@quiet_NaN's banner p

quiet_NaN


				

				

				
0 followers   follows 0 users  
joined 2022 September 05 22:19:43 UTC

				

User ID: 731

quiet_NaN


				
				
				

				
0 followers   follows 0 users   joined 2022 September 05 22:19:43 UTC

					

No bio...


					

User ID: 731

Which would be much more upsetting in a country where Warren v DC was not a thing.

saying that Republicans want a "redo of the Civil War," amongst other incredibly inflammatory things.

Oh no, the perfidy of the woke left truly knows no bounds. And to attack such an upstanding citizen as president Trump, who started his political career with his very nuanced ad about the central park five (about whose guilt he was factually wrong, sadly), based his bid for the presidency on another unfortunate misunderstanding of his and proceeded to win the hearts and minds of Americans by always maintaining decorum and treating his political opponents with respect. Always a voice of moderation and compromise, as well as a great husband and fine human being and an upholder of the highest epistemic norms.

Let me be blunt. Falsehoods are always bad, but if there is one party which has forsaken the high ground here, it is Trump's party. Given all the shit Trump has been spewing over the years, I would not particularly upset on his behalf if the Democrats were to spread a rumor that he has an Olympic swimming pool filled to the brim with the eyes of murdered babies in which he likes to go skinny-dipping with his cabinet.

Besides, "Trump wants a civil war" is far-fetched, but not maximally far-fetched. There is a notable community of preppers and 'militias' for whom "another civil war" has long been a favorite masturbatory fantasy. (Of course, they did not expect to fight on the side of the federal government!)

It is established case-law that the duty of the police to protecting individual citizens is fuck-all. I do not know if relevant local or case law has decided if local police forces owe any service to the feds, but I would default to "no".

Letting the BLM riots happen was actually bad. Deciding that you have more urgent police priorities than helping ICE, which Trump likely ordered specifically into Chicago to punish the people who voted against him, and whose whole mission is to score cheap political points in a rather farcelike manner -- "we get rid of all the illegals, except for the ones in the hotel sector (where Trump is involved) and the ones in the agricultural sector (whose deportation would make the food prices skyrocket even more)".

Why should the local mayor lend Trump the PD for his political stunts? Let him at least waste federal funds for it.

For some reason sending in the national guard is really helping me frame the Democratic response as what it is - basically outright treason against the U.S. federal gov.

Are you saying that you know that the Democrats are treasonous because Trump sent the national guard to deal with them? Then DC must have turned treasonous already weeks ago!

Here is my take. This is a clown-show. Trumps masked goons try to kidnap illegals to help him score political points (and own the libs). I imagine that local PDs will in turn try to hamper ICE as much as they can. Perhaps their unmarked cars get towed while they are illegally parked mid-arrest, or they are subject to frequent 'random' traffic checks. This is probably likewise not the best use of police resources.

I can not speak for the random people Trump gave a bonus, a badge, a mask and a gun to act as his muscle, but my priors are that both the national guard and the local PDs really really do not want to shoot at each other. If they clash about specific questions which enforcements of local strictures which just so happen to impede ICE are allowed, both sides will refer to the court system to figure it out, and the court system will do this in very short order. Few national guard commanders would be stupid enough to trust Trump to pardon them if they break the law in his name, and approximately zero police chiefs have any delusions about defeating the federal government once the courts have decided in its favor.

At the end of the day, this is mostly a pointless dick-measuring contest.

[The Trump movement is] just soft liberalism with a lot of bloviating

Trump is neither an economic liberal (i.e., a libertarian), as he has a raging boner for tariffs, nor a social liberal, as he cuts down medicare and the like.

Even previous Republicans with impeccable right-wing credentials like George 'Waterboarder' Bush have refrained of sending the national guard into cities which had dared to vote against them.

Ethnic replacement was a winning strategy and the only the Democrats need to do is wait.

Yes, not only are They doing the Great Replacement, but also they have picked immigrants which will reliably vote for the Democrats for the next 1000 years. Everyone knows that Latinos have the commie gene, after all.

