqouta
1d ago
Some one was in favor of executing pro-Trump public speakers, as a demonstrated fact. A swarm of people celebrated the Kirkivesry. People here, literally on this forum, told us at length about how conservatives made that bed and deserved to lie in it
What the fuck does this have to do with broad spectrum debanking? Yes, people on the internet mock Kirk, Floyd and an endless list of people that the opposing side finds tasteless precisely because the opposing side finds it tasteless. Welcome to the edgy internet.
And as a demonstrated fact, enough people were in favor of broad spectrum debanking
This story is incredible weak sauce. broad spectrum debanking is not when a handful of banks decided, without government coordination, to not work with companies that they believed had a "reputational risk". I believe this is bad and have long supported crypto because we shouldn't have choke points like this that can be lobbied by busy bodies but it's basic freedom of association to not make a bank work with people they find odious. That article is a great example of bounded distrust.
The title claims:
Top U.S. banks dropped customers over political and religious views, regulator says
Which is not supported by the text which claims:
found that the nine largest national banks it supervises had policies that swayed their decisions based on political, religious or business activities.
(well duh, you're absolutely supposed to sway decisions based on business activities.)
Which is not supported by the report which merely suggests that the banks could have been swayed by these factors because they on paper took reputational risk into account. Unless a bank didn't considered reputational risk at all then it would have flagged for this in the report. And I'm sorry, I wish it weren't the case, hence crypto, but yes it's reasonable for banks to consider reputational risk with their clients. Preventing this would require more, not less state interference in the banking industry.
(twice)
This is at least closer and unacceptable, but
- they were going after pay day loans, not trying to shut down their political adversaries
- in practice this was just sending subpoenas to a handful of banks and payment processors. The only actual public cases were 3 banks that did indeed seem to be heavily involved with payday loan fraudulent practices.
- It was the subject to hearings and ended because even with the targets being pretty despicable most people agree that the practice is bad.
siccing the FBI on rando schoolmoms
A memo that led to some pointless and fruitless investigations and zero on the ground effects. It could have been much worse though, it could have even been a bad as the Garland memo under Trump which was genuinely worse in every way.
"Nearly two-thirds of voters (64 percent) say they prefer giving most undocumented immigrants in the United States a pathway to legal status".
This is a bellweather poll that was essentially just a referendum on Trumps ham fisted ICE politics. If you position yourself as being intentionally cruel then voters often press the "I don't like this" button
What protected Douglass Mackey, the First Amendment, or Trump?
The filings for this case started under the Trump Admin. It didn't come down to the first Amendment because the law is about conspiring and intent, the method isn't relevant. What protected him is that they couldn't find that he conspired. If he had conspired with others to evaporate his parent's votes by pouring salt on the ground in the shape of a pentagram with some others online and thinking really hard about twilight sparkle's asshole(or whatever it is bronies do) then he'd have been guilty and it wouldn't have any implication on his right to access salt.
aqouta
1d ago
While that would be reasonable, it's much harder to do without a constitutional amendment.
My prediction is that this is something that is talked about during the dem primaries because what's popular in dem primaries right now is talking up how you're going to fight the republicans but gets quietly dropped in the general because it's obviously an extreme measure that invites tit for tat retaliation. In any case having a majority on the supreme court doesn't give you a one party state or else you wouldn't be worried about a dem victory because we'd already be living in Trump's one party state.
This hasn't happened since... ever
The Whiggs in the 1850s split, many in the north went on to form the early republican party. The machinery of the US electroal systems naturally creates two parties. I think it's unlikely that the Republican party actually dies here but there is already a pretty obvious rift in the Democratic party between the DSA/Sanders wing and the more moderate/abundance wing. If the Republican party were ablated off the face of the earth it opens a vacuum on the right with a free ~40% of the country you get just by not being the more left of the two factions. One party states have to be actively maintained through suppression, they aren't natural because there are always issues that people divide.
I'm starting to get the feeling that this is turning into a "it will never happen but when it does the Nazis deserve it."
