@ThenElection's banner p

ThenElection


				

				

				
0 followers   follows 3 users  
joined 2022 September 05 16:19:15 UTC
Verified Email

				

User ID: 622

ThenElection


				
				
				

				
0 followers   follows 3 users   joined 2022 September 05 16:19:15 UTC

					

No bio...


					

User ID: 622

Verified Email

I don't think that this conspiracy is likely or that anything is so rationally planned, but Israel might prefer a more multipolar world. China doesn't care which tribe occupies a particular piece of land or how it governs it, so long as it's pliant to China's national interests. A realignment with China would mean much less finger wagging, no threats of boycotts, no constraints around solving security issues. It's not altogether clear that access to weaponry would even suffer too much, especially a decade from now (and if Israel's primary problem is controlling an insubordinate population, China already has the US beat there in technology). And, if nothing else, Israel could play the US and China against each other, hoping to get the best deal from both. The only big question is whether Israel can offer more to China than China's Middle Eastern allies can (which seems unlikely, but maybe China could find a way to thread the needle and work with both).

  1. As far as the vigilantism, more likely than not it would be fine. There's the risk of getting stabbed, and I'm a relatively petite guy. But getting stabbed with a hepatitic knife is kind of a medium case scenario. The worst case is the city figuring out it's some kind of vigilantism, deciding to make a grand symbolic point, and putting in the effort to ruin my and my family's lives. A year or two back a guy took a hose and sprayed down a homeless woman shitting in front of his business during open hours, and the city suddenly took far more interest in charging that assault than in homeless people stabbing each other. Additionally, there are hundreds of people doing the exact same thing or worse; it would be a drop in the bucket for little payoff.

  2. People do do that, but it's taking a risk. You're surrounded by a 24/7 surveillance system called your neighbors. I'm on good terms with all of them, but if someone wanted to screw me over, it would be very easy.

I live in a city where the government will punish you more (up to putting a lien on your house and, if you don't pay the rapidly accumulating fines, appropriating it) if you change your windows to be double-paned without getting the appropriate permit than if you regularly go to elementary schools, expose yourself, and masturbate to the children. And god forbid if a taxpaying resident decides to perform any vigilante activism against the public masturbator. And, of course, the chronic masturbator can throw a rock through your window, and if you don't respond appropriately and request permission to fix it through the city channels, the same appropriation process begins.

This colors my views.

The State already is a panopticon. The only fly in the ointment is that it's only a panopticon for members of society who are well-integrated into the economy, own assets, and generally "have something to lose."

This extends the panopticon to those who have nothing to lose. So we lose the anarcho part of anarchotyranny. Perhaps losing the tyranny part instead would be better, but throwing out the rulebook altogether seems a bigger lift than just ensuring it's applied universally.

I've had an idea in the back of my head, which is essentially to have the human brain "train" an attached, minimal, and highly plastic (compared to the brain) ML model through some direct high bandwidth connection (not targeting any particular objective function). The hope would be, once the ML model has converged to some near equilibrium state, it would have learned most of the distributed representations that comprise human values. That model could then be scaled up the much higher compute and used as a foundation model that is then highly aligned with (that particular) human's values.

Aside from being really speculative, it seems very likely to me that this would require feedback connections to actually work. And so the model is training the brain at the same time the brain is training the model.

That's a big part of it, yes. Finance is about abstracting away all the messy realities of the real world into a single self consistent symbol--money--so that humans can accurately act on knowledge of the real world without having to know any of its concrete details. Software is also a big part of it, as MadMonzer points out. And we are best in the world at both of them: there's no reason for us to make a lot of widgets when we can manipulate symbols to create information that's worth 100x as much. Trade policy has enabled us to make this tradeoff, and we do.

You can also frame higher education, law, and media as symbol manipulation industries, though I see their successes as more downstream of the symbolification of the US than a cause of it.

It wouldn't be wrong to say that the entire point of logistics is the abstraction away of place.

I feel this summed up my thoughts decently, but it lacks my passion of hate I have for our system.

I'd prefer vitriol and passion over AI slop. Not reading.

That said, if debt is going to be issued, something like credit scores, implicit or explicit, are pretty much required. People likely to repay debt and who want the conveniences of a good credit score are going to usually get better scores than those who don't.

I have some sympathy for immigrants who come to the US without a credit score and need one. On the other hand, did you miss some payment recently? If so, the credit score is functioning exactly as it should.

Don't ask me; that was Claude's take. Mostly just kvetching here about Western LLMs.

Cats were domesticated (or self domesticated) through their interactions with granaries. They hunted rats, which was useful in and of itself. But their independent streak didn't lend itself to many other economically productive activities, so they were creatures that were mostly for domestic purposes, and so they ended up feminine coded.

Wolves were more amendable to selecting for a broader range of economically useful activities: hunting, herding, guarding. All of these activities are masculine coded, and so dogs eventually gained the masculine coding.

I wonder how universal this coding is, though. Egyptian Bastet and Norse Freyja were both female and associated with cats, and I don't know of any male gods associated with them.

Can you share the original, if you're comfortable revealing your country?

Deepseek tells me it's a traditional Japanese saying (while Claude scolds me for asking questions about misogynistic sayings).

Trade policy has changed the composition of the domestic elites who make bad political decisions. It's caused (in part) the specialization of the American economy into one of abstract symbol manipulation. Although it turns out that's probably the highest value thing anyone can do, the winners of a symbolic economy create an unmoored society. And the issues you list are all downstream from that: real physical and safety issues have become secondary to the symbol.

I don't think Trump's tariffs are good or will do much to reverse this trend. They are, however, a strong symbolic strike against the ruling elite, which will have unfortunate side effects on the material wellbeing of Americans.

