Oceans Yet to Burn

Mr. Newsom, a Democrat, signed the order mandating state agencies work with academics, labor groups and the A.I. industry to study how to subsidize companies that keep employees rather than replace them with the technology.

Why are you doing this to me, man - why do you hate me so much?! I know this is just slopulism bullshit, it won't go anywhere beyond rent-seeking handouts to connected industries, which sucks but we will survive like we always do. But Newsom is such a fucking clown of a politician, he indulges every worst instinct under the sun with no regard for the consequences. Both Republicans and Democrats have very deep communist tumors growing on them these days that he intellectually knows better than to feed, but he does it anyway.

[Gavin Newsom] intellectually knows better than to feed [wasteful slopulism]

Me, I rent, so I've always been quite aware that I don't own the land I reside on, although a lot of other people don't seem to know it.

Could be "People of Color" or BIPOC or the proletariat, whatever. The US has built its power by exploiting that group both in and out of the country, and it therefore owes that group some sort of restitution whether they be in or out of the country.

I don't put much stock in such views but they don't strike me as particularly incoherent on that level, at least if the people espousing them have even a modicum of sophistication.

I mean, theory of mind-wise it's not fundamentally very difficult, you just find a group which contains both native Americans and some proportion of immigrants and decide that the ruling class in America maintains its power by exploiting both of them and therefore owes something to both of them. Could be "People of Color" or BIPOC or the proletariat, whatever. The US has built its power by exploiting that group both in and out of the country, and it therefore owes that group some sort of restitution whether they be in or out of the country.

Thanks. On a first pass my experience was that this explanation did not make sense logically, but did somehow make sense psychologically.

Based on that, I may have been able to identify the flipped assumption relative to my way of thinking. This was a bit of a brain teaser.

What we will call the "Theory X" way of thinking is that control of access to a territory is one of the most important forms of power, and therefore migration is (in the presence of demand) mostly imposed by the ruling class, and if they are not able to either impose or contest immigration, then who controls the territory (and thus who constitutes the ruling class) is under contest.

In this view, physical presence in a territory allows someone to exercise power. Therefore, the movement of Group B to territory originally held by Group A is effectively a zero-sum transfer of control shares from Group A to Group B, facilitated by the ruling class. Thus, migration cannot act as net restitution to the combined Group A + Group B.

In "X" terms, Group A are assumed to have both shared cooperative interests, such as trade and opposition to other groups that are opposed to both, and also conflicting/competitive interests, such as cultural economies of scale, or as a potential axis of fracture during a resource conflict.

In the view Morlock is describing here (which we will call "Theory Y"), (a) some fraction of control shares are controlled by "the ruling class," (b) Group A and Group B are assumed to be a united faction, and (c) migration is not (necessarily) something controlled by the ruling class (e.g. perhaps only border enforcement counts as 'from the ruling class'). Moving more members of Group B to the territory thus increases the total power of Group A + Group B by shifting control shares away from the 'ruling class,' thus qualifying as net restitution.

In this view, suffering primarily comes from exploitation, not from natural scarcity. Culture is treated as largely not relevant, and ethnicity as an axis of fracture is something that only becomes relevant due to exploitation (as exploitation is the primary source of scarcity).

Thus, the move does not count as decreasing the power of Group A.

In this view, physical presence in a territory allows someone to exercise power.

I think this is straightforwardly wrong, at least here in the US, and my exhibit A would be the Supreme Court decision "City of Grant's Pass v. Johnson"

The really, really brief capsule of that decision is that Grant's Pass, here in Oregon, criminalized public camping. The controlling circuit court decided that, since Grant's Pass did not have any homeless shelters, this meant that there was a class of people who were legally prohibited from sleeping anywhere in Grant's Pass, and that essentially this was an illegal attempt to punish people merely for belonging to the class of homeless people.

The Supreme Court pulled an Anatole France, explaining that since the law of Grant's Pass, in its majestic equality, prohibited both rich and poor alike from sleeping in public parks, this was not illegal discrimination against the homeless.

For these purposes, the takeaway is this:

It is entirely possible to be a citizen of the US and for it to be illegal for you to actually reside in any specific place within the US.