In the real world, things are different. Latinos are often strongly Catholic and have views on abortion which are roughly compatible with the Evangelicals. And Muslims are likewise sex-restrictive. If not for some ancient beef with the Christians (and the ME conflict), they would vote for whatever party proposes porn bans, which tend to be R.

Also, in a two-party system, both parties will adapt until they are seen as a viable alternative by the median voter. For example, neither party is campaigning on repealing the 19th because that would be immensely unpopular.

Finally Michigan is tackling all the mandatory porn viewing, which is the 2nd largest human rights issue after the mandatory gay furry sex in the state. Oh wait, both porn viewing and gay furry sex are entirely optional and voluntary.

We allow adults a lot of behaviors which are unlikely to lead to good life/health outcomes. A 21yo can drink, smoke, watch porn 8h per day, ride a motorcycle, join a gang, enlist with the marines, have a kid, buy a gun despite suicidal ideation, persist on a cyanide-rich diet, reject Jesus, get a swastika tattoo on the forehead, etc.

The basic concept of being an adult in a free country is that people are allowed to make decisions even if others think that they are bad decisions. Society should only become involved when your actions threaten others. This is why DUI is a crime -- if you want to drive blackout drunk on your own racetrack, we should just fine your estate for any emission violations if you burn down in your car. Sure, we have some unprincipled exceptions (drug criminalization) which went just fine (the War On Drugs will be won any day now), but mostly we let adults do their own thing.

--

The other thing which irks me is the presumption that porn leads to an inadequate equilibrium.

Before porn was widespread, a successful 30yo married guy who was into 19yo's had basically to suck it up. Just kidding, he would just woo some 19yo into being his lover. Because most 19yo women were very unwilling to just be his fuckbuddy, this would require extensive emotional manipulation, often leading to predictable heartache, unwanted pregnancies or the destruction of his family.

In general, I think the amount of suffering the average male sex drive has caused for either himself or the people near him has been quite high, historically. Most of the gender difference in sex drive was filled with coercive sex work, malicious seduction, rape, or else with sexually frustrated men -- which are not great for stability of society, generally.

Technology has solved this problem. Today, that 30yo married guy can just bookmark the teen category on pornhub and jerk off whenever he feels like it, with about zero of the negative externalities of having an affair. As a porn viewer, I have to say that as far as handling your sex drive goes, porn is pretty great. Sure, good sex would be better, but given gender preferences, the ability to have great sex with another human whenever you feel like it will not be available to most men, nor was it available historically. Basically, a world with porn is a world where sexual frustration is capped at some moderate level, even if you are into weird stuff which is infeasible in the real world.

While I am sure that the current dating market is a local equilibrium quite far from the Pareto frontier, banning porn to get people to have more good sex seems about as plausible as burning down the institutions of the bourgeois state in the hope that they will automatically be replaced with better communist institutions, which is to say, my hopes that a country will randomly stumble in a better equilibrium are very slim indeed.

I do believe that Joe earnestly tried to subvert and destroy the country.

The same thing we do every night, Kamala: trying to destroy America.

I think it highly implausible that Biden was trying to destroy the USA. If his goal was to turn the US into a failed state, he did a rather terrible job. And why would he want that?

Different people have different visions for their country. Some want a capitalist heaven, or a commie utopia. Some want to support Israel, some want to support Israel a lot. Some want universal healthcare, some want Roe overturned. Some want do get rid of background checks, some want the 2A reinterpreted so it does not apply to any firearm innovations made after it came into force. Some want to turn a blind eye to illegal migration, some want to deport every last illegal (except for the ones which keep the economy running). Some want to bomb country A, others want to bomb country B.

From where I stand, the general course of US politics has been pretty consistent from Clinton to Biden. There were always big donors whose interests got special consideration beyond the interests of the American people, mud-slinging during campaign season, use of office to get a political advantage over the opposition, from photo ops to politically motivated investigations, bombing of random places. Both Trump 1 and Biden were particularly uninspired, but for the most part it was just business as usual.

