LessWrong 2.0 Reader

I am working on it! Will post here in the coming week or two about how it’s going.

links 12/03/2024: https://roamresearch.com/#/app/srcpublic/page/12-03-2024

• https://sashachapin.substack.com/p/my-mind-transformed-completely-and Sasha Chapin on how meditation changed him
  • it doesn't seem clear to me whether this is better or not!
  • reduced anxiety seems great, but reduced sense of narrative drama is a big cost. part of what makes life seem meaningful to me is the sense of being part of a story, and if anything i feel like my current arc involves gaining abilities to envision myself as inside a narrative.
• https://www.wired.com/story/murderbot-she-wrote-martha-wells/ Martha Wells seems like a lovely person
• https://www.orcasciences.com/articles  recommended by Ben Reinhardt, great example of rigorous analyses of potential future technologies.
  • https://www.orcasciences.com/articles/checking-my-prejudices-on-materials-decarbonization eg: where does it make economic sense to use electrochemical or biological manufacturing? (compared to "thermochemical", fossil-fuel-powered). For biomanufacturing, only for complex molecules like proteins; for electrochemical processing, mostly metals and things with big voltage potentials in the chemical reaction (zinc, cobalt, copper, lithium, etc) but not simple organic molecules (methane, ethanol, etc)
• https://www.biotech.senate.gov/press-releases/interim-report/ "US National Security Commission on Emerging Biotechnology", a congressional advisory committee led by Jason Kelly of Gingko Bioworks
  • their purpose seems to be getting biotech-friendly policies through congress, with the rationale that this is good for national security/defense.
  • a lot of naive boosterism about biomanufacturing without engaging with the question of "is this better than alternative manufacturing techniques?"
• https://www.aria.org.uk/request-for-opps/ new opportunities for program managers at ARIA: lead a scientific research program!

If you already live in a city, I think you are unlikely to move away to start the family at a cheaper place. You still need to keep a job, and city is where the jobs are.

So the necessary amount of money is:

• to buy (not rent) a place to live
• in a city
• with at least three rooms
• (optionally) enough that the wife can stay at home with kids, if necessary

Hm, maybe we could build a lot of new apartments, and sell them for a subsidized price to families with small children. With the condition that they cannot sell the apartment for N years, or they need to pay the subsidy back. Also, build kindergartens near the apartments (maybe even in the same buildings).

I think "stay within bounds" is a toy example of the equivalent to most alignment work that tries to avoid the agent accidentally lapsing into meth recipes and is one of our most important initial alignment tasks. This is also one of the reasons most capabilities work turns out to be alignment work (and vice versa) because it needs to fulfill certain objectives. 

If you talk about alignment evals for alignment that isn't naturally incentivized by profit-seeking activities, "stay within bounds" is of course less relevant.

When it comes to CEV (optimizing utility for other players), one of the most generalizing and concrete works involves at every step maximizing how many choices the other players have (liberalist prior on CEV) to maximize the optional utility for humans.

In terms of "understanding the spirit of what we mean," it seems like there's near-zero designs that would work since a Minecraft eval would be blackbox anyways. But including interp in there Apollo-style seems like it could help us. Like, if I want "the spirit of what we mean," we'll need what happens in their brain, their CoT, or in seemingly private spaces. MACHIAVELLI, Agency Foundations, whatever Janus is doing, cyber offense CTF evals etc. seem like good inspirations for agentic benchmarks like Minecraft.

Paid sick leave decreases, rather than increases, fertility.

Is that correlation or causation? Like, maybe there is some X that decreases fertility and makes states more likely to legislate paid sick leave.

Also, sorry I am not familiar with English terminology here, does "sick leave" only apply to women who are sick, or also women who are at home with their sick children? Because if it's the latter, the question is what is the alternative? Suppose you have a sick child and no way to take a break from your work, would you... hire a babysitter? ask the grandma to babysit? become a stay at home mom? Maybe it's women staying at home, or people keeping good relations with their parents (because they need them to babysit their kids) that lead to people having more children.

Also, people can send their sick children to schools and kindergartens (if it is nothing too serious), if there is no other option. And if everyone knows that there is no other option, it becomes socially acceptable. Legislate sick leave, and suddenly it is no longer socially acceptable. Now the children's sickness becomes more expensive for the parents, making children in general more expensive.

Hungary-like fertility policies flatly don’t work.

The explanation seems correct to me. Assuming that we (Slovakia) have a tax system similar to Hungary, it means that only a small part of "the part of your income the state takes from you" is called "tax"; the majority of it is called "social insurance" and "health insurance". But it is not insurance in the literal sense of "your payment is proportional to the risk"; it is just another redistributive mechanism. You have twice the salary, you pay twice the health insurance, but receive exactly the same health care as anyone else. Also, the "tax" part is progressive, that is with a very small income you pay nothing, and then you pay more; but the "insurance" part is a more or less fixed fraction of your income. So "no tax" in practice means about zero difference if you get the minimal wage (because you wouldn't be paying "tax" anyway, only "insurance"), and it is a difference between maybe 40% and 50% of your salary taken away at the high income levels. It's definitely not what it sounds like.