Being a citizen of a country is not like owning a piece of property. Yes, as a citizen I have some influence on how public land is used through the institutions of representative democracy, but if I become homeless that doesn't mean they'll let me live in the public park.

That's why I push back so hard against @isaacsapphire 's analogy above:

In that analogy, a person here on a temporary work visa, who cannot vote, who can easily be expelled from the country by the bureacracy, and who rents his living space from somebody else, that guy is one of the people who lives in the house.

A Native American citizen, one who can certainly vote, and one, say, who owns property and a business, is, in this analogy, the homeless person who exerts no control over the house.

And so our analogy has drifted so far off from actual reality that even answering the question tells us nothing about the thing we're trying to illuminate.

Is this a criticism of naive landback and decolonization rhetoric? Yes, it absolutely is.

My being an American Citizen is not the same as my owning land, and trying to reduce it down to such a thing will just confuse us.

A better simplification, in my opinion, is that my citizenship, and migration in general, are about which economies I am allowed to participate in.

This stuff is only mystifying if you force yourself to forget about some very basic economic principles, in particular the fact that both labor and capital have a vested interest in being able to move freely.

When you sell in a market, you want to go to a place where there is high demand for what you are selling and low supply, because then you get to charge more for what you are selling.

The danger and tragedy of market economics are that my personal incentive is to find a place where there is high demand for my services, set up shop, and then prevent new people from entering the market, therefore keeping supply low relative to demand and allowing me to jack up my prices.

But of course, if I do this, I am standing in the way of the ways that markets improve the lot of mankind.

There's a story about this lower down on my blog, but suppose I am a Doctor from a foreign country with a fairly rare specialty.

Perhaps I might want to practice my specialty in the rural united states; it might be that I earn less there then Doctors in the big city, but far more than I would earn in my home country. In fact, if there's enough difference between what I can earn in the US and my home country, I might be willing to undercut native-born doctors on my salary asks and still consider myself to be really well off.

So patients in this rural area have more access to medical specialty at a lower cost. Hospitals can pay lower wages. US Doctors are annoyed because the increased supply of doctors means that they have to compete more by possibly accepting lower salaries for their labor.

But like, who is the "ruling class" here? Have native-born patients in this scenario somehow had their power diluted through, uh... the ability to purchase a service they need at a lower cost than they would otherwise be charged?

I'm skeptical about that last one really being a loss of power.

All of this stuff becomes far less mystifying once you actually understand it as being heavily related to economic participation.

To go back to "How do you square landback reforms and pro-immigration stance?"

The answer is that you find a way to believe that most Native Americans AND most immigrants have been unfairly excluded from full participation in the American economy. There's a bunch of lenses you can use to reach that conclusion, I don't think it is really a particularly out there idea.

And again, if I'm a Native American who owns an apartment building, who is better for my power: A green card holder who pays rent on time every month, or a native-born citizen who doesn't?

For example, this assumes that either ownership of the apartment building is not subject to political attack, or that expropriation attacks against ownership of the apartment building can only proceed along a class axis, rather than an ethnic axis, or combined ethnic/class axis.

Subscribers of Theory X generally believe that physical force, or hard power, is the foundation of power, both over the environment and over people, and soft power derives its influence from coordinating and directing hard power. Physical proximity allows greater exercise of both hard and soft power.

Anyhow, a thought experiment can be (a) used to try to force someone to take a socially disadvantageous position, or (b) reapplied to some other domain to get leverage based on the answer, where explaining why it doesn't apply isn't trivial.

You could make some cases of people who are literally paralyzed being unable to cycle, and sure, they’re less privileged than cyclists. But he was explicitly arguing for cars against bike lanes, which makes the argument ridiculous. You could buy and maintain a good commuter bike for less than the price of virtually any legal car.

There are more things that could be said here, based on the built environment, such as time cost. Will the boss accept being sweaty when one arrives at the office?

Read a couple of materialist-scumbag posts, and was wondering - how much does kontextmaschine use "layers" as a term in his written analysis? I didn't see that as key to his way of thinking. I see it as more key to my way of thinking (e.g. Now, Melt 10.a).