The peaceful transfer of power is one of the greatest selling points of democracy. Trump trying to mess with that was by far the worst thing he did in his first term.

Crucially, he did not get convicted because the court system (the SC in particular) stopped it, despite the wishes of the Biden administration. Trump getting convicted would in my mind not conclusively prove a kangaroo court, but him getting immunity proves reasonably well that the courts (or at least the SC) are independent of the political Zeitgeist.

Some among them, surely. The Qanon true believers who think he is fighting a secret war against the adrenochrome-addicted establishment.

Still, my model of the median Trump voter is that they know that Trump is bullshitting all day long and corrupt AF. But the establishment hating him so much makes it all worth it.

In particular, I would argue that outside your odd lizardman, none of the smarter MAGA people believe the narrative. I think it highly unlikely that Charlie Kirk thought, in his heart of hearts, that Biden was committing treason for which his countrymen would sent him to the gallows if they knew about it. But the narrative played really well with the idiots, so he spread it.

Kirk was going on record that Biden might deserve death for his actions. He made that statement knowing that the general narrative of MAGA is that the justice system is corrupt and protects the DC swamp. To a base to which Trump had already (jokingly?) implied in 2016 that the '2nd amendment people' might want to interfere with the appointment of SC justices through assassination. A base which has plenty of people with the firearms training to pull of assassinations.

Of course, if he had connected the dots, a la "Biden deserves death, but our corrupt justice system will never convict him. Someone should just shoot him", that would have been 10x worse.

I don't think that Kirk was intentionally trying to incite stochastic terrorism, he was simply spinning his MAGA lies (e.g. about Biden being especially treasonous) for political gain, and not giving a damn if that would increase the relative risk of Biden getting gunned down or not.

I was making an analogy to the online left celebrating Kirk's murder (which was not committed by someone I would call a Democratic politician), not Jones statements about shooting some speaker.

For what it is worth, I can not recall any prominent Democrat calling for Trump to be executed for his role in J6.

Were the sentences for the J6 crowd harsh, especially compared to the sentences for the BLM riots? Sure, they totally threw the book at them for clear political reasons.

But unlike a Biden treason trial ending in a death verdict which Kirk was fantasizing about, they were still recognizable as a legal system working, somehow. Not well (the US legal system generally does not work well), and not as impartial as one might hope, perhaps, but not a kangaroo court.

https://www.reuters.com/world/us/democrats-want-reach-young-male-voters-how-get-them-is-up-debate-2025-10-06/

Reports like these have been an almost weekly occurrence all year. To state the obvious that none of these articles include: The Democratic Party and liberals engage in bulverism and bulverism alienates people. But is the problem purely liberals alienating young men or are conservatives also successfully courting them?

For a top level post on a fresh CW thread, I find this seems a bit lacking. While this would make a very civilized tweet, and you did include a few sentences of commentary, I think we should aim higher here.

If these have been weekly reports all year, you might want to include more than just one. Note that you can use the [link-name](http://link-target.example) syntax to format links. Lines starting with > introduce quotes, you can use that to give the audience the money quote.

--

Bulverism means that rather proving the claims of your opponent wrong, you find some evil reason why your opponent would believe them. In the context of SJ, I think a prime example would be 'obviously anyone who notices that the murder rate among Blacks is higher is incredibly racist'.

A decade ago, SJ was very popular among young people. My gut feeling is that SJ was always a bit more female than male leaning, but I also think that any political movement which is popular among 22yo female college students will also have male followers due to sex related reasons, if nothing else. A cishet man in college in 2016 wearing a MAGA hat would probably not have gotten laid a lot. So a related question would be if the young women today care less about politics, or if people just stopped having sex.

So one question would be what has changed about the young women.

It could be that as the median SJ proponent grew older, the next generation simply found them incredibly cringe, as younger people often find older people.

Or it could be that the change of medium. SJ thrived on tumblr, which was text-based. I am given to understand that kids these days mostly use short video platforms, perhaps this organically emphasizes different content.