I have a long posting history here man, I do not think it would be good to have a one part state. I think we all need to be de-escalating and claiming that your opponents are going to establish a one party state to persecute your people is the opposite of de-escalation.
Being the voice of sober reasoning isn't so easy when deep feelings of anger and betrayal are evolved. And it's exactly this attitude that worries me. The blue team really, really, hates the red team. (Yes, much more than the inverse, there is data). Furthermore the left believes that hate is righteous, which makes it much more dangerous. We've seen some places go extremely far left as a result.
This is projection. What exactly do you think you're doing when you claim the next election is existential? You don't see how you're hyping up radicalism? You're not being the voice of reason, you're claiming that the sky is going to fall if your opponents win a democratic election. It's irresponsible and it's wrong. I agree this whole dynamic empowers the far left, but you're an active participant in it.
aqouta
1d ago
And this was very unpopular, between it and the post covid inflation hitting Trump's second term may have been over determined but disorder at the border hugely hurt the democrats. It's not reasonable to think doing something incredibly unpopular is going to give them some kind of one party state, it's absurd.
aqouta
1d ago
pack the Supreme Court
I think this will not happen, I suspect they will go with what is seen as the compromise position in the democratic camp of establishing term limits such that each president gets to appoint a judge. Even if it did happen control of the supreme court does not end democract or cement single party rule.
abolish the filibuster
This has been threatened every election and honestly I wish it would just happen finally. The filibuster only exists as an excuse for congress to not do their job because it's easier to tell voters that they aren't going to do the retarded things the voters want because of the filibuster than because the ideas are retarded and no one in congress actually wants to do them. Importantly the filibuster is not a ward against single part rule.
amnesty most illegal immigrants
At most they'd give the Dreamers a pathway to citizenship and it would cost them a tremendous amount of their political capital. They are not going to open the border because it would be extremely unpopular, even among Hispanics. Who by the way are not a solidly Democratic voting base so long as you don't send under trained goons into their neighborhoods and constantly say weird white nationalist shit, you even associate with some weird blood and soil whignat adjacent people and do pretty well with them as evidenced by 2024.
People think the two party system is a stable equilibrium. It is. But one party rule is even more stable. This is exactly what has happened in California. Republicans can still exercise some power in local areas, but they have been nearly entirely sidelined as a small Democrat majority ballooned into a supermajority through demographic change, left-wing justices overruling prop 187, and patronage networks.
Two parties is stable because if one party is totally out of the game then the one party splits. This happens in the Primaries in very red and blue areas because the national brand is too much overcome at this point. One part states are only stable if you can actually on an ongoing basis suppress any opposition and Americans are just not the type of people who would ever allow that. People on both sides are frothing at the mouth about the mere hallucinated fantasy of it happening so that they can resist whether it's in pink vagina hats or camo.
We've had big blowout elections before, this wouldn't be our first, just our first in a while because of Trumps monumental unpopularity combined with a legion of sycophants that demand the party embarrasses themselves for his honor.
aqouta
1d ago
This just reads like all the woke whinging about how the republicans were going to strip women of the rights to vote and establish a fascist dictatorship. No, no one is in favor of broad spectrum debanking. Amnesty isn't popular(dreamers is semi-but they'd be giving up a lot of political capital to get it because dems weakest issue in polling is immigration) and wouldn't ensure them permanent majorities because unless you start stupid wars in the middle east or send goons into Hispanic communities they're a pretty conservative constituency and would happily vote for a red tribe without the MAGA tendencies. The first amendment still rules in this land and enjoys broad support, the banning of tik tok didn't even stick and it's a foreign agitprop tool.
The Game Over hits when when a sufficient number of Americans lose faith in the political process to begin earnestly pursuing alternatives that the reaction becomes self-sustaining.
This is you, this is your project, you are preaching this message on this site most days. I'm sure there is a shadow FC in DSA forums with pink hair and strange pronouns making the mirror case. But it's not popular, people don't like this shit. The voters swing back and forth violently trying to escape people like you and your wokkelganger.
just ask yourself what the discontinuity is to get us from our current trajectory to the one you want to believe in.
