TsviBT

Karma: 10,829

TsviBT 25 Feb 2026 23:36 UTC
4 points
0
on: The optimal age to freeze eggs is 19
Note: some places outside the US appear to have significantly less expensive egg retrieval clinics. E.g. even if you’re traveling to Taiwan from the US, you might save a few thousand dollars even over CNY. There could be other disadvantages, e.g. I don’t know about quality, and there could be a significant language barrier.

TsviBT 25 Feb 2026 23:15 UTC
6 points
2
in reply to: Lydia Nottingham’s comment on: The optimal age to freeze eggs is 19
I don’t immediately see how AI progress would help super much here, though IDK and of course it could be surprising. IDK why IVF is still so expensive. See the thread here w/ GeneSmith + me: https://x.com/BerkeleyGenomic/status/2026173816351674879. But the hypotheses there are things like:
1. You need certifications to do egg retrieval. (Not obviously helped by AI)
2. Meds are expensive. (But they are much cheaper elsewhere, so I’d guess this is a regulatory thing, so not really helped by AI?)
AFAIK there’s currently no practical alternative to this type of egg retrieval (stimulation with meds and then surgically extracting whatever grows). (There are presumably worse alternatives like surgically extracting ovarian tissue, used for fertility preservation across chemotherapy.)

The hope would be in vitro oogenesis. AI might help with some bits of that, IDK (like designing molecules for epigenetic control, or doing big literature searches, or something). But it’s mostly a hard problem, requiring slow experiments, with some experts working on it, which I doubt would just be solved with a magic wand.

(There are several other ways things can improve, e.g. better understanding of how to personalize drug regimens, better ways of freezing and fertilizing, maybe methods for in vitro maturation of immature retrieved eggs, less painful Menopur, etc.; but none of these avoid the big hassle of standard egg retrieval. In theory there could be some much more convenient way to administer drugs?)

So my weak guess would be that waiting is basically a bet on IVG coming in time for your desired time to start a family. I’m hopeful about IVG but I would not bet on that.

TsviBT 25 Feb 2026 23:07 UTC
4 points
2
in reply to: Lydia Nottingham’s comment on: The optimal age to freeze eggs is 19

a ~100-hour endeavor

I’m not sure whether you’re factoring in the cost of feeling bad and not being able to do most exercise, here. If not, then I think that estimate isn’t way off, but my impression is that it’s maybe kinda high. In terms of actual time spent (finding clinics (10hrs?), consultation(1-3hrs?), baseline and monitoring appointments(5-10hrs?), getting meds(1hr?), figuring out and doing injections(5-15hrs?), retrieval appointment(3hrs?)), and assuming I’m missing some costs and vaguely accounting for switching costs / energy costs or something, I’d guess it’s more like 50ish. Which is still a lot of course.

TsviBT 25 Feb 2026 20:26 UTC
4 points
0
in reply to: Cole Wyeth’s comment on: TsviBT’s Shortform
I’m not following (or I disagree). Even without that sort of feature, it would provide Bayesian evidence of lots of things (e.g. “this is feasible as opposed to not feasible” and “here’s some details about how this can work well in some cases”), as well as accomplishing other perfectly legitimate purposes of discourse (such as “making the possibility feel real” and communicating about social intentions such as social reward for leaving).

TsviBT 25 Feb 2026 19:36 UTC
2 points
0
in reply to: Cole Wyeth’s comment on: TsviBT’s Shortform
Plausibly, yeah, or at least many versions. Could you expand a bit more on what you mean / what you’re imagining / what would go wrong? My thought was that if it’s actual people leaving and then talking about that, it would be real + maybe impactful. E.g. think @Daniel Kokotajlo leaving OpenAI. (Don’t mean to imply Kokotajlo is an omnicider, just an example of loudly walking away from a frontier AI company.)

TsviBT 25 Feb 2026 6:20 UTC
48 points
28
on: TsviBT’s Shortform
Off-the-cuff suggestion: help omniciders build lines of retreat. E.g.:
1. Find some people who quite working on AI capabilities for ethical / moral / risk reasons.
2. (if needed) Help them out with getting a good different job not working on AI, and having a good life / community / exciting things to work on.
3. Have them talk about how things went for them after quitting.
4. Make an informational website about how to leave AI research.

TsviBT 24 Feb 2026 5:10 UTC
LW: 2 AF: 1
0
AF
in reply to: Cole Wyeth’s comment on: Glass box learners want to be black box
Ah I see (I’ll try a bit to remember to say “by so and so”).

TsviBT 24 Feb 2026 4:58 UTC
LW: 2 AF: 1
0
AF
in reply to: Cole Wyeth’s comment on: Glass box learners want to be black box
It’s an understandable mistake, but that post is by @Kaarel , not me!