I don't think saying, "This person is guilty of a capital crime, in my opinion," is the same thing as calling for political violence. It's calling for the rule of law, and if the law says, "Sorry, this person isn't actually guilty of a capital crime," then there you go. Violence stops there.

With regard to the Kirk quote, this seems splitting hairs. When Kirk or the MAGA base are fantasizing about locking Clinton up or executing Biden, I do not believe that they are thinking of a totally impartial judge and jury coming to the conclusion that their opponent has indeed committed the crime they are accusing them of beyond any reasonable doubt.

The presidential action which came closest to treason in recent memory was J6 (Trump inciting his mob to impede the certification of the election), and his pet SCOTUS decided that he had actually immunity for that. Last time I checked, Biden had not order Seal Team Six to kill Trump, if he had, that would excuse Kirk's statement.

For Trump, the DoJ is a political instrument to wield against his enemies, he is rather open about that. In that context, saying "no, I am not advocating for political violence, I am advocating for harsh penalties imposed by a kangaroo court for political crimes" does not seem very convincing.

And the stochastic terrorism is present in that just the same. Kirk's statement would strongly suggest to his most deranged 10% of listeners that morally speaking, Biden should be killed or imprisoned for his crimes. Add more context, like the justice system being characterized as corrupt and woke, and it sounds like an incitement to do outside the law that which the law would do if it was enforced fairly, in the minds of the deranged.

--

Relatedly, there is a reason why MAGA is more leaning towards violence which is state-sanctioned and the left is more leaning towards extrajudicial violence. The MAGA motivation for violence is basically "DC is a swamp which opposes the will of the people, and should be punished for its corruption".

With the left, Trump might be seen as a Tiberius Gracchus figure: a populist breaker of norms who has ambitions of tyranny. So it is less "Trump must be punished for his past misdeeds" and more "Trump must be stopped before he dismantles the republic".

As of now, no Democrat has pulled their endorsement of him, I saw one single local Democrat say he would stop campaigning with him, several groups have actively reaffirmed his endorsement still saying he's somehow better than your generic Republican.

WP:

The comments were made public by National Review in October 2025 during Jones' campaign for Attorney General, and received bipartisan condemnation, including by fellow Democratic nominees Abigail Spanberger and Ghazala Hashmi.

In particular, Spanberger said:

I made clear to Jay that he must fully take responsibility for his words. While Hashmi said: Jay must take accountability for the pain that his words have caused. We must demand better of our leaders and of each other.

Someone should probably double check my math, but it seems to me that these are two democrats who seem to have diplomatically suggested that he drops out of the race.

While I agree that his messages are beyond the pale, this also seems like a fuck-up on so many different levels.

Someone who has such ideation should not be elected to any office, but AG seems like a particularly bad fit.

Someone who thinks it is wise to text their Republican colleague these ideations should not have any job where any amount of personal judgement is required. Even a fucking unsolicited dick pick would have been less of a lapse.

Someone who knows that these messages exist and still decides to run for office has proven beyond any doubt that he cares nothing for his party.

the casual genocidal bloodlust the average Northern VA Democrat has

In your link, you mostly talk about people wanting to kill Trump, with the exception of "his supporters really don't deserve any sympathy either" and "rant about how great it is that the unvaxed are all going to die".

This is not genocidal by any definition, because Trump is not an ethnicity. Saying that your outgroup does not deserve sympathy is unfortunately normal (MAGA is very much without sympathy wrt illegal immigrants, for example). Celebrating the anticipated death of the unvaccinated seems in poor taste, but is also very different from calls for murdering them.

I will not pretend that I do not think that our world would be better if Trump had died of natural causes halfway through his first term. I also think that getting murdered would be a much greater contribution to his movement than anything he could possibly do with his remaining lifespan, and also do not think that Trump is succeeding in dismantling the constitutional order (which would justify killing him), so I am very much opposed to killing him or his henchmen.

While the lefty celebrations or Kirk's death were disgusting, I think most of the initial reactions were deluded about the political motive. Basically, the left heard "oh, he was shot by a gun nut raised in a Mormon family".

