Man’s best friend

Steve Hsu has suggested that we might be able to vastly increase human intelligence via genetic engineering, by finding the many small-effect alleles that affect IQ and engineering embryos with all-plus versions of those alleles – leading to people whose IQ would be 30 standard deviations higher than us yokels.

There are some possible problems with this scenario. The new & improved humans might not like us. They might be crazy: the risk for certain kinds of crazy increases with the number of plus alleles. Smarter than human, they might realize that life is not worth living, then rush out and do us all a favor.

It risks the death or domestication of the human race.

Now presumably intelligence in dogs is also highly polygenic. We could find those alleles and engineer super–dogs that were 30 std smarter than dogs today.

This might be a better idea. Dogs actually like us, which is not at all guaranteed for supermen.

Of course, they’d dominate internet poker, but that’s a a small price to pay.

This entry was posted in Uncategorized. Bookmark the permalink.

92 Responses to Man’s best friend

  1. Sandgroper says:

    Terence Tao has one brother who is an autistic savant. He is an international chess master and has a double degree in mathematics and music, but I’m guessing that for all practical purposes (and this is going to sound bad) in doing some kind of useful job, he is probably of no use to anyone.

    And not all dogs like everyone. In fact, a lot of dogs would be useless in the jobs that they do if they did like everyone. American data show that most human fatalities from dog attack are attributable to pit bulls. Rottweilers come second. A super smart pit bull or rottweiler could be a truly terrifying animal.

    At least Terence Tao’s brother doesn’t go around biting people. At least, I don’t think he does. Winning chess games seems pretty harmless, although utterly useless for all practical purposes.

    • “although utterly useless for all practical purposes.”

      Hard to say if that’s really true. Even the least worldly of Mathematicians can be very useful.

      Speaking of dogs: Sandgroper, do you happen to be a Dingo?

      • Sandgroper says:

        Not if he is incapable of communicating with other humans. I’m not dissing Mathematicians, I hold them in high regard, and consider that a good working knowledge of Mathematics and Statistics is absolutely essential in any sort of professional level occupation now. But autistics generally are not known for their ability to work with, or even communicate with, other humans. And a Mathematician sitting locked in a room by himself can only be productive to some very limited point, unless he truly is a genius, which autistics typically are not.

        I am not personally a Dingo, no, but I had a pet Dingo from when I was 7 years old until she had to be put down because she had advanced cancer when I was 21. She was the most marvellous companion for a boy growing up, and I have refused to have any other dog ever since. So I guess I might behave a bit like a Dingo – by the time she was 2 human years old and I was 9, the role reversal thing had happened, and she became the one who took care of me and kept me in line, rather than me taking care of her when she was a pup.

        The only other dog I had a lot of time and respect for was a Kelpie sheepdog (who incidentally are alleged to have been bred partly from Dingoes, but I have never seen that proven genetically) – she was a great companion for going out hunting, but would brook no kind of affectionate contact at all. I daren’t try to pat her, she would have taken my hand off. She made it very clear that she was not my friend, she was just my hunting partner, so long as she and her pups got their fair share of what I shot and butchered. But in terms of eye contact communication of what she wanted me to do when we were hunting, that dog was absolutely brilliant.

        • I thought it might be you, the rambling autism of your reply proves it. I’m a listener of your podcast, kudos.

          Please see if you can get Greg Cochran on, Greg’s previous podcast with James Miller about Stalin’s unrecorded biowarfare programs ended too soon, and needs followup.

    • MawBTS says:

      Even Tao himself seems a little “off”. I don’t know how to put it. Watch his interview with Colbert. Not autistic, but not fully normal, either. He’s like if someone made a robot that can pass the Turing test, and Terence Tao is the unreleased alpha of that robot.

      Most celebrated high IQ humans are kind of dysfunctional.

      Rick Rosner is a bouncer/stripper who spent 10 years in high school with a series of fake IDs (apparently to live out some fantasy of being a cool guy, a fantasy he would never quite achieve). Now he works as a comedy writer.