The $30 direct payment and the $100 tax reduction can be important, for the kind of people in Eastern Europe who are not software developers, but the "no tax" part is only designed to sound impressive. Also, you start to get it too late (a few years after your last child was born), so it has no impact on the decision to start the family. And I suspect that starting the family young is the important thing if you want to have many kids.

Here are my policy proposals:

1) Lower the retirement age for people who have more kids, both men and women. The advantage is that this will appeal more to people with long-term thinking.

2) Make a law that companies have to provide part-time work opportunities. I guess this is outside the Overton window in USA, but European countries could do this. The reason is that big families traditionally meant that the mom stayed at home; you probably can't make men stay at home instead because it is still women who get pregnant so that would mean losing both sources of income for the few months; but having both parents work part-time could be functionally equivalent.

If we only let ‘model parents’ have kids there won’t be many kids.

Exactly. In my experience, few people feel like model parents, because everyone made some mistake at some point; and also when people compare against each other, one is better at X, the other is better at Y, both can feel bad about it. For example, my kids are smart and well behaved and know a lot about computers, but maybe they should have better social skills and spend more time in the nature, so we worry about that; and then there are families with extraverted children who have a lot of fun outside, and they are worried about the grades at school.

I lived in homes with 2 different girls who became mothers at age 16, then both had another kid at 18, and another in their 20s (all different fathers). Interestingly, they had daughters only.

There probably is an evolutionary adaptation that influences (at least probabilistically) the child's sex depending on the social situation. From the perspective of reproduction, girls are the conservative choice, boys are "all or nothing". (Like, maybe not now, but definitely in our evolutionary past. A few men were chieftains and had plenty of wives and plenty of kids with them; most men died young in a battle. Women mostly had a few kids each.) So if your children are going to be rich and powerful, it makes more sense to have boys; if they are likely to be poor, it makes more sense to have girls. If you are a young woman who has a problem to keep a partner, yes, your children are likely to be poor. Unless they inherit the genes of some superior alpha male.

(This is not a thing discussed in a politically correct society these days, but the word "interestingly" suggests surprise, and surprise suggests having a blind spot in your model of the world. I am trying to fill in the gap.)

Has anyone ever talked to you about how aging changes your fertility? She said no. No one. Not one person.

Well, I would expect to get yelled at a lot and be called a sexist if I ever dared to mention this out loud. Then someone would mention that they actually know a woman who had a child at her 60s... well, they do not exactly know her in person, but they have read about her in a newspaper... Then everyone would clap.

You can also have the situation in between, when people are vaguely aware that too old is bad for fertility, but can't translate it to numbers, whether "too old" means 30 or 40 or 50 or maybe 60. So they kinda know it, but at no moment of their life they consider themselves too close to being "too old". Because as you get older, your perspective shifts. (Though there seems to be a Shelling point for women in some cultures to panic about marriage when they approach 30.)

The reviewer here noted that the marginal cost of additional children seems to decline. That has been my experience as well. There are dire warnings that two is more than twice as hard as one, or three will be so much harder than two.

In my experience, two is much harder than one when they are small (below school age). They are too incompatible, and you lose the option of "one parent babysits while the other takes a break". When they reach the age that they can nicely play with each other, the costs of parenting become sublinear. Cooking for two kids is not more difficult than cooking for one (unless they have allergies), if you teach something one child, they will be often happy to teach the siblings, etc.

Your childcare regulations must be really harmful if parents respond by having noticeably fewer children.

At some places it is almost impossible to build new kindergartens; the regulations are too strict, many existing ones simply got grandfathered in.

The obvious response to the situation in South Korea would be to opt out of it. Accept that your children might be low status in the eyes of others, but if you can pass on the willingness to accept this and keep going, you inherit the country. [...] Alas, this is not a popular way of thinking.

Maybe there are some risks? Like, you will have a large low-status family, maybe even find a few friends that will do the same thing, so you are now a subculture... and a decade later, suddenly the social service comes and takes away your kids, using "they are so many, and so low-status" as evidence for you being a horrible parent?

(Not sure how much this is actually a risk in South Korea, but I have heard similar stories about Sweden. Doing things differently than the majority of people is... not technically illegal, but... can have consequences.)

Here Johann Kurtz extends the argument that status is the thing that counts, and that the newly low status of stay at home moms is the thing we have to fight. This seems super doable if we decide that we care. The issue is that so far we don’t care enough.