Is men away from home frequenting sex workers really that much more common in Africa? Genuine question: I would bet prevalence of sex work is really high everywhere with a large group of transient men without much education and weak public health. But Africa's HIV rates are far higher (10-50x) than e.g. India's, and Africans aren't having 10x the amount of risky sex as Indians (I think?)

Maybe the graph has more clustering for Indians, which would limit spread, but I don't see how that would cause the rate to be so much lower.

Given the heterogeneity of prevalence even within Africa, I think reaching some bare level of competence in government/public health actually makes a difference here.

That happened to another of the cult members: three of them got together to try and murder their 80 year old landlord by stabbing him multiple times with a fake samurai sword, knocked him out, couldn't figure out how to decapitate him and dissolve his body before he regained consciousness, and then on coming to he shot two of them, killing one.

There's a lot of truth to this, but it misses a bit of my perceived gender dynamics. In the past, the wife and both their families would pressure the husband to go ahead and have kids. That pressure is much rarer nowadays: it seems gauche for someone not in the couple to ask them to do something. And, for whatever reason, wives pressure their husbands about kids less nowadays. Everyone has shifted their preference away from the couple having kids, in favor of other priorities. Boomer would-be grandparents prefer their obligation-free vacations, wives prefer Michelin dinners, and husbands prefer ???.

I think oftentimes we focus too much on the PMC, in large part because it's the milieu that populates places of cultural production. And I think they're more or less a lost cause: they have plenty of agency to change their lives and they choose not to, and even if you promised them double their real income, it wouldn't move the needle much. There's no world in which they'd consider raising a child more prestigious than eating out at overpriced mediocre New American restaurants, and so they'll never have kids.

If we're concerned about increasing fertility rates, we should be focusing on the bulk of people who are not PMC, who already show a willingness to have kids at all, and so nudges there are likely to have a greater marginal impact.

It's all simply applying best practices and not fucking up, almost boring.

It's pretty weird: there's nothing there that any of the big labs in the West should have trouble replicating a hundred times over, and DeepSeek still managed to make something that can trade blows with them (and subjectively win, more often than not).

Might it really be just clarity of purpose leading to focusing on what matters? About a week ago, I remember Claude lecturing me, apropos of nothing, a bit about how it's best to buy from local bookstores instead of online retailers in response to me asking about what kind of textbook would be used for a particular course. I've not experienced DeepSeek doing anything even close to that, and it makes me wonder if the extraneous post-training being lathered on is the real difference here. Western models get distracted and are pulled in a thousand different directions, while DeepSeek can focus on what's relevant.

They provide human generated data for other companies to train on, which is generally hard and expensive.

In theory, at least. In practice, at best you get a bunch of data from low wage Filipinos. At worst, you get data generated by existing models and laundered through the workers trying to hit quota.

There is the potential for a kind of recursive growth, once you have access to some kind of external verifier. A model of a certain size performs a search; external verifier gives it back a reward signal for good searches; and the model learns and gets better at the search, allowing the process to begin anew. E.g. AlphaZero.

Where it gets murkier in my head is whether LLMs can act as their own verifiers, even with arbitrary compute. As a proof of concept, humans can think a long time to come up with a novel insight and learn it, but it still seems we learn best when there is some kind of objective/external feedback signal.

I would reject the idea that he's not done much. Not for consumers, no (his best product was Facebook circa 2004), but he's built an empire of information. WhatsApp has been a qualified success, and Facebook's cloning of Snapchat features has basically killed the latter company's trajectory. The Metaverse is an abject failure, but, you win some and you lose some.

And DeepSeek was undoubtedly a panic inducing setback, but the LLaMA play wasn't to establish dominance, but to prevent competitors from displacing Facebook through access to proprietary models. If someone else does that instead... well, he still has his spigots of cash and information.

In terms of his legacy, it's still too soon to tell. He's well-positioned to continue to make an impact: he has GPUs, and nothing matters more than them at this moment in history. If DeepSeek managed to build a better model than one costing an order of magnitude more, then it's not unreasonable to think Meta can now just copy that architecture and throw an order or two of magnitude more compute at it and make something even better.

The big question is whether Meta is able to excise the Google-style cruft and bureaucracy to be able to execute more quickly and effectively.

Most SS fraud is going to be concentrated in SSDI, which is like 15% of its budget. And although I bet there is a shocking amount of fraud in it, it's not going to be 100% fraudulent.

How would Scott self-immolating have helped anything? It's not like he was sitting on some special knowledge that no one else in the universe had or that people couldn't read about in other blogs if they were inclined to. If he had explicitly pushed against the Overton Window on it, the Overton Window would have thrown him into the outer darkness. You push against the things you can shift, not sleeping tigers that are blocking the path.

Sometimes the Kolmogorov Option is the right one.

I think we need more data to find a convincing explanation. We don't even really know when IQs diverged: has it been that way since time immemorial? Or maybe it was a process that took a millennium? Or maybe it was strong selection over a century or two (something with the Black Death providing a selective pressure for more effective immune systems?)

This doesn't seem like information that's been lost irrecoverably to history, either: looking at modern and ancient DNA likely would make it possible. There just would have to be a social and scientific willingness to open that Pandora's box.

Reminds me a bit of Mankiw's Optimal Taxation of Height parody paper. IQ is observable, inelastic, and correlated with income, and so the government should collect IQ measurements and tax those with high IQ more.

He was controversial; I vaguely remember a Gould polemic against the Bell Curve, back in the day (thought it was in the Mismeasure of Man, but that was published in 1981).

The difference was controversial meant "lots of people criticize you and imply bad things about you," not "you get blocked from all media organs for all eternity, as does anyone you're associated with."