      Chris Langan is basically the Timecube guy. Apparently he’s blocked from Wikipedia for disruptive editing. He works as a bouncer.

      Kim Un-Yong has spoken at length about how depressed and isolated his intelligence makes him feel.

      Marilyn Vos Savant seems socially normal, but she hasn’t done much beyond write a “Dear Abby” type advice column. Either her intelligence has been overstated or she’s an underachiever.

      Ron Hoeflin lives in a flat above a laundromat. Or so says the last scrap of biographical information I can find on him.

      William J Sidis is the biggest mystery of all. You might have heard that his IQ was 250-300 – it’s now thought that this was actually his placement on the list of people who took the exam. Nobody knows how smart he was. We do know that he died in a hovel at age 46.

      Greg once said something like “stupid people believe x, smart people believe y, really smart people believe x”. Mutatis mutandis, it seems generally true that stupidity is dysfunctional, intelligence is functional, and extremely high intelligence is dysfunctional.

      • Sandgroper says:

        During all the years I lived alone, I could have done with living above a laundromat; preferably a laundromat next door to a Pizza Hut.

        But yeah – where you need to have people working cooperatively as an integrated team, extremely highly intelligent people can be very disruptive. Unless you can lock them in a room somewhere and just pay them to keep producing innovative ideas (a bit like Greg, now that I think about it), they can definitely be too disruptive and more trouble than they are worth. I’m joking about Greg, btw – I’ve seen him on videos and he’s nowhere near as disruptive, disagreeable or uncooperative as the kind of people I’m thinking about.

      • Marduc says:

        “Even Tao himself seems a little “off”. I don’t know how to put it. Watch his interview with Colbert. Not autistic, but not fully normal, either. He’s like if someone made a robot that can pass the Turing test, and Terence Tao is the unreleased alpha of that robot.”

        He looks like he is trying to talk to a child but does not have much experience talking to children.

        Which is probably a reasonable description of what he is doing.

      • Marduc says:

        “William J Sidis is the biggest mystery of all. You might have heard that his IQ was 250-300 – it’s now thought that this was actually his placement on the list of people who took the exam. Nobody knows how smart he was. We do know that he died in a hovel at age 46.”

        Sidis was hot-housed by his parents (who were Ashkenazi Jewish professors and doctors) in an attempt to prove that nurture trumps genetics in determining adult intelligence. Since his parents were clearly stupendously intelligent themselves, they had to be able to claim he was not just a very intelligent man, but one of the smartest who have ever lived. They were wrong about nurture, so Sidis was probably a few standard deviations above average but not as smart as they claimed he was.

        • MawBTS says:

          Yes, many claims about Sidis come from his family and aren’t trustworthy.

          Boris Sidis claimed his son could speak eight languages (Latin, Greek, French, Russian, German, Hebrew, Turkish, and Armenian) by age 8…but what does “speak” mean? Could he hold actual conversations in those languages? Or did he just he just learn a couple of Turkish words, and his father thought “yep, he speaks Turkish.”

          There’s evidence that he was exceptional, perhaps very exceptional (passing a Harvard entrance exam at 9, for example), but yeah, lots of question marks there.

          Perhaps his most enduring accomplishment is this 300 page book on streetcar transfers, often cited as the most boring book ever written.

        • Cpluskx says:

          Btw you can’t have 250-300 iq. Pop size/avg iq/sd isn’t large enough.

      • Chasm says:

        That reads like a list that could be called “anecdotal examples of very high-IQ people who are socially maladjusted.” What about the thousands (or tens of thousands) of IQ 160-ish people who lead normal and largely quiet lives as professors, quants, analysts, lawyers and the like? Publicity-seeking, high-IQ oddballs like the ones you mention don’t represent (very) high-IQ people in general; in fact, their weird publicity-seeking behaviors qualify them as outliers.
        Also, I looked up that Terence Tao clip on Colbert and he didn’t seem Aspergy to me; he has a bit of an odd accent, as someone who grew up in Australia to Chinese parents, but otherwise he seemed pretty normal.

