LessWrong 2.0 Reader

GeneSmith gave some more details about his background in this episode of the Bayesian Conspiracy podcast: https://www.thebayesianconspiracy.com/2025/02/231-superbabies-with-gene-smith/

Well we have it in cows. Just not in mice.

The lack of good population genetic information in animal models and deep phenotyping of complex behavioral traits is probably one of the biggest impediments to robust animal testing of this general approach.

I think many people in academia more or less share your viewpoint.

Obviously genetic engineering does add SOME additional risk of people coming to see human children like commodities, but in my view it's massively outweighed by the potential benefits.

you end up with a child whose purpose is to fulfill the parameters of their human designers

I think whether or not people (and especially parents) view their children this way depends much more on cultural values and much less on technology.

There are already some parents who have very specific goals in mind for their children and work obsessively to realize them. This doesn't always work that well, and I'm not sure it will work that well even with genetic engineering.

Sure we will EVENTUALLY be able to shape personality better with gene editing (though I would note we don't really have the ability to do so currently), but human beings are very complicated. Gene editing is a fairly crude tool for shaping human behavior. You can twist the knobs for dozens of human traits, but I think anyone trying to predetermine their child's future is going to be disappointed.

The tremendous effort involved in trying to fit the child to the design parameters betrays a lack of belief in the child's inherent value as themselves, and they will be able to tell.

The thing about this argument is you could easily apply it to other interventions like medicines or education. "The tremendous effort involved in trying to fit the child to the design parameters through tutoring and a specialized education program betrays a lack of belief in the child's inherent value as themselves, and they will be able to tell."

Does working hard to give your child the best shot of a healthy, happy and productive life show a lack of true affection for them? I think it shows the exact opposite; you loved them so much that you were willing to go to extra lengths to give them the best life you could. I think this is no different than parents moving to America to give their child a chance at economic opportunity, or parents working extra shifts to send their children to a better school.

But no "super" people can exist in an ethical system where people are of equal intrinsic worth.

The term "super" is not a description of the relative moral worth of these future children. It is a description of their capabilities and prospects for a healthy life.

Good genes enable human productivity and happiness. They don't determine moral worth. That exists independent of ability.

Confering a genetic immunity to HIV on a child might help them out, but it does not, for example, license them to win the trolly problem.

Agreed. I don't get the sense we have any disagreement about the moral worth of people being tied to their genetics.

It's written to explore the principle that there are no bad genes, only genes badly adapted to their environments, and our heroine is an aspiring apprentice baby designer with sickle cell. While it's a challenging position to take, I'm not sure it's a bad guiding principle for somebody made of genes.

I think we need to separate judgment of genes from judgment of the people who have them. You are not your genes. Sure they shape you and influence your experience of the world, but I think a lot of these kinds of books make the mistake of starting with the mistaken premise that our worth IS determined by our genes, and then ask how we can still be equal.

I think the premise is just wrong. It's like saying that you are your trauma, or you are your leg injury. People are much deeper than their experiences or their predispositions, even if all those things have a strong influence on their behavior.

I would love to try this in mice.

Unfortunately our genetic predictors for mice are terrible. The way mouse research works is not at all like how one would want it to work if we planned to actually use them as a testbed for the efficacy of genetic engineering.

Mice are mostly clones. So we don't have the kind of massive GWAS datasets on which genes are doing what and how large the effect sizes are.

Instead we have a few hundred studies mostly on the effects of gene knockouts to determine the function of particular proteins.

But we're mostly not interested in knockouts for genetic engineering. 2/3rds of the disease related alleles in humans are purely single letter base pair changes.

We have very little idea which specific single letter base pair changes affect things like disease risk in mice.

MAYBE some of the human predictors translate. We haven't actually explicitly tested this yet. And there's at least SOME hope here; we know that (amazingly), educational attainment predictors actually predict trainability in dogs with non-zero efficacy. So perhaps there's some chance some of our genetic predictors for human diseases would translate at least somewhat to mice.