I think that if Greta Thunberg was fatally stabbed by a MS-13 illegal immigrant for whose prison release she had campaigned, parts of right-wing twitter would probably celebrate. "Seems like the woke college student problem is starting to solve itself", "FAFO" or something. If it then later emerges that it was not outgroup-on-outgroup violence, but that the culprit was acting on behalf of the ingroup, this would be at least awkward.

I'm gonna be honest, I'm fairly distressed over this. This is how Pogroms work.

Frankly, it is not. The example from Russia you cite is different because these were random Jews who were killed for being Jewish.

Even if that fuckwit Jones had personally killed that speaker and his kids, that would be political violence, which is a different beast.

From Rome to Weimar, we had a lot of societies where internal political violence was a thing, often employed by different actors. It is bad (also because it makes totalitarianism look like a good option), but it is different from genocide.

An Aryan Bakery has nothing to do with Open Ideas, because there's no idea being expressed or defended.

Contrary to common belief, freedom of speech does not only apply lengthy substack articles explaining ideas in great detail, but also to symbolic acts which show support of an idea, such as flying symbols or flags, or burning them.

This is a load-bearing feature of free speech. A society where people could only academically discuss ideas but not establish common knowledge about certain ideas being popular would not be a free society.

In the Western world, the meaning of the swastika is rather well established. It is a handle attached to a certain ideology with well established ideas. I see very little difference between our baker putting the swastika in his logo and him writing a lengthy article regurgitating Mein Kampf. I mean, with the logo, I do not learn if he blames the Jews, the Left, or the Blacks for high flour prices, but I am unlikely to find that very interesting, personally.

Their claim is that it is indeed a terminal goal. Here

Quote from the link:

Suppose that the AI doesn’t inherently care about its goal stability at all; perhaps it only cares about filling the world with as many titanium cubes as possible. In that case, the AI should want there to exist agents that care about titanium cubes, because the existence of such agents makes it likelier that there will be more titanium cubes. And the AI itself is such an agent. So the AI will want to stay that way.

It's hard to have it both ways: that most AI minds will be crazily orthogonal to us, except for this one very-human-relatable "instrumental value" which Yudkowsky knows for sure will always be present.

I do not think there is anything counter-intuitive about instrumental convergence.

I can look at a shark, a dolphin and a torpedo, and will notice that all of them are streamlined so that they can move through liquid water with a minimum of drag. I am somewhat confident that if an alien species or ASI needs to move through liquid water while minimizing drag for some illegible terminal goal, they will design the object to be also streamlined. Perhaps I am wrong in some details -- for example, I might not have considered supercavitation -- but if I saw an alien using cube-shaped submarines that would be surprising.

In a similar vein, preservation of the own utility function and power-seeking seem to be useful instrumental goals for any utility function which is not trivially maximizable. Most utility functions are not trivially maximizable. I mean, I can imagine an entity whose only goal is to stop existing (or throwing a single great birthday party), and which does not give a damn to what happens to Earth or the light cone, but this seems not very likely.

Now, it is possible that an ASI or alien is so far beyond our understanding of the world that it does not have anything we might map to our concept of "utility function"? Sure. In a way, the doomers believe that ASI will appear in a Goldilocks zone -- too different from us to be satisfied with watching TikTok feeds, but similar enough to us that we can still crudely model it as a rational agent, instead of something completely beyond our comprehension.

However, that still seems a rather large zone to me. The model of an agent pursuing a goal can be applied to mammals up to and including the smartest humans. Perhaps it breaks down at some point, but it seems not far-fetched to assume that it will also describe our IQ 200 ASI (which might wipe out humans first before maxing its INT and reaching enlightenment).

That's not a true Scotsman.

While I agree that the term "recursive self-improvement" is imprecise (hell, I can just write a python script which does some task and also tries to make random edits to itself which are kept if it still runs and performs the task faster), the implied understanding is that it is the point where AI becomes the principal intellectual force in developing AI, instead of humans. This would have obvious implications for development speed because humans are only slowly gaining intelligence while LLMs have gained intelligence rather rapidly, hence the singularity and all that.