I see two status-lowering forces around me. One is the pressure to be a perfect parent. It is assumed that proper parenting takes a lot of money and time, so if you have too many kids (more than two? depends on whom you ask), you are not giving them enough, which makes you an irresponsible parent. (Perhaps you are too dumb to use birth control properly?) Here the Caplan's book [LW · GW] could change minds. The other force is, uhm, how to put it, feminism. Like, it is not enough to give women the choice of a career, you must also actively make sure that they take it, by shaming those who don't. Let me quote a popular song that used to play here in the radio twenty years ago almost every day, very popular among young wannabe feminists. (translation mine)

"I don't want to be a little domestic woman / and I don't want to get married at all / I don't want to be a wife and have stress / I don't want to have a husband and five children // I don't want to behave sensibly / and say no to all men / and be every Wednesday at a women's club / that is not for me, that is a bad dream // To live like an artificial flower / and wait for someone to pollinate me / to be nice and good / I hope that won't happen to me."

Well, many of them got their wish now, and avoided the dreaded fate of "a little domestic woman". Not sure how many actually enjoy it. (Also, fascinating frame control: a woman having kids is "like an artificial flower", i.e. unnatural. Seriously?)

there is discussion of raising taxes on childless families

This could be one of those ideas that sound good if you haven't spent 30 seconds actually thinking about it. Every person with children was previously childless. Taxing childless people doesn't just mean "taxing people who never want to have children". It also means e.g. "taxing people who don't have children yet -- but want to have them as soon as they save enough money to buy their own house". The more you tax them, the later (or possibly never) that happens.

Possible solution: the money you pay as the childlessness tax is saved in a special personal account, and you get that money back after your first child is born (or maybe 50% at first child, 30% at second, 20% at third).

Those policies [of TMSC] seem fine, but not exceptional, and as others noted company benefits don’t seem to move fertility decisions much. Total compensation matters far more.

Yeah, it seems like the policies should not have such large effect. But maybe it is the kind of situation where providing many things together has stronger impact than providing them separately. Maternity leave and paternity leave and child subsidy and a kindergarten at workplace... and if that starts a wave of people having kids, then you also have many colleagues who have kids, and you are not damaging your career at the company. Basically, the effect of taking all these benefits at the same time, and being surrounded by people who do the same. Also, the child subsidy gives me the impression that the company really wants you to have kids; while e.g. the kindergarten could be perceived as merely removing an obstacle so that you can better focus on your job.

Sometimes the implementation details matter a lot. For example, I had a job that provided a subsidized kindergarten, but... the company paid 50% of the cost, and the kindergarten was twice as expensive as the private alternative next door, so ultimately I paid the same. (The benefit was that having the kindergarten at the same building made the logistic easier.) So it was the kind of a thing that is more impressive on paper than in reality. It would be quite different if the company paid e.g. 75% of the cost, or if the kindergarten would cost the same as the alternatives. I don't know the numbers for TMSC, so I don't know if this was the case.

Why is the Israeli birth rate so high, even outside religious communities? The hypothesis offered here is that those religious communities are integrated with more secular ones

I think it's not just status, but generally how things are set up. A society where people have kids is a society that provides services for people who have kids. Are apartments built small or large? Is it acceptable to bring a child to a restaurant? Does the restaurant offer some toys for the kids, and a place to sit down and play with them? Will people throw a hissy fit if they see someone breastfeeding? If your child is often sick and you take too many leave days to care for it, will you get fired (for completely "unrelated" reasons, of course)? If someone organizes a local rationality meetup, is it possible to bring your kids along?

People with kids can make having kids normal; and when having kids is normal, it is easier to have kids.

if you’re gonna spend a few billion on a harebrained scheme that might not work...

If we are going to literally pay people for having kids, then if the scheme won't work, at least we won't spend too many billions, I guess.

half the comments are saying, yes, you find this woman in a church, obviously.

My advice would be to find a woman who has friends who have many (more than two) kids. Chances are, those will be religious people, but she doesn't have to be.

Step one, don't be an asshole, and make some religious friends... who will know that you are an atheist (i.e. don't pretend to be something that you are not). Especially some religious friends who have kids, but chances are this will happen to many religious friends sooner or later anyway. Then, meet their friends. Chances are, you will meet a woman who is also an atheist, but is exposed to people who have kids, which makes her likely to copy their lifestyle.

(Which is how I met my wife, by the way.)

If you mean the Manhattan Project: no. IIUC there were basically zero Western groups and zero dollars working toward the bomb before that, so the Manhattan Project clearly sped things up. That's not really a case of "centralization" so much as doing-the-thing-at-all vs not-doing-the-thing-at-all.

If you mean fusion: yes. There were many fusion projects in the sixties, people were learning quickly. Then the field centralized, and progress slowed to a crawl.

As a point of comparison - do you think the US nuclear programme was substantially slowed down because it was a centralized government programme?

I think probably we don't have that great a reason not to roll this out to more users, it's mostly a matter of managing UI complexity

I think this is missing the most important consideration: centralization would likely massively slow down capabilities progress.

thanks! added to post