We do need to do more thorough investigation of this but I'm not really that hopeful.

I think a far better test bed is in livestock, especially cows.

We have at least a few hundred thousand cow genomes sequenced and we have pretty well labelled phenotype data. It should be sufficient to get a pretty good idea of which alleles are causing changes in breed value, which is the main metric all the embryo selection programs are optimizing for.

You’d like me to point out where you’ve only referenced individual papers for particular ideas and concepts? I can do that but I’d appreciate an answer to my question first if that’s okay.

Glad to hear it!  If you want more detail, feel free to come by the Discord Server or send me a Direct Message.  I run the welcome meetings for new members and am always happy to describe aspects of the org's methodology that aren't obvious from the outside and can also connect you with members who have done a lot more on-the-ground protesting and flyering than I have.

As someone who got into this without much prior experience in activism, I was surprised how much subtlety and counterintuitive best practices there are, most of which is learned through direct experience combined with direct mentorship, as opposed to written down & formalized.  I made an attempt to synthesize many of the code ideas in this video--it's from a year ago and looking over it there is quite a bit I would change (spend less time on some philosophical ideas, add more detail re specific methods), but it mostly holds up OK.

Agreed, this is could be much more convincing, we still have a few shots, but I still think nobody will care even with a much stronger version of this particula  warning shot.

But I think it is wrong to instrumentalize children in this way.

I agree with this as a significant thing to keep in mind, and have written about it here: https://berkeleygenomics.org/articles/Potential_perils_of_germline_genomic_engineering.html#objectification

I think a pretty core lesson from this concern is that communication to parents is very important. Parents should understand:

1. What the traits do and don't mean that they are selecting for, including plausible consequences.
2. What uncertainties exist in the PGSes that are being used (generally lots of uncertainty), e.g. are they accidentally tracking something else as well, or might they perform less well than expected.
3. How much variation is still being left up to chance or environment; pointing out important things that aren't being tracked.
4. That overall the methods will have uncertain outcomes.
5. How to raise kids well regardless of their genomic foundation (i.e. cultural tech for parenting so your kids flourish).
6. That, at least in the scheme of genetic variation, the nudges applied by germline genomic engineering are a drop in the bucket.

And the framing here, that "superbabies" is a category of people who would be generally better to have around, is incompatible with the value of equality.

I agree with this and commented this on a draft. It's not a good way of thinking of germline engineered kids, and inaccurately implies there's some gradation and some single direction of desirability or superiority.

But no "super" people can exist in an ethical system where people are of equal intrinsic worth. A superbaby and a regular baby are both worth unity.

I agree with this and agree it's pretty crucial, and possibly threatened by germline engineering, and possibly threatened by thinking of them as "super". I've written about this a bit here: https://berkeleygenomics.org/articles/Potential_perils_of_germline_genomic_engineering.html#loss-of-human-dignity

For that matter, a hearing baby and a Deaf baby are both worth unity.

Right, also true. And we (society) should be oriented around not abandoning people who become less typical because of germline engineering. https://berkeleygenomics.org/articles/Potential_perils_of_germline_genomic_engineering.html#centrifugal-force-on-marginalized-people https://berkeleygenomics.org/articles/Potential_perils_of_germline_genomic_engineering.html#erasure-of-some-kinds-of-people

Also, I think that, although some may find it immoral, and clearly it has substantial negative first-order consequences, deaf people should be allowed to choose for their children to be deaf. This falls under propagative liberty (https://www.lesswrong.com/posts/DfrSZaf3JC8vJdbZL/how-to-make-superbabies?commentId=ZeranH3yDBGWNxZ7h [LW(p) · GW(p)]).

All this said though.... I think you're going really wrong here. The benefits are just enormous. And while I hold the positions above, I don't think they can justify you imposing on me and my child the requirement to have a chance to be deaf or HIV or sickle cell.

Great write up!

Why don't you do this in a mouse first?  The whole cycle from birth to phenotype, including complex reasoning (e.g. bayesian inference, causality) can take 6 months.