        • JayMan says:

          The Terman study is the only one that comes to mind that shows statistically that certain forms of dysfunction are more common in people with very high IQs. That said, the majority of the super-smart are quite normal.

          This needs to be looked into more.

          • gwern says:

            The Terman kids weren’t very smart, for the same reason that the Hunter Elementary kids* underperform their ‘150 IQ’: a noisy test in childhood + inherent age change = severe regression to the mean. (I should hardly need to point out the role of measurement error to JayMan.) Hunter High School/SMPY/TIP are much better in this respect: that few years’ delay in testing makes a big difference in reliability of the test scores, and it shows in the accomplishments of their alumni.

            Any dysfunction is probably for other reasons (eg that Australian study where the subjects were recruited from… child psychologists; yeah, that’s sure going to be an unbiased sample), because the genetic correlations of education/intelligence with all mental illnesses and with bad personality traits are negative except for a small one with autism spectrum – take a look at the references in my handy dandy new WP article:

            And as far as that ASD correlation goes, because it’s so expensive a disorder, anyone doing embryo selection/iterated embryo selection/genome synthesis is going to be doing multiple selection (, which is much more efficient in the first place, and will be placing considerable weight against ASD and some weight on good personality traits like agreeableness and low neuroticism.

            Having company will also help a lot. Given the economics, iterated embryo selection and genome synthesis won’t make any sense unless you are creating hundreds or thousands of modified embryos. One can hardly be an ‘outsider’ when there are so many peers.

            So I’m not particularly worried about them turning out to be mean as snakes.


            • et.cetera says:

              To call the “agreeableness” dimension of personality “good” without any sort of qualification is sloppy thinking. If you want your kids to succeed in life, it’s probably best they are more disagreeable than average, not more. Actually, high agreeableness is so damn bad for you and anyone around you that I am compelled to ask: are you actively trying to screw people over or is it merely accidental?

              A world full of extremely intelligent, overbearing mothers who treat everyone around them like helpless babies — there are very few more hellish, draconic futures I can think of than that.

              • gwern says:

                High agreeableness is not that bad for you, and comparing it to helpless babies and asking if I am deliberately trying to screw people over is absurd. Psychopathology and other issues go with low agreeableness, not high. Sure, in the Terman study lower agreeableness did predict higher income, but screwing over other people and making money by being mean as a snake is the exact opposite of what we want extremely high IQ people to be like, for the reasons in OP. They’re going to make plenty of money without playing negative-sum games.

          • Actually, high range IQ test designer, Paul Cooijmans claims that the higher the IQ, the less mental problems one has, according to his data of people taking his tests.

            • Sandgroper says:

              Cooijmans is moronic on the subject of human evolution, so I wouldn’t take anything he says as Gospel. Not saying he’s wrong, but he is disturbingly way off mark on some pretty obvious, known, well established things. Like he thinks farming was discovered in Europe contemporaneously with other regions. Duh. No intelligent, reasonably well read person could be excused for thinking that.

      • Anecdotes are not data. If you read reviews of high ability people, this is not the pattern one finds. It’s your lucky day because a new, great review just came out.

        David Lubinski – From Terman to Today: A Century of Findings on Intellectual Precocity

        • MawBTS says:

          Thanks – interesting reading.

          Most of those studies seem to focus on the top 1% or thereabouts. Can we say this pattern holds all the way to the asymptote – the top 0.01%, and the top 0.0001%, etc? My impression was that it’s very hard to get a sample of such individuals to study.

        • Bob says:

          There are lots of anecdotes though. If you read biographies of significant figures like Newton, Henry Cavendish, John Dalton, Kant, etc., many of them come across as weird shut-ins who’d be described as “autistic” today.

          • Jim says:

            Henry Cavendish was certainly pathologically shy but people like Gauss or Dirichlet seemed to be pretty normal. Kant was supposedly a brilliant lecturer.

            Despite his shyness Cavendish served in some official positions without any apparent great problem.