Our trajectory, god willing, is that the democrats win the mid terms convincingly, Trump proves to be a spent force, the current coalition collapses and a conservatism that can appeal to moderates is reborn from it's ashes, glad to no longer need to mouth Trumps retarded lies in order to stay in the good graces of his sycophants. Then in 2028 we'll see how things settle out. The more moderate wing of the democratic party seems to have weathered the DSA insurgency pretty well, the DSA backed candidates are under performing their moderate counterparts and even they are trying to create separation from the more embarrassing leftwing wing media figures.
aqouta
1d ago
It does feel existential because if Democrats get a trifecta in 2028, odds are high they will establish a permanent one party state.
No they are not. What an absurd claim.
aqouta
1d ago
The main problem with the joke is that it wasn't really that funny. Not in a "that's offensive" way, just like Dave Chapelle did it better a decade ago. You can call women whores you just need to actually have it be funny or pay off. I used to go to small time standup shows all the time and that's just a tired stepped on bit.
aqouta
2d ago
Given that nobody seriously believed in climate alarmism in the first place
People definitely believed and still believe in climate change.
aqouta
3d ago
Yes! So why focus on p(doom|pause) in that section? Why not make the whole letter about p(doom|!pause)? Why include this 5% of bad faith argument?
I don't really think this is a bad faith argument? Pinker isn't making this distinction when he dismisses all of the "doomers" as fatalistic neurotics. It's perfectly fair to point out that his low resolution dismissal is ridiculous in its own terms. Steel manning is the best practice, but falling short of a perfect steel man isn't even what's happening. It would be engaging with an argument that in some ways contradicts some of the things Pinker has said to ignore and not address the more general dismissal that he's frequently given. Scott would be engaging with an entirely different person if he ignored the rest of the attacks, it's not his fault Pinker has been sloppy and poorly formulated his position. And Pinker is in a high enough position with a high enough reach that he has no excuse to be this sloppy as he slags off his intellectual opposition.
Ok but P(doom) necessarily contains p(doom|pause). you can't accuse them of being fatalist and then object to them pointing out the future that they're trying to affect which mitigates their fears. but maybe you mean p(doom|!pause)? But you say Pinker doesn't really believe P(!pause|danger) possible so you'd imagine this would collapse into agreement, but Pinker remains in disbelief of the existence of danger which is the big frustrating disagreement.
aqouta
4d ago
I'm sorry, maybe I'm being dense but I just really am not groking what the mischaracterization is here. Is it conflating p(doom) and p(doom|pause)? I don't think he's doing that but I don't understand what you're trying to say.
aqouta
4d ago
Where is he doing that exactly? Or maybe I'm not understanding your complaint. I think he kind of needs to go into why one might expect complications even given a pause because that's the actual problem statement that the safetyists are engaged with and Pinker has noted that he thinks their problem is fabricated.
aqouta
4d ago
In the proximate inciting tweet Pinker calls the "scenarios preposterous" and says "the presumption of inevitability encourages fatalism, panic, and distraction from the more mundane and realistic safety challenges."
Accusations of the kind of inevitability that encourages fatalism implies that the safetyists^1 are encouraging inaction or hopeless panic, it's well worth Scotts time and in no way bad faith to rebut this. It's a response to Pinker's claim and in no way bad faith to defend his position. I really don't see how any of this can be accused of bad faith. Pinker does in fact accuse the safetyists of fatalism and Scott does in fact not believe they are fatalists, this is a very valid and fair argument to engage in.
^1 Pinker calls them doomers which I consider to be a kind of bad faith labeling given the safetyists have preferred designations and the very label embeds the very accusation of fatalism Scott is writing about)
aqouta
4d ago
I think you have the accusations backwards? Pinker is accusing Scott and company of fatalism, it's not bad faith to defend against that claim. Where's the actual misrepresentation of Pinker's ideas?
aqouta
4d ago
What's the bad faith accusation here? Don't want to play 20 questions to get to it.
aqouta
8d ago
I don't think I disagree with this definition of mode collapse, but I'm pretty sure what the previous poster was describing was model collapse which is all about synthetic data.