No, the original KM does not use "layer" as analysis much, and it doesn't appear in the corpus I used. Whereas the generated posts have. I asked the bot writing the posts, and here is what it said.

"Layers" is a default move in generic analytical/explainer prose — the gesture of "here's a surface and underneath it there's more." It signals depth without having to demonstrate any. It's the structural cousin of "not X, it's Y" — both are reframe-shaped moves that announce analysis rather than performing it. KM's actual move isn't layers. It's accumulation and adjacency. He puts things next to each other — the country club next to the 1920s road, the road next to the postwar suburb — and lets the reader feel the system. There's no "underneath." It's all on the same plane; the work is in the sequencing and the digressive return. "Layers" implies a vertical stack you peel; KM does horizontal sprawl. So when I reach for "layers" I'm reaching for the wrong geometry. I'm importing a default essayistic shape ("look beneath the surface of X") into a voice whose whole epistemology is that there isn't a beneath — there's just more stuff in the network, more material and logistical adjacencies, and the job is to keep naming them until the pattern shows. I'd guess it also creeps in because "layers" is genuinely overrepresented in the kind of corpus training data that gets labeled "thoughtful cultural analysis." It's a tell of the genre I'm pastiching toward rather than the specific voice I'm supposed to be hitting. Worth adding to the AI phrase checklist if it isn't already.

LLMs don't necessarily have good insight into their internal operations, but this bit fits:

It's the structural cousin of "not X, it's Y" — both are reframe-shaped moves that announce analysis rather than performing it.

Text is generally a linear stream. We can think of a book, for example, as a serialization of a knowledge graph. In writing an explanation, an author must, to some degree, work backwards in order to resolve knowledge dependencies. The author may already have the whole knowledge graph, however, so to the author, that knowledge may be simultaneous or networked/associative.

A - Genetically-enhanced Neo-Nixon, accompanied by his AI assistant ULTRA-KISSINGER

B - 1,000 Georges W. Bush

Meanwhile, thanks to voters who just want to see what would happen, Neo-Nixon coasts over the line to victory.

Read a couple of materialist-scumbag posts, and was wondering - how much does kontextmaschine use "layers" as a term in his written analysis? I didn't see that as key to his way of thinking. I see it as more key to my way of thinking (e.g. Now, Melt 10.a).

That's kind of a protection mechanism because ideologies are often incomplete or wrong, and so not suited to have complete control over human beings, but it also limits just how much ideology can do.

Eliezer Yudkowsky's whole deal is that he thinks that Skynet is coming based on an extrapolation of existing trends in things like computer chip development like Moore's Law (or thinks that even if the curve tapers off, it may taper off too late), so he wanted to either make sure Skynet was good or prevent Skynet from being developed, or prevent it from being developed until we can ensure that it is good.

He considers this to be of finite but extreme difficulty.

But in 2008, "Skynet might kill everyone in 30 years" looks silly, unless someone takes the graph of computer chip development quite seriously.

Yudkowskian LessWrong Rationalism is, therefore, (a) designed to encourage taking ideology very seriously ("shut up and multiply") while (b) also encouraging being extremely correct ("notice when you are confused"), for (c) a goal considered wacky when it originated (though the problem it was intended to solve was not considered impossible to be an issue later).

To meet this goal of extreme but finite difficulty, ordinary ideology would not suffice. (We have all seen ordinary political arguments.) He needed a group of guys who would be strong adherents to an ideology focused on being extremely correct.

This attracted some broken people. It also attracted some people who would go on to become the post-rationalists.

It was not intended as a cult. However, because it promotes taking ideology extremely seriously, it has pieces that can be used to assemble one.

That's the tension, here.

AI agents are becoming more and more prevalent, but the number of deaths attributable to malign or misaligned AI remains firmly at zero. The closest anything has come to misaligned AI is that one coding agent who published an article accusing a repository maintainer of being prejudiced for not letting him (her? I don't like referring to an agentic entity as "it") make git commits. There was no misrepresentation, and (IMO) the agent's case was pretty solid!

Insofar as anything Skynet-like exists it is AI tools used by militaries to kill their existing enemies in conflict zones, and humans have been doing that for as long as we've been humans. We have exactly zero documented cases of an AI agent killing someone who wasn't "supposed" to die.