      • I wrote often about Sidis a few years ago, thinking that his story needed debunking. Short version: He was smart, but not any smarter than a few folks in tow here. Nothing like that 250-300 score. No one is. How he got the reputation is interesting.
        Marilyn Mach vos Savant’s claim has some weaknesses. So does Chuck Schumer’s 1600 SAT claim.
        I used to correspond with Hoeflin and few of the guys way up the list in the 1980’s but have dropped out of that realm. Not all were strange though all were at least a little eccentric.

      • Sandgroper says:

        To be fair to Terence Tao, I thought he behaved commendably well considering that Colbert was behaving like a rude idiot. He even laughed politely at Colbert’s moronic jokes.

        The one slight strangeness is that he talked too fast. But he might not always do that, or it’s possible that he just has so much information to impart in a tolerably short amount of time that he has to do that. Or maybe his thinking is so far ahead of his mouth that he is rushing to keep up.

        A little ‘off’? No, I wouldn’t have said so – he seems to be a tolerant, sociable, good natured kind of person. Under the same circumstances, I think I would have wanted to tell Colbert to shut, stop interrupting and quit playing the idiot; or else just smack him in the mouth.

        If you grow up with a sibling or other close relative that clearly has ‘problems’, which I assume his autistic brother does, it can be a fairly humanising and humbling experience.

        I know intelligent people who are less intelligent than Terence, who behave a damned sight worse and far more ‘off’.

        He doesn’t have an Australian accent. I would put his accent closer to Oxford or some other shade of upper class English, with a hint of Cantonese on the side. Mind you, people from Adelaide tend not to have very strong Australian accents – it is the only major city in Australia that did not develop from a convict colony, it was built by free settlers, so it has always been a bit snooty. It also has a pretty good medical school, which I presume was the attraction for Terence’s father (a surgeon) when they migrated.

        But now I’m sounding like a rambling autistic.

    • Chess, chess, can ruin everything =)

    • Luke says:

      Excellent example. Adding intelligence, no matter how much to a human being doesn’t change the fact that they’re still human beings. Bounded by physical limitations, needing the same things every other human beings require, vulnerable to the same circumstantial game of luck that is life.

  2. Anonymous says:

    As a physicist, Hsu should know that almost everything is locally linear and globally nonlinear.

    • ckp says:

      “To first order” — that all-purpose incantation

    • masharpe says:

      I think the point of doing the global estimate is to demonstrate that it’s unlikely that the smartest existing humans have achieved some sort of maximum human intelligence. The actual maximum is probably not +30 SDs, but it’s also probably bigger than the +6-ish SDs of the smartest existing humans.

  3. MawBTS says:

    Of course, they’d dominate internet poker, but that’s a a small price to pay.

    You’re funny. I hope they kill you last.

  4. I don’t think they’d be able to hold the cards. They’d be spilling them everywhere. No, if dogs used their intelligence like the only other example we know of – us – they’d just watch cat videos.

  5. namae nanka says:

    “They might be crazy: the risk for certain kinds of crazy increases with the number of plus alleles. ”

    Lewis Terman and Leta Hollingworth’s studies on gifted students diverged on this aspect, maybe because Hollingworth had smarter gifted students who would more acutely feel being ‘Outsiders’. 2-3 SDs from mean is good but go further out and you start running into problems.

    Maybe there’s another difference in that his study took place in California while hers was in New York so there might have been different ethnicities in play.

    • JayMan says:

      “Maybe there’s another difference in that his study took place in California while hers was in New York so there might have been different ethnicities in play.”

      That’s another factor I wonder about. What are the super high IQ from different ethnic groups (or their equivalents) like?

    • another fred says:

      Maybe there’s less crazyness because bright/different people aren’t getting beat up as much.

  6. Anonymous says:

    Why bother with dogs when we could improve much better species? Cats love us so much more…

    Think of the potential!

  7. Agent J says:

    Koontz’s novels with superintelligent dogs have an interesting take on what they’d be like. They despair at their lack of ability to communicate, and at being so alone (though with many intelligent superdogs around, that part might not be so bad). Plus, once they realize they have lifespans typically 10-15 years, that would be another source of sadness. Look at what people do as a result of knowing our mortality; is it even ethical for us to do that to another species?