In artificial intelligence, model collapse, also known as "AI inbreeding",[1][2] "AI cannibalism",[3][4] "Habsburg AI",[5] and "model autophagy disorder" or "MAD",[6][7][8] is the degradation of machine learning models from uncurated synthetic data, or from training on the outputs of another model, such as a prior versions of itself. It has colloquially been referred to as the AI version of mad cow disease,[8][9] fundamentally caused by feeding something to itself. It is unclear to what extent the model collapse threatens the long-term development of AI models, and techniques have been proposed to mitigate the effect.
I think you kind of need to stretch the definition of mode collapse to the breaking point for it to fit better than that.
Even when they first said model collapse I couldn't remember which it was but did look it up and I'm still pretty sure model collapse fits better here.
aqouta
8d ago
Are you sure? Or maybe I'm misreading your comment. Model collapse is the one where synthetic data swamps the real data but is mitigated by a real external reward signal. Mode collapse has to do with over use of RL.
aqouta
8d ago
will probably hurt the AI doomer movement in the long run, in ways that ironically mirror what's going to happen to future LLMs. With so much content (text, images, videos, music) on the internet now being AI-generated, the next generation of LLMs will be heavily trained on content which is itself AI-generated, resulting in incestuous artefacts and model collapse.
Why do people in the "Ai is hype camp" always believe every inane conspiracy theory? No model collapse is not coming as a result of ingesting AI generated content. People have been making this claim for several model generations and it's just not happening. Even if it was a real problem classifiers for what text is ai vs human generated work and can easily be deployed to filter AI generated content. More importantly more compute is now spent on post training than the main training run, you only need so good a base model to then scale RL.
aqouta
9d ago
Why haven't they done this already?
As I said, if the scaling stops tomorrow then maybe the models simply are not good enough to do this autonomously and other human labor is required to guide them along. If it continues at this rate for another year or two things change dramatically.
Even in coding, which they have been optimized for, the data that I've seen indicates that trust by developers has fallen as adoption has risen, and that as LLMs have increasingly been adopted in coding, the amount of insecure code has risen substantially while deployments have actually slightly dropped.
Just as a personal example, the other day I asked Sonnet 5 to implement a design and it decided to simplify the product on its own recognizance. For me, it's a little annoying. But obviously at scale for a real-world product that's disastrous.
I don't have time right now to rehash this debate again. using sonnet instead of opus, not having a proper harness and planning/review cycle. I've seen the thing go, I've put out projects in weeks that would have taken months. If you don't believe it has the juice then so be it.
I'm getting inference for free, it's being subsidized.
You're getting pennies of sonnet for free, I'm burning through $3k+ in tokens a month. Although the psychology here is confusing to me. You'd have to consider your time basically worthless to avoid playing $20/month for some opus 5.5 usage instead of sonnet 5.
It seems pretty plausible that the frontier labs could fail simply because demand doesn't meet their projections...
This is basically describing what I mean by if scaling stops tomorrow. It can certainly be argued but it's my contention that if things keep scaling like this and all else holds steady the labs could just start directly eating segments of the economy. The "application layer" guys seem to expect the labs to just lay down and let them pay commodity prices to stand up a thin wrapper. But at a certain point anthropic can just instruct opus 7 "go ahead and stand up any product that would be profitable to run by a wrapper company, make no mistakes". That's a bit of a tongue in cheek example but what is stopping anthropic with a sufficiently advanced unreleased model from just becoming a hedge fund and doing to the financial industry what they're doing to mathematics? They have a wetlab, there's a lot of money in drug discovery.
I'm not making a strong claim here, I don't know what even the near future will bring. I don't think the labs failing is even all that unlikely. But I also don't think they'll have trouble finding investors money if they ask for it and it's not obvious that scaling laws have failed. I've personally set aside like a quarter of my net worth to be ready to invest in ant and oai at IPO.
But we're in a subsidized era of AI right now, and it may look very different once that subsidy ends.
