Anyway thanks for engaging, I appreciate the contention and I found it helpful even though you’re so RAWNG.
You are welcome. It has been fun inventing the PERFECT government policy and giving so many 100% CORRECT takes.
(Also remember, even the best possible policy cannot survive execution by an incompetent and untrustworthy government. My policies are only good if they are actually followed.)
If we had ASI we could just let the children choose their own genes once they grow up. Problem solved.
I do not have autism/ADHD/bipolar/dyslexia/dysphoria or a non-heterosexual orientation. If I woke up tomorrow with one of those, I would very badly want it reverted.
However, it seems obvious to me that if being a little schizoid made you a free thinker, able to see things most others can’t see, able to pursue good things that most others won’t pursue, then it does not count as “unambiguous net harm” and the government should have no say in whether you can pass it on. That’s not even close to the line of what the government should be allowed to prohibit.
The question is if it really is their opinion. People often say things they don’t believe as cope or as tribal signalling. If a non-trivial number of people who perceive themselves and their ingroup as intelligent, were to say they anti-value intelligence, that would update me.
Under my system we can ask people with below-average IQ whether they are happy to be below-average intelligence. If they are unhappy, outlaw gene editing for low intelligence. If they are happy, then either allow it, or decide to overrule them.
You want to be careful about overruling people. But intelligence is uniquely tricky because, if it is too low, people are not competent to decide what they want. Plus, people with low IQs have bad objective measures (e.g., significantly lower life expectancy).
The topic has drifted from my initial point, which is that there exist some unambiguous “good” directions for genomes to go. After reading your proposed policy it looks like you concede this point, since you are happy to ban gene editing that causes severe mental disability, major depression, etc. Therefore, you seem to agree that going from “chronic severe depression” to “typical happiness set point” is an unambiguous good change. (Correct me if I am wrong here.)
I haven’t thought through the policy questions at any great length. Actually, I made up all my policy positions on the fly while talking to you. And I haven’t thought about the coalition-building aspect at all. But my current position is that, if we had a highly competent government that could be trusted to reasonably interpret the rules, I would want them to enforce the following:
Don’t allow unambiguous net harm. (Reasonable tradeoffs are fine. Err on the side of permissiveness.)
The best experts on whether “unambiguous net harm” was done are the people who were edited.
Although in rare cases we may have to overrule them, such as the cult example above. This is especially the case if cult members have objectively bad outcomes (e.g., high rates of depression and suicide) despite claiming to be happy.
If we have high confidence that the edited people will have regrets (e.g. based on observing existing people with the condition) we can prohibit the edits without running the experiment. Allowing “unambiguous net harm” edits to be performed for a generation has a high cost.
In some cases I am more permissive than you are. I don’t think we have enough evidence to determine that removing the emotion of fear is “unambiguous net harm”, but it would be prohibited under your “no removing a core aspect of humanity” exception. (Perhaps a generation from now we would have enough data to justify banning it under my rules. But I suspect it has sufficient upside to remain legal.)
Brief reactions to things you said:
some even say they anti-value it [intelligence]
I think a lot of people who say they anti-value intelligence are coping (I am dumb therefore dumbness is a virtue) or being tribalistic (I hate nerdy people who wear glasses, they remind me of the outgroup). If they perceived their ingroup and themselves as being intelligent, I think they would change their tune.
Also, intelligent people strongly value intelligence. And since they are smarter, we should weight their opinions more heavily :P
There’s a big injustice in imposing your will on others
In this case, we are preventing the parents from imposing their will on the future child.
If the parent is compos mentis, then who the hell are you to say they can’t have a child like themselves?
I am the Law, the Night Watchman State, the protector of innocents who cannot protect themselves. Your children cannot prevent you from editing their genes in a way that harms them, but the law can and should.
Fair point, I glossed over the differences there. Although in practice I think very few blind people who wish they could see, would be in favor of gene editing for blindness being legal.
Ok possibly I could see some sort of scheme where all the blind people get to decide whether to regulate genomic choice to make blind children?
Yes, blind people are the experts here. If 95% of blind people wish they weren’t blind, then (unless there is good reason to believe that a specific child will be in the 5%) gene editing for blindness should be illegal.
(Although we might overrule blind people if they claimed to be happy but had bad objective measures, like high rates of depression and suicide.)
By “best policy” I meant “current most preferred policy”.
“prospectively ban genomic choices for traits that our cost-benefit analysis said are bad” is not my position. My position is “ban genomic edits that cause traits that all reasonable cost-benefit analysis agree are bad”, where “reasonable” is defined in terms of near-universal human values. I say more about this here.
I skimmed the article earlier, and read through it more carefully now. I think “edited children will wish the edits had not been made” should be added to the list of exceptions. Also, if we can already predict with high confidence which changes will be regretted, why wait until the next generation to ban them?
By “reasonable” I meant “is consistent with near-universal human values”. For instance, humans near-universally value intelligence, happiness, and health. If an intervention decreases these, without corresponding benefits to other things humans value, then the intervention is unambiguously bad.