  8. pyrrhus says:

    I think there are probably innumerable ways in which the law of unintended consequences could bite Steve Hsu’s idea in the ass….The notion that we understand how Mother Nature works well enough to take over from her is an extremely dangerous one, and is constantly refuted by research.

  9. The Z Blog says:

    I’ve written a lot about this:

    For starters, normals are suspicious of the super intelligent. People have no trouble respecting those above them in the normal range, but once someone moves beyond the normal range, average people suspect they lack the moral coding that defines normality. It’s why Hollywood loves the evil super genius. The audience believes it.

    Once you lurch into the hyper-intelligent, like evolving AI, the most likely outcome is nothing like the futurists imagine. The best outcome is the super intelligent robots quickly get bored and leave to explore the universe. The next most likely outcome is they go insane and destroy themselves. There’s also the strong chance they create an artificial dream state for themselves. In other words, the super intelligent robot future will look like an opium den for retired terminators.

  10. Greying Wanderer says:

    what if high IQ isn’t a sufficient condition for whatever you want high IQ for?

    and instead it’s a necessary but not sufficient reason and there’s an x factor involved as well?

    for instance i want it for fusion and FTL and if people focus solely on IQ when it’s IQ + x then it’s not gonna happen

    so hopefully there’s another Steve Hsu somewhere looking for the x.

  11. Greying Wanderer says:

    A second thought on the too-smart-to-function people is the internet should make it possible for people like that to cooperate in a group while still being on their own.

    For example this Tao person might be socially dysfunctional to the point where he can’t fit into a normal or even academic environment but give him a monk’s cell on a campus somewhere connected to one or more research groups and sent math problems to work on on his own he might be fine.

  12. Paul Conroy says:

    I wonder if extremely high IQ, 250+ sd15 say, would select for non-domesticated traits. By which I mean that from the time of large brain Neanderthals, humans have been undergoing antagonistic selected for both efficient brains (increasing IQ) and domesticated traits – usually seen as lowering IQ in other species.
    So what if an IQ 250 guy is a smelly, shaggy, unsociable, wild man?!

  13. jb says:

    Is there any good reason to believe that an “all-plus” human would really be 30 standard deviations above average? A lot of those small-effect alleles could be functional duplicates, and wouldn’t sum linearly if they were put together. As an example, East and West Eurasians have different skin lightening alleles, but you don’t get people who are super white when you cross them. They’re independent, but not additive.

    For example, you might have a thousand alleles that affect only 50 different pathways. Imagine a simple model where each of those pathways can have a value of either 0 or 1, and where a single positive allele on a given pathway bumps its value from 0 to 1, but having more than one positive allele on that same pathway has no additional effect. A total of zero across all those pathways makes you a complete moron, while the genetically fortunate real-life super-genius might have a total of, let’s say, thirty. It seems to me that if plus alleles were rare enough, such a model would look additive when you are doing statistics on more or less average people, and yet the perfected plus-50 superman might not be as much of an improvement over someone like Terrance Tao as one might imagine. (Note that I’m making no claim that the specific numbers I’ve use here are reasonable; I’m just trying to generalize the point I made with East and West Eurasians. Feel free to come up with better numbers if you think the underlying argument makes sense.)

    In any case, the proper route is not to go for everything at once. What you really want to do is produce a large number of people who are four or five standard deviations up, and let them figure out the next step.

  14. Pingback: Genetic engineering | Candide III

  15. Matt says:

    Smart dogs would have a whole different life history. Need to engineer them for longevity as well as smarts, so they have enough time to learn enough to be useful. Human language specific variation might also help. Or perhaps those factors are just bs – Canis lupus superior can brute force the language problems without any of our specific adaptations with enough g, and even with mayfly lives by human standards, if they’re smart enough they’ll be productive?

    I don’t like the idea though. Say they’re smarter, don’t like people (imagine they like us because we look smart to them and their species dominance hierarchy puts the smartest at the top) and they win, then I’d probably rather our successor species was recognizably human descended than dog descended.

  16. j says:

    Greg is right to worry that 30 SD folk may not be fond of us. Did our ancestors love Neanderthalers? Maybe if well done.