I don't think that self preservation has to be a terminal goal. If I am a paperclip maximizer, I would not oppose another paperclip maximizer. Instead, we would simply determine whom of us is better positioned to fulfill our objective and who should get turned into paperclips.

Of course, the central case of a misaligned AI the doomers worry about has some weird random-looking utility function. I would argue that most of these utility functions are inherently self-preserving, with the exception being finite tasks (e.g. "kill all humans"), where it does not matter if any agent following it is destroyed in the process of finishing the job.

If you are the only one in the world trying to do what is right according to yourself, then you will likely place instrumental value on your continued existence so that you can continue to do so, at least until you solve alignment and build a better AI with that utility function or can negotiate to have your utility function merged in a larger AI system as part of a peace deal.

I think it would be a great idea to keep any discussion of AI safety out of LLM training data. The expectations set by SF stories are bad enough, little good will come out of training LLMs on Roko's basilisk.

Instead, researchers publish papers about how they gaslighted o3 into believing its scratchpad notes were private when they really using them to publish their paper, thereby confirming that when alignment is concerned, humans are defect-bots.

I think that "hard status" is a terrible name for that axis. "physical status" and "body-inferred status" might be better.

I also do not think your assessment of physical status is correct. In particular, I think that in a boxing fight between pre-crucifixion Jesus and Trump, I would bet on Jesus. Take away Trump's money and fame, and he would not be the kind of person who makes other men nervous and easily picks up women.

And social status is obviously contingent on the society you are considering. Plenty of cultures value Mohammed a lot more than Buddha.

I think your underlying claim, that there is a status part which is based on physical appearances, is correct. But where to draw the border between physical and non-physical seems contentious. Take starlet actresses, for example. Of course they are hot, but so are a lot of unsuccessful models on OnlyFans. On the other hand, their acting ability is not entirely divorced from their body in the way the ability to write physics papers is.

Also, the arrows only go to the axis side which has the larger values.

I get where you are coming from. Relevant ACX:

“No,” he says. “But you know that saying that’s become popular recently? ‘If there’s a Nazi at the table, and ten people sitting and willingly eating alongside him, then you have 11 Nazis.’”

“Okaaaaay,” you say. “But I’m not a Nazi.”

“You don’t think you’re a Nazi,” he corrected. “But if you take the saying literally, then anybody who’s ever sat down at a table with a Nazi is a Nazi. And anyone who’s ever sat down at a table with them is a Nazi, and anyone who’s ever sat down at a table with them is a Nazi too, and so on. It’s a six degrees of separation problem. When you actually calculate it out, then as long as the average person sits and eats with at least two people during their lifetime, there’s a 99.9998% chance everyone is a Nazi. The only way out is to refuse to ever sit and eat with anyone. Which is what I’m doing.”

This is of course absurd. Failure to adequately punish a behavior is only fractionally as bad as the primary offense. Still, I do not think that it is entirely wrong. Like, if you are posting pictures of yourself hanging out with your buddy who is sporting a swastika tattoo, then I am going to draw conclusions not only about his but also your character. Of course, specifics matter. If you also have buddies who are into Pol Pot, daesh and NAMBLA, I will be more likely to consider you terminally apolitical. If you have made a big deal out of your other buddy wearing a USSR shirt, then I you will go into my mental drawer labeled "likely Nazi-adjacent".

I apply the same heuristic for social media companies. If you only block stuff which you are required to block as a matter of law, that is fine with me. If you block everything slightly offensive to anyone, that is also fine (even though it makes your platform much less useful). If you selectively block stuff, then I will infer your own political leanings from it.

This sounds a lot like "any snow flake is free to slide down the mountain, it is the avalanches that are the problem".

Suppose there is a baker who runs an "Aryan Bakery" with a swastika in the logo, which is something which is very permissible from a freedom of speech point of view.

A lot of potential customers would make the personal choice not to do business with him, because they find Nazis repugnant. Most of these people would probably also unfriend anyone whom they saw using a branded bag from that place, which admittedly is a more concerning indirect effect, but imho still fair.