Bioanchors 2: Electric Bacilli

TsviBT24 Feb 2026 1:07 UTC

44 points

1 comment7 min readLW link

TsviBT 21 Feb 2026 22:15 UTC
9 points
2
in reply to: Rafael Harth’s comment on: Rafael Harth’s Shortform
Just registering my wish for more investigation (theory and practice) of pluralistic filtering / karma schemes. Something in the genre of eigenmorality or something. I don’t know how to just pick, or iterate towards, good + feasible voting norms or ideals / shared intentions. (Cf. https://tsvibt.blogspot.com/2025/11/forum-poweruser-forum.html )

TsviBT 20 Feb 2026 7:30 UTC
6 points
4
in reply to: Cole Wyeth’s comment on: “We are confused about agency”
Because of the question of reflective (in)stability in general, I think it’s quite hard to get a handle on anything really important in a mind other than by really understanding mind/agency. Otherwise you have no coordinates for what the mind “really is” in the sense of what elements of the mind will actually stick around.

TsviBT 20 Feb 2026 2:16 UTC
6 points
3
in reply to: Linch’s comment on: Linch’s Shortform
Also, IIUC, if you control all the causal inputs, then there is no regression to the mean. E.g. if there were a trait entirely controlled by genes + controllable environment (not the case usually), that trait would not have regression to the mean if you are selecting on / controlling everything. (And if the environmental part is totally tranmissive (which it generally isn’t), then phenotype selection also gets the causality, so there’s no regression.)

TsviBT 19 Feb 2026 6:52 UTC
6 points
0
in reply to: Wei Dai’s comment on: Genomic emancipation contra eugenics
the many examples of positional arms races throughout human history suggest that this alone is unlikely to work.

Are there one or two examples that are most compelling to you on this point?

I agree, it’s a significant issue. In general, there’s a significant tension in my position: on one hand, I want to push decisions down to the most personal level that’s feasible; on the other hand, you can’t do that with every decision. Positional races are one issue; another issue is simply that some choices aren’t acceptable. A few notes:
- I don’t feel satisfied with the below comments; I’m open to suggestions. Maybe there does have to be some kind of ban, though that seems pretty rough.
- I’m not an absolutist about genomic liberty, and in general these principles shouldn’t be absolute, just like free speech shouldn’t be an absolute right. In other words, it’s fine for the state to intervene in certain genomic choices (only, or almost only, to prevent certain active changes). A fortiori, I’m not infinitely opposed to these other influences playing a substantive role. I think this is a necessary complication to the sort of norms I’m proposing. See https://berkeleygenomics.org/articles/The_principle_of_genomic_liberty.html#exceptions-to-the-genomic-liberty-principle
- My current weak guess for the least bad implementation is to keep the state out of it, but to have really strong professional norms for reprogenetic clinicians against pushing height really high. [ETA: glancing back at the essay, I also said that tall sons could be an exception for social stigma, i.e. it would be fine to stigmatize that choice. I guess at some point I could list a whole bunch of genomic choices and make up ideas of what sorts of pressures I think would be ok in that case, in order to figure out how / whether I’m actually applying my stated ideals.]
- Generally, clinics should be refusing to do really dangerous genomic changes. So they should at least refuse to intentionally make a future child have a high chance of being, say, >7 feet (or whenever is some reasonable cutoff that carries too much disease risk). This doesn’t prevent some racing, but hopefully would at least provide a stopgap. If the really dangerous uses are being prevented, then the outcome doesn’t seem that bad to me in the grand scheme of things, compared to the overall benefits of reprogenetics, but it’s pretty bad.
- On the other hand, letting the race get going could be adding more fuel to the fire because of FOMO and race dynamics. That argues for a stricter moratorium.
- Drawing a line about height for purposes of regulation seems a bit awkward, or at least it would need some working out. How tall is too tall? What probabilistic distributions over heights are acceptable? Do we use changes in expected height, or change in probability of being above a threshold? How does this interact with the parental expected height?? (A natural answer would be about disease risk, but again, that still leaves a fair bit of headroom as it were.)
- Enforcement would be awkward because there are many tradeoffs. What if a couple wants to select an especially tall son, but that’s because that embryo is significantly healthier (less diabetes risk, say) compared to their other embryos? (This problem can be innovated away; strong reprogenetics removes tradeoffs between biologically separate traits.)
- Society should change to be somehow less heightist, lookist, etc., but this is pretty pie-in-the-sky.
- Society should change to be somehow less arms racey. I’m not sure how pie-in-the-sky this is.
- I’m imagining that there’s one or a few professional ethics/standards that reprogenetics clinics can hold to / qualify for. These would prohibit a clinic from intervening to make a future child to be likely to be 7″. This is de facto somewhat-centralized control. However, there are a couple things that make it relatively less bad compared to state control:
  - You can make a clinic that is out of norm. It will probably be a worse clinic, because fewer clinicians are willing to work with you and you’d have fewer clients, and clients would have to pay more / travel further. But at least it’s possible to get the out-of-norm treatment if you really want it. This is kinda like taxing cigarettes rather than banning them outright or being totally laissez faire. It feels a bit like “driving at half speed”; isn’t that genomic choice either acceptable or not? But I think maybe soft disincentives are better e.g. because they set up a regime that is at least somewhat responsive to the intensity of people’s private desires, simply because if you really want it you can still get it.
  - I think problems like the tall sons race are quantitative, so preventing most people from doing that addresses most of the problem. (Though it could be that you set off the arms race.)
  - It’s sort of “direct democracy” in a sense. If literally all clinics refuse to provide a service, that’s a strong indication that the service shouldn’t be provided—significantly stronger than some state process making that decision. In other words, it seems quite difficult to fully ban something unless it’s really bad.
- A totally different kind of idea, sorta in the vein of the previous sub-bullet point, would be to take some sort of set of democratic direct referenda, specifically of the form “should the state ban genomic choice X”. And you only ban it if 95% of people say yes (or something).
- In general, I would want society to get long-term feedback about how things are going with reprogenetics, and then course-correct if necessary. (Hence some of the importance of multi-generational feedback on reprogenetics: https://berkeleygenomics.org/articles/Genomic_emancipation.html#habermas-and-multigenerational-feedback )