  17. engleberg says:

    We could breed for ‘working temperament’ as well as intelligence. Not the same as agreeableness.

  18. Sean says:

    Artificial intelligence would make the current elites’ main advantage worthless within a generation, so the smart move for the US intellectual elite (few of whom are scientists) is to try and cast doubt on the moral and scientific integrity of genetic engineering .Which is what they are doing. My bet is genetic engineering for IQ is going to be anathematized and disbelieved in by the American elite for the foreseeable future, because that is to their advantage.

    The US will have the edge with artificial intelligence, (Kurzwell’s position at Google shows that is where serious money is going) although IE it is probably impracticable for now. The Chinese elite will go with genetic engineering.because that is easiest way to steal a march on the US. How capable the Chinese are of making GE for massively boosted IQ all on their own is an open question, but if Hsu is correct it might be well within their capacities to do it in a generation. And by that time China will not only have unprecedented super-brains, it is likely going to be something like a giant Hong Kong, with the attendant capacity to outspend the US on artificial intelligence research. Game over.

  19. whyteablog says:

    I bet you $20 that if anyone is dumb enough to try this, the super-babies don’t even make it to term.

    • RaceRealist says:

      And if they do, they’d die almost immediately I’d assume. We obviously have no idea what would occur. It’d just end up being a huge waste of money (not that the people who would be able to afford it would care).

      What if they do make it to term and survive? What do you think would happen then?

      • whyteablog says:

        They probably would be insane on the off chance they survived.

        “I have an idea guys. Let’s take the most complicated part of the body and change it as much as we possibly can. And we’re not stripping as many functions as possible, we’re adding them.”

        The best case scenario is having them die, which is a waste of money and ethically wrong, if ethics are real or relevant. Below that, you have genius lunatics running around, which sounds pretty dangerous. Worst case scenario is that, against all odds, they’re otherwise normal people, in which case Greg is probably right- they’d take over the world. They would correctly view us as incompetent and themselves as our natural successors.

        Anyone who deliberately creates something that renders its creator obsolete… get that guy away from the control box.

  20. masharpe says:

    I assume for this experiment that you’d want to start with dog breeds that already have high intelligence. Given that those breeds have been bred for intelligence over a long time, I wonder whether there’s much potential left in their gene pool. Might be inexpensive to check by creating a test and seeing if there’s enough variability of results among different individual dogs of one breed.

    How many dogs would need to be measured to create the models? For humans, it seems necessary to have at least hundreds of thousands or even millions. If that’s the same for dogs, collecting the initial data might be expensive.

    But maybe it’s commercially viable. If you were to create models for all the traits typically used when breeding, then the genomic methods could replace traditional breeding. Sounds like there might be money in that. Whether it’s enough to cover the up-front cost, I don’t know. Agriculturalists must be doing this already, right?

    • Sandgroper says:

      Yes, more specifically animal herders have been for a very long time. Sheep dogs are notably smart, as are cattle dogs (but cattle dogs are mean, sneaky, nasty bastards, because they need to be – sheep scare easily, cattle are likely to turn around and fight back).

      Plus other kinds of working dogs, of which there are now a whole range. Sniffers of all different kinds (trackers, drugs, explosives, corpses, living trapped humans, etc.), guards, attack dogs, seeing eye dogs for the blind; all sorts. Those dogs are very expensive (I know this because I have a cop in my extended family who is in charge of a big police dog training facility – the dogs all come from Germany and they cost a bomb – like RMB1M each). So, you can bet that for that price, a hell of a lot has gone into the breeding of them. If the cops pay a million Yuan for a dog and get a dud, they’re not going to be happy.

      • j says:

        Yes, the individual dogs that breed the best descendants are very, very expensive. Like a prize winning horse. But there is no demand for intelligence but much for obedience. It would be embarrassing that the dog was smarter than its police officer…

        • Sandgroper says:

          I know some smart cops. You’d be surprised. Chinese, of course.

          The dogs need to be intelligent enough to train to do relatively complex things.