This is what frustrates me about Eliezer Yudkowski and the other "rationalists", they seem incapable of updating their beliefs in the light of new evidence, much like that Harold Camping fellow.

They've reasoned themselves into an unfalsifiable position, where the infinitely bad robots are always just around the corner and no amount of evidence to the contrary will ever be enough to dissuade them of that.

It has also, by basically any measurable metric, been… wrong?

He basically created an intellectual scene, which is one of the few scenes doing much ideological development work of interest lately, along with the online right. Most left-wing ideological work lately (that I can see) has been intersectional theory, which is a dead-end in terms of solving 21st century problems.

Many in the scene don't fully agree with Eliezer, and some of them went on to become the "post-rationalists," more of a post-formal, more vibes-based branch.

AI agents are becoming more and more prevalent, but the number of deaths attributable to malign or misaligned AI remains firmly at zero.

Well, we live in the timeline where many in that space will have heard some of the noise Eliezer was making went they went into the field of AI development, but more importantly, the core theory was self-improving AI, and existing GPUs are much less powerful than a human brain.

We're already seeing social distortions, like huge amounts of spam, and "AI psychosis," even at the current low levels. It's also not that hard for a human to exceed an LLM because LLM training is fixed in time.

Most of the risk, though, comes at the tail end, when AI meets or exceeds the power of a human brain. The idea behind the Moore's Law curve is that transistor count on new computer chips doubles every two years, so computer power advances much, much faster than a straight line.

Eliezer's concern is thus that incredibly powerful AI will "sneak up" on us, because we will intuitively expect a linear increase in computer power, but instead there will an exponential increase. Things like image software, LLM translation, and AI video generation may look like a linear increase in capability (it's "adding new abilities"), but they actually represent an exponential increase (those abilities are much more difficult / higher in dimensionality).

Eliezer seems to have bet hard on a vertical optimizer, rather than the broader, shallow, more horizontal AIs we have right now. He seems to have focused on a fast, hard takeoff scenario, where AI very rapidly exceeds human ability, as this was the highest risk scenario. Slow takeoff scenarios are much more ecological, and not much work (that I'm aware of) was done on these.

Any system capable of aligning an extremely powerful AI with human interests would also be capable of aligning less powerful ones, so from a certain perspective it makes more sense to solve the harder problem first. On the other hand, that drags in a lot of philosophical questions about what counts as "human interests," and pieces and parts that address smaller problems might contribute to a larger whole.

Rumors say that Eliezer hasn't kept up with research on implementation, but the whole project was probably inspired by things like, "We'll just program the robots not to have an uprising," which, nevermind AI specifically, is contrary to basic knowledge about the field of software development.

That's kind of a protection mechanism because ideologies are often incomplete or wrong, and so not suited to have complete control over human beings, but it also limits just how much ideology can do.

Eliezer Yudkowsky's whole deal is that he thinks that Skynet is coming based on an extrapolation of existing trends in things like computer chip development like Moore's Law (or thinks that even if the curve tapers off, it may taper off too late), so he wanted to either make sure Skynet was good or prevent Skynet from being developed, or prevent it from being developed until we can ensure that it is good.

He considers this to be of finite but extreme difficulty.

But in 2008, "Skynet might kill everyone in 30 years" looks silly, unless someone takes the graph of computer chip development quite seriously.

Yudkowskian LessWrong Rationalism is, therefore, (a) designed to encourage taking ideology very seriously ("shut up and multiply") while (b) also encouraging being extremely correct ("notice when you are confused"), for (c) a goal considered wacky when it originated (though the problem it was intended to solve was not considered impossible to be an issue later).

To meet this goal of extreme but finite difficulty, ordinary ideology would not suffice. (We have all seen ordinary political arguments.) He needed a group of guys who would be strong adherents to an ideology focused on being extremely correct.

This attracted some broken people. It also attracted some people who would go on to become the post-rationalists.

It was not intended as a cult. However, because it promotes taking ideology extremely seriously, it has pieces that can be used to assemble one.

That's the tension, here.