Instead of “the principle of genomic liberty”, I would prefer a “do no harm” principle. If you don’t want to do gene editing, that’s fine. If you do gene editing, you cannot make edits that, on average, your children will be unhappy about. Take the following cases:
1. Parents want to genetically modify their child from an IQ of 130 to an IQ of 80.
2. Parents want to genetically modify their child to be blind.[1]
3. Parents want to genetically modify their child to have persistent mild depression.[2]
People generally prefer to be intelligent and happy and healthy. Most people who have low intelligence or are blind or depressed wish things were otherwise. Therefore, such edits would be illegal.
(There may be some cases where “children are happy about the changes on net after the fact” is not restrictive enough. For instance, suppose a cult genetically engineers its children to be extremely religious and extremely obedient, and then tells them that disobedience will result in eternal torment in the afterlife. These children will be very happy that they were edited to be obedient.)
A concrete example of where I disagree with the “principle of genomic liberty”:
Down syndrome removes ~50 IQ points. The principle of genomic liberty would give a Down syndrome parent with an IQ of 90 the right to give a 140 IQ embryo Down syndrome, reducing the embryo’s IQ to 90 (this is allowed because 90 IQ is not sufficient to render someone non compos mentis).
I think our disagreement over ideal policy spills over into practice as well. To you, “the principle of genomic liberty” is the best policy, while to me it is one of many policies that is less bad than status quo.
I think the future opinion of the gene-edited children is important. Suppose 99% of genetically deafened children are happy about being deaf as adults, but only 8% of genetically blinded children are happy about being blind. In that case, I would probably make the former legal and the latter illegal.
I did read the linked comment, and I agree that gouging a child’s eyes out is different. But I don’t see a difference between “immediately after birth, have the doctor feed your child a chemical that painlessly causes blindness” and gene editing. To me, both seem about the same level of soul-damaging. Many of the linked arguments don’t apply at all to the new scenario, and I didn’t find any of the rest remotely convincing, but I don’t feel like taking the time to create a point-by-point response.
It sounds like your goal is to build a political coalition, and I am talking about my ideal policies. I would be happy to accept “the principle of genomic liberty” over status-quo, since it is reasonably likely that lawmakers will create far worse laws than that.
Is your position that at least one parent must be blind/deaf/dwarf in order to edit the child to be the same? If so, that is definitely an improvement over what I thought your position was.
I’m not sure what the difference is supposed to be between “blinding your children via editing their genes as an embryo” and “painlessly blinding your children with a chemical immediately after birth”. The outcome is exactly the same.
I don’t understand your position. My position is:
1. Higher intelligence and health and happiness-set-point are unambiguous good directions for the genome to go. Blindness and dwarfism are unambiguous bad directions for the genome to go.
2. Therefore, the statement “there are no unambiguous good directions for genomes to go” is false.
3. Since the statment is false, it is a bad argument.
Which step of this chain, specifically, do you disagree with? It sounds like you disagree with the first point.
But then you say the fact that “any reasonable cost-benefit analysis will find that intelligence and health and high happiness-set-point are good” is irrelevant to your argument.
So it seems your argument is “even if all reasonable cost-benefit analyses agree, things are still ambiguous”. Is that really your position?
Government intervention comes with risks, but if I had an iron-clad guarantee against slippery-slope dynamics I would not want it to be legal to genetically engineer a healthy embryo to be have Down syndrome, or Tay-Sachs disease, or be blind. It is already illegal to blind your children after they are born, and this is a good thing imo.
I don’t think parents should be required to use genetic engineering to increase their children’s intelligence, health, and happiness set point. However, I don’t think parents should be allowed to harm their children along these axes. (Just like it is already illegal to feed your children lead in order to decrease their intelligence.)
I am quite certain that, even after thinking about it for years, I would still be against feeding children lead or genetically altering them to be less intelligent. That being the case, I don’t think that this topic would be a good use of several years of thinking time.
For what it’s worth, I agree that the state should be involved as little as is reasonable. But if it would be illegal to do something through non-gene-editing means, it should also be illegal to do through gene editing. “You cannot blind your children, unless you do it through gene editing, then it’s totally fine” does not seem to me like a reasonable public policy.
This reminds me of the Scott Alexander article, Against Against Autism Cures.
Would something be lost if autism were banished from the world? Probably. Autistic people have a unique way of looking at things that lets them solve problems differently from everyone else, and we all benefit from that insight. On the other hand, everyone always gives the same example of this: Temple Grandin. Temple Grandin is pretty great. But I am not sure that her existence alone justifies all of the institutionalizations and seizures and head-banging and everything else.
Imagine if a demon offered civilization the following deal: “One in every hundred of your children will be born different. They will feel ordinary sensations as exquisite tortures. Many will never learn to speak; most will never work or have friends or live independently. More than half will consider suicide. Forty percent will be institutionalized, then ceaselessly tyrannized and abused until they die. In exchange, your slaughterhouses will be significantly more efficient.”