  21. I hope new and improved humans get here as soon as possible. Let’s just say for the sake of argument it is not that they don’t like us, they think there should be a hell of a lot less of us.

    So do they get into a war, with us? Probably not, after all they are improved. So they are smarter and they want less of us, a pretty reasonable conclusion to make. They don’t want us dropping dead from a pathogen they create and unleash upon us, too messy, too unethical, and too much risk for backlash. After careful consideration of the problem of too many dumb people they decide to let loose a pathogen that has no side effects except that it reduces fertility.

    If they are smart enough to do that much than they can be very sneaky and focused with who in the population has the greatest reduction in fertility and they could very gradually turn off the faucet of human fertility so that the impact is never blamed on them.

    Super intelligent humans could finally get just too disgusted with us and then breed super intelligent dogs with hands so that lawns get mowed, houses get fixed and meals get made. An underclass of angry dumb humans sounds real risky. If golden retrievers had human hands and IQ’s of 90 to 100 they would get rid of us altogether. Hired help jumping around all happy to see you every time you open the front door sounds far preferred to a grumbling dangerous underclass.

    • Tonight dark forces are gathered in Cleveland. They have flown a drone armed with a fishing net onto the top of the scoreboard at the Cleveland Indians stadium the night before the seventh game of the World Series. When the Cleveland Indians hit a routine fly ball to center field the drone will spring into action. Swooping down on the center fielder and hovering directly over his glove the ball drops directly into the fishing net and the drone then sweeps off into the night.

      The Umpires follow the rules and declare the routine fly out a ground rule double and the Cleveland Indians beat the Chicago Cubs because of the run that scores from it. Riots break out in Chicago. Upon closer inspection the drone was wearing a Donald Trump wig!

      Ohio a swing state in the upcoming presidential election votes for Trump! But the Cubbie blue states are mad as hell.

      “Rigged, everything is rigged!” they scream. History decides Hillary won the election because of the Grab them by the Pussy tape and a stolen World Series.

  22. syonredux says:

    RE: IQ,

    According to Chisala, Scrabble and Checkers prove that Blacks are really smart. Except for American Blacks, who are strangely dim…

  23. stephen hsu says:

    From the blog post linked below:

    The prediction I’ve made about the consequences of additive genetic variance in intelligence is not that we’ll be able to realize +30 SDs of cognitive ability. That would only be true if we could ignore pleiotropy, nonlinear corrections to the additive approximation, etc. What I claim is that because there are +30 SDs up for grabs in the first order approximation, it seems likely that at least a chunk of this will be realizable, leading to geniuses beyond those that have existed so far in human history (this is the actual claim). To doubt this conclusion one would have to argue that even, say, +8 or +10 SDs out of 30 are unrealizable, which is hard to believe since we have examples of healthy and robust individuals who are in the +6 or +7 range. (These numbers are poorly defined since the normal distribution fails to apply in the tails.)

    I could make further, more technical, arguments that originate from the fact that the genomic space is very high dimensional. These suggest that, given healthy/robust examples at +X, it is very unlikely that there is NO path in the high dimensional space to a phenotype value greater than X while holding “robustness” relatively fixed.

    Greg comments on whether super smart people can have “normal” personalities. This is obviously not necessary for them to be viable contributors to civilization (and even less of an issue in a future civilization where everyone is quite a bit smarter on average). He posits that von Neumann might have been radically strange, but able to emulate an ordinary person when necessary. (The joke is that he was actually a Martian pretending to be human.) My impression from reading Ulam’s autobiography, Adventures of a Mathematician (see also here), is that von Neumann was actually not that strange by the standards of mathematicians — he was sociable, had a good sense of humor, enjoyed interactions with others and with his family. He and Ulam were close and spent a lot of time together. I suspect Ulam’s portrait of vN is reasonably accurate.
    The University of Chicago conference on genetics and behavior Greg mentions, which was hosted in James Heckman’s institute, is described here, here, and here (videos).

    A masochist in the comments asked for the actual argument, so here it is:

    Here’s a simple example which I think conveys the basic idea.