TsviBT 19 Feb 2026 2:16 UTC
2 points
0
in reply to: Cole Wyeth’s comment on: “We are confused about agency”

I don’t agree with this. It is often easier to design some X that works predictably, than to verify whether an arbitrary X works. For example, we can write verifiable programs, but we can’t verify arbitrary programs.

Your interpretation of this paragraph is reasonable, and I agree that the statement you’re hearing is incorrect; but rereading the whole section ( https://www.lesswrong.com/posts/NqsNYsyoA2YSbb3py/fundamental-question-what-determines-a-mind-s-effects#The_word__a_), I largely stand by what I wrote—that paragraph functions in context as part of a development of thought that is dealing with the tension between (1) we just need to design one mind that works, and (2) you can’t really do that without understanding minds somewhat more generally (IDK how much).

This statement is narrower and more useful than “We are confused about agency.”

I think it loses a pretty important + useful idea, which is that it’s NOT the case that:

we understand agents and [their goals and how they relate to their goals or something], as long as you’re not trying to change [their goals and how they relate to their goals or something].

TsviBT 19 Feb 2026 2:02 UTC
6 points
0
in reply to: Cole Wyeth’s comment on: “We are confused about agency”
“What determines a mind’s effects” does not seem likely to have a clean answer in general.

I can quibble with this, and of course it’s an open question (e.g. cf. https://www.lesswrong.com/posts/NvwjExA7FcPDoo3L7/are-there-cognitive-realms). But I certainly take your point that you maybe don’t need to understand all or most minds. (I may have written something contradicting that in the “fundamental question” post; if I did, then I was mistaken or wrote unclearly or something.)

I am not sure that there is a finite set of concepts for understanding agency. The problem with a theory of agents is that agents invent theory, so agent theory keeps pulling more stuff into itself.

I think I agree with this. Cf. https://www.lesswrong.com/posts/nkeYxjdrWBJvwbnTr/an-advent-of-thought

I guess that the point of agent foundations is to focus on only the core theory of agency. But what characterizes the core theory?

I would say that alignment is pretty close to the core theory; let’s say, ¹⁄₄ of core theory is alignment, and ¹⁄₂ of alignment is core theory (numbers made up, but just to give relative sizes or something). Or IDK haha. But what I mean is, core theory would tend to be
- Elements that an individual mind converges to over time. (Yes, there are many elements X that may not converge, e.g. because the mind is permanently deranged, or because X is entangled with ongoing / infinite creativity; and in some sense all X are entangled with ongoing creativity, and therefore don’t fully converge in terms of their full meaning for the mind… but like, still, there’s obvious senses in which things do converge.) (Some of these elements are mind-general things like Bayesian reasoning; others are mind-specific / value-laden / cognitive-realm-specific. The latter are converged to by choice (choice of what to be).)
- Elements that determine the mind’s ultimate effects.
- Elements that are stable under reflective self-modification. Cf. https://www.lesswrong.com/posts/Ht4JZtxngKwuQ7cDC?commentId=koeti9ygXB9wPLnnF
If something tends to get self-modified away, it’s probably not core.