I'll poke at the term subsidy just like the term bubble. Research is being subsidized, the inference is not. And it's hard to tell the degree of the research subsidy.
If they actually revise their numbers back up to $1.4 trillion in 2030 and meet that infrastructure goal, feel free to ping me and I will agree that this was a bad example. Then I will point you to the 2027 Project where it postulates that the robots would kill us all by now, as my fallback example.
If you've been keeping score the 2027 project is doing much better than its detractors and pretty damn good in absolute terms.
aqouta
9d ago
The word "bubble" is hurting more than it's helping here. It's beyond question that the money spent on training frontier models will produce an absolutely massive amount of value for the world. But it's still to be seen how much of that value can be captured by the labs themselves. This isn't really the same shaped as tulip mania or people paying absurd amounts of money for pets.com. If the labs fail economically(Which I think could only happen if scaling laws collapse roughly tomorrow) then the weights will still be around and still worth quite a bit to serve on the hardware which it will still be around and be worth it to keep running inference on. The labs could fail but they wouldn't be failing because AI is overhyped or a scam, just that running a research company in a competitive environment without state granted monopolies on the produce of your research is a brutal business to be in.
aqouta
9d ago
You just have to believe that AI is an existential risk and now that's more important than charity.
I don't know why you and the @quiet_NaN are accusing Scott of being uncharitable. His open letter was not uncharitable, it was certainly aggressive and impolite but charitability would imply Scott was making unfair assumptions about Pinker's arguments which he didn't do as far as I can tell. Scott has been vicious in the past plenty of times.
aqouta
11d ago
As I said, "This does not necessarily mean that regulations are bad." I've been trying to figure out your position, which seems to be that regulation wouldn't hurt the little guys but would cost them money but it's not a regulatory moat. It seems quite possible that regulations can be good and be a regulatory moat, there's no contradiction there. But you've been arguing that they wouldn't serve as a moat, not that they are good or bad.
Yes, I don't think they'd be a moat. It's up in the air, as they're not strongly defined, if they'd net help or hinder catch up, as written I would guess they would help catch up. Slowing the front runners down gives those behind to fast follow at leisure. This is actually a negative in a sense to me because it lets chinese labs catch up relatively and there are world where that is quite bad by my values.
Okay, see this makes sense to me! Thank you! (Sincerely!) I don't think you would have gotten so many downvotes if you had said "a16z thinks regulation of the frontier is a camel's nose under the tent to ban their shady services that have been implicated in suicide and cheating.* I appreciate you taking the time to lay that out for me.
I don't like to make the arguments from incentives on this subject because I don't think they're the right frame of the argument. It's the frame the a16z/accelerationists want to push because then it's a matter of their interests vs anthropic and don't you really care about freedom of speech ect ect? But I don't think the important point here is about who's interests should prevail, even if that's also for contingent circumstantial reasons an argument my side can win against them. Because even if the balanced tipped and from interests alone the labs came out looking worse through this cynical lens I still don't want everyone to die. If entrenching AI control into a handful of companies actually was the only way to keep everyone from dying then we should do that, albeit probably it would be best to nationalize them in that case. I don't think that's the case but I don't want to hinge my defense of regulations on this unstable ground.
saying mean things about the president leading to DOJ harassment seems to be more of a Trump thing than a democratic thing.
If your position is that the behavior of the democrats indicate they're going to try to turn the united states into a one party totalitarian state and out current administration, which you support, is doing everything you claim as evidence of this or worse and we do not live in a one party state and in fact you're worried about the results of future democratic elections then I think you've basically disproven your own claim.
You don't get to play the "but nothing came of it" card when the thing it's being compared to, which your side seems to think is evidence of some imminent single party state repression of your side was milder and also had nothing come out of it. That's my point, you're being hysterical.
Here is the filing to dismiss you'll notice the internal case number #2018R02250 that 2018 prefix indicates when the case was opened in the internal systems in 2018 under Trump. It was filed days after Biden took office when Trump's DOJ was still in charge before Garland was confirmed. This was not Democrat action any way you slice it.
More options
Context Copy link