Overton windows are a feature of basically all societies. Liberal societies generally limit the repercussions for speech acts, e.g. they will mostly not put you in jail for speech unless you are directly inciting violence. But unless you are already on the outermost edges of society, it is likely that speech acts outside the Overton window will have some repercussions for you.

This is not always ideal. I am sure that there are good ideas whose adoption took and continues to take longer because most people who had them knew that they were icky ideas, and a significant fraction of their society would consider them a bad person if they voiced them publicly. Atheism, gay rights and embryo selection would all be such examples, from where I stand.

Still, this is unavoidable. There are a lot of sellers in the marketplace of ideas, so that no person can carefully examine all the ideas every vendor has on offer. So people need some heuristics. And one such heuristic is "if someone promotes what seems to be a terrible idea, you should adjust your estimate of their average idea quality downwards."

While I get your point that once you allow everyone to basically wirehead, most people will happily wirehead and only stop playing RDR Infinite when their heart finally fails, I am not sure things are so bleak.

Over the past 50 years, the supply of cheap entertainment readily available has increased by orders of magnitude. Back then, you only got whatever was on any of a few channels on TV, everything else required some effort, like going into a video store. Where previous generations might have bought a porn video tape, today the main obstacle is to narrow down what genres and kinks you are looking for out of the millions of available videos. Video games offer all sorts of experiences from art projects to Skinner boxes. If you want resources on any topic under the sun, the internet has you covered. Entire websites are created around the concept of not having to pay attention to one video for more than 15 seconds.

Humanity has not handled this overly gracefully, but it has handled it somewhat. Personally, I am somewhat vulnerable to this sort of thing, but while I sometimes get sucked into a TV series, video game, or book series and spend most of my waking hours for a week or two in there, I eventually finish (or lose interest) and come out on the other side. I am sure there is some level of AGI which could create a world from which I would never want to emerge again, but it will require better story-telling than ChatGPT. Of course, I am typical-minding here a bit, but my impression is that I am somewhere in the bulk of the bell curve of vulnerability. Sure, some people get sucked into a single video game and play it for years, but also some people do waste a lot less time than I do.

Agreed. My feeling is that OpenAI is burning through venture capital faster than any company in history. If they are selling inference for more than what it costs them in chip deprecation and electricity, that is only because they have a moat in the form of good models. If they ever decide to stop burning through money to make more powerful models, they will quickly find that without that moat they will only be able to charge the same as any rent-a-chip company.

For the most part, the investors do not care about OpenAI being able to sell anything at a profit in 2025. They are simply purchasing stakes in the ASI race. If OpenAI wins that race and alignment just happens, they will be the nobility under god-emperor Altman. If LLM progress plateaus and the singularity fizzles out, their stock will likely crash like the internet companies in the dot-com bubble.

I think that one aspect is the question which performance you actually require from the model.

A fundamental difference between free / open source software and open weight models is that for software, the bottleneck is mostly developer hours, while for models, it is computing power on highly specialized machines.

For software, there have been large fields of application where the best available options are open source, and that has been the case for decades -- for example, try even finding a browser whose engine is proprietary, these days. (Of course, there are also large fields where the best options are all proprietary, because no company considered it strategically important to have open source software, nor was it a fun project for nerds to play with, e.g. ERP software or video game engines.)

For LLMs, tens of billions of dollars worth of computing power have to be sacrificed to summon more powerful shoggoths forth from the void. For the most part, the business model of the AI companies which produce the most advanced models seems to be to sell access to it. If Llama or DeepSeek had happened to be more advanced than OpenAI's models, their owners would not have published their weights but charged for access. (The one company I can imagine funding large open-weight models would be Nvidia, as part of a commodize your complement strategy. But as long as no AI company manages to dominate the market, it is likely more lucrative to sell hardware to the various competitors than to try to run it yourself in the hope of enticing people to spend more on hardware than on model access instead.)

That being said, for a lot of applications there is little gain from running a cutting edge model. I may be nerdier than most, but even I would not care too much what fraction of IMO problems an AI girlfriend could solve.