You might say that nothing is stable, everything gets self-modified away? We could debate that. But I have a perhaps not very accountable sense that it is possible for me to decide some things about myself permanently, even though I’m fairly gung-ho (in the long run) about radically changing / growing / transcending myself. E.g. I think I can decide to never pointlessly torture a person—I mean presumably someone could fairly feasibly mess with my head a bunch to change that property, but I mean that left to my own RSI devices I would never change my mind on that. Do you agree that some things can be stable like that?

TsviBT 19 Feb 2026 1:30 UTC
4 points
0
in reply to: Cole Wyeth’s comment on: “We are confused about agency”
Mhm. I think those concepts are quite central to what an agent/mind is.

TsviBT 18 Feb 2026 22:47 UTC
2 points
0
in reply to: Cole Wyeth’s comment on: “We are confused about agency”

“What determines a mind’s effects” does not seem likely to have a clean answer in general. What determines the trajectory of embryonic development? What determines the trajectory of an ecosystem? In your words: “It may be that there is no answer. It may be that no small assembly of small elements of the mind comprehensively determines the mind’s effects.”

Yeah, not in fully generality. IDK how generally it should have an answer. I think embryonic development, and especially an ecosystem, are bad comparisons because they aren’t sufficiently mind/agent-like. The embryo is executing a bunch of narrow adaptations (various homeostats, and similar, I imagine); it isn’t on a very determined trajectory, and it can be fairly easily disrupted, and it won’t pull itself back onto the trajectory. Even more so for the ecosystem. A mind/agent, on the other hand, can aim itself—at least to some pretty large degree.

TsviBT 18 Feb 2026 22:40 UTC
2 points
0
in reply to: Cole Wyeth’s comment on: “We are confused about agency”

Also, I have a bundle of intuitions and models from theoretical computer science that suggest there should be not be elegant complete theories for bounded rationality.

I agree with this for at least two reasons, one of them being “The concrete is never lost” (https://tsvibt.blogspot.com/2025/11/ah-motiva-2-relating-values-and-novelty.html#values-require-reference). But I don’t view this as bearing much on questions like “can we get a major compression of the important elements of the domain”. All complex, context-embedded things will have a bunch of complication that doesn’t reduce, but that doesn’t mean you can’t much much more deeply understand things. E.g. the OS of a laptop computer ~irreducibly must have complexity to interface with mouse, keyboard, screen, wifi, audio, different memory elements, etc.; but it’s still very principled (written in multiple layers of languages, using algorithmic ideas for high leverage) and is much more understandable than a huge pile of machine code that some compiler spits out (or however that works).

TsviBT 18 Feb 2026 22:33 UTC
2 points
0
in reply to: Cole Wyeth’s comment on: “We are confused about agency”

Build a unified picture of agency from the perspectives of the fields that formally describe it, which “explains” agency about as well as evolution + genetics “explains” biology. Perhaps a little worse.

Oh. Well that’s kind of a low bar. Maybe we don’t disagree about this then, not sure. We agree that it’s not nearly enough for alignment, right?

TsviBT 18 Feb 2026 22:32 UTC
4 points
2
in reply to: Cole Wyeth’s comment on: “We are confused about agency”

Before Lamark biology seems to have lacked unifying principles to address its central questions.

Before Linnaeus it lacked unifying principles.… But after Linnaeus, we have categorized everything.

...By which I mean to suggest, I think we might be miscommunicating about the questions at hand… If a post-Linnaean biologist is looking back and saying “sure, we had lots of information about species, but we didn’t have a conceptual scheme; now we do, so we are no longer confused about biology”, then the problem isn’t that ze is wrong that a big advance happened, but rather that ze is… failing to imagine that there could be major future insights, after which the prior state of knowledge would seem fundamentally conceptually impoverished.

I think that a biologist in 1900 is confused about several fundamental things about life. For example, they don’t know about GRNs. My guess would be that biologists in 2026 are also confused that way. I wouldn’t know specifically what they are confused about, but for example they may lack concepts of [metastable self-reinforcing gene regulatory states] or [bundle structures within the space of biological functions corresponding to characters such as cross-species-hands and cross-species-eyes and cross-species-T-cells]. (You would definitely find papers about both of those topics, and probably lots of other interesting theoretical topics in biology, but I’m saying it would be unsatisfactory, where it could eventually be satisfactory.)

TsviBT

Bioan­chors 2: Elec­tric Bacilli

Bioanchors 2: Electric Bacilli