I feel like Screwtape would facepalm, then force him into remedial Not-Sounding-Like-An-Obvious-Demon classes.
Even if we give Bill Thurston’s vision problems credit for his mathematical insights, it’s not clear that that makes up for the suffering caused by over 1% of people having strabismus. And even if we decide that the existence of strabismus is good on net, that doesn’t justify the existence of other disabilities. Have blindness or deafness or dwarfism ever caused someone to have unique insights?
My intent with this lesson is to give some pause to those who think that there’s an unambiguous universal “good” direction for genomes to go.
I predict any reasonable cost-benefit analysis will find that intelligence and health and high happiness-set-point are good, and blindness and dwarfism are bad[1].
I think good arguments for “protection of genomic liberty for all” exist, but I don’t think “there are no unambiguous good directions for genomes to go” is one of them.
Deafness is probably also bad on net, but it does allow you to join the tightly knit deaf community. Such tight-knit communities are increasingly rare in our atomized society, so if you really want one deafness might be an acceptable price to pay.
As a LW reader, I find it to be a great summary!
Eliezer wrote an article about this phenomenon, where people think they are writing for a less sophisticated audience than they actually are.
I procrastinated on this, but donated $200 today (the day after the fundraiser ended). I have been reading the site for 4 years, and it has been worth significantly more than $50 per year.
According to the site metrics there are around 15,000 logged-in monthly active users, so if everyone donated $200 that would exactly equal the $3,000,000 target. This made $200 seem like a reasonable number.
Is there literally any scene that has openly transgender people in it and does 3, 4, or 5?
If you can use “they” without problems, that sounds a lot like 4.
As for 3 and 5, not to my knowledge. Compromises like this would be more likely in settings with a mix of Liberals and Conservatives, but such places are becoming less common. Perhaps some family reunions would have similar rules or customs?
After thinking about this some more, I suspect the major problem here is value drift of the in-person Rationalist communities. The LessWrong website tolerates dissenting perspectives and seems much closer to the original rationalist vision. It is the in-person Berkeley community (and possibly others) that have left the original rationalist vision and been assimilated into the Urban Liberal Monoculture.
I am guessing EAs and alignment researchers are mostly drawn from, or at least heavily interact with, the in-person communities. If these communities are hostile to Conservatives, then you will tend to have a lack of Conservative EAs and alignment researchers, which may harm your ability to productively interact with Conservative lawmakers.
The value drift of the Berkeley community was described by Sarah Constantin in 2017:
It seems to me that the increasingly ill-named “Rationalist Community” in Berkeley has, in practice, a core value of “unconditional tolerance of weirdos.” It is a haven for outcasts and a paradise for bohemians. It is a social community based on warm connections of mutual support and fun between people who don’t fit in with the broader society.
...
Some other people in the community have more purely intellectual projects, that are closer to Eliezer Yudkowsky’s original goals. To research artificial intelligence; to develop tools for training Tetlock-style good judgment; to practice philosophical discourse. But I still think these are ultimately outcome-focused, external projects.
...
None of these projects need to be community-focused! In fact, I think it would be better if they freed themselves from the Berkeley community and from the particular quirks and prejudices of this group of people. It doesn’t benefit your ability to do AI research that you primarily draw your talent from a particular social group.
Or as Zvi put it:
The rationalists took on Berkeley, and Berkeley won.
...
This is unbelievably, world-doomingly bad. It means we’ve lost the mission.
...
A community needs to have standards. A rationalist community needs to have rationalist standards. Otherwise we are something else, and our well-kept gardens die by pacifism and hopefully great parties.
...
If Sarah is to believed (others who live in the area can speak to whether her observations are correct better than I can) then the community’s basic rationalist standards have degraded, and its priorities and cultural heart are starting to lie elsewhere. The community being built is rapidly ceasing to be all that rationalist, and is no longer conducive (and may be subtly but actively hostile) to the missions of saving and improving the world.
Its members might save or improve the world anyway, and I would still have high hopes for that including for MIRI and CFAR, but that would be in spite of the (local physical) community rather than because of it, if the community is discouraging them from doing so and they need to do all their work elsewhere with other people. Those who keep the mission would then depart, leaving those that remain all the more adrift.
I welcome analysis from anyone who better understands what’s going on. I’m just speculating based on things insiders have written.
Yes, it’s not a law, so it’s not a libertarian issue. As I said earlier:
Any community is free to have whatever standards they want for membership, including politically-coded compelled speech. But it is not exactly shocking if your membership is then composed 70% of one side and <2% of the other.
By “compelled speech” being a standard for community membership, I just meant “You are required to say certain things or you will be excluded from the community.” For instance, as jefftk pointed out,
The EA Forum has an explicit policy that you need to use the pronouns the people you’re talking about prefer.
Haha, well, at least I changed your mind about something.