    Suppose you have 10k variants and that individuals with 5.5k or more + variants are at the limit of cognitive ability yet seen in history (i.e., at the one in a million or billion or whatever level). Now suppose that each of the 10k + variants comes with some deleterious effect on some other trait(s) like general health, mental stability, etc. (This is actually too pessimistic — some will actually come with positive effects!) These deleterious effects are not uniform over the 10k variants — for some fixed number of + variants (i.e., 5.5k) there are many different individuals with different levels of overall health/robustness.

    Let the number of distinct genotypes that lead to (nearly) “maximal historical” cognitive ability be n = (number of ways to distribute 5.5k +’s over 10k variants); this is a huge number. Now, we know of many actual examples of historical geniuses who were relatively healthy and robust. The probability that these specific individuals achieved the minimum level of negative or deleterious effects over all n possibilities is vanishingly small. But that means that there are genotypes with more than 5.5k + variants at the same level of general robustness. These correspond to individuals who are healthy/robust but have greater cognitive ability than any historical genius.

    You can make this argument fully realistic by dropping the assumption that + effect sizes on cognitive ability are uniform, that effects on different traits are completely additive, etc. The point is that there are so many genotypes that realize [cognitive ability ~ historical max], that the ones produced so far are unlikely to maximize overall health/robustness given that constraint. But that means there are other genotypes (off the surface of constraint) with even higher cognitive ability, yet still healthy and robust.

    • whyteablog says:

      You’re saying that we identify a bunch of + alleles that don’t have negative effects? How long would this take?

      I wonder if it’d be easier (read:cheaper and with quicker payoffs) to just do it the old fashioned way, breed for traits you like as if we were cattle. Take Americans with 2100+ SAT scores, offer them to be part of a research experiment. Test them for neuroticism, conscientiousness, agreeableness, extraversion etc, weed out the ones with shitty personalities. Weed out the ones with genetic disease risk factors. I’d even weed out the ugly ones, because I’m shallow and arbitrary. Pay the remaining ones for their eggs and sperm.

      I’ll bet you could harvest eggs and sperm from 16 year olds and do several generations of this within a single lifetime.

      Instead of supergeniuses you get ordinary geniuses with winning personalities, killer smiles, healthy bodies, whatever political preferences you want and, I assume, great hair.

      Costs nothing to get SAT scores from the College Board if you can convince them to give them to you. Personality tests cost the time of a trained psychologist, genetic testing for disease associated alleles costs ~$100 per head. Pay random people to rate their attractiveness, buy their eggs and sperm. Should cost less than $5,000 per baby, not including the costs that having an ordinary child incurs, unless I have something wrong. And we could start this program today.

    • namae nanka says:

      ” we have examples of healthy and robust individuals who are in the +6 or +7 range. (These numbers are poorly defined since the normal distribution fails to apply in the tails.)”

      Besides that, SMPY’s multipotentiality study didn’t find it. So I doubt you could really call it IQ if it doesn’t look like a general factor at that point.

  24. Cpluskx says:

    ”They might be crazy”
    Bullied autistic nerd creates artificial super intelligence in his garage. It would be a bad day for the jocks.

  25. Anchises says:

    On the other hand, augmented intelligence in common laboratory mice has been observed to induce the megalomaniacal desire to subjugate the entire planet- not to mention a bizarre physical resemblance to Orson Welles.

    “The new & improved humans might not like us. They might be crazy: the risk for certain kinds of crazy increases with the number of plus alleles… We could… engineer super–dogs that were 30 std smarter than dogs today. This might be a better idea. Dogs actually like us, which is not at all guaranteed for supermen”.

    I recently read a sci-fi book, “The Judge of Ages”, which employed something similar. To outsmart his enemy (who has access to a hyper-intelligent self-aware supercomputer copied from his own brain), the hero builds his own hyper-intelligent supercomputer, but copies the brain of his faithful sorrel horse, to ensure its permanent loyalty and sanity (and confuse the heck out of the villain).

  26. melendwyr says:

    It’s worth noting that IQs are normalized. If the average actual score people get on a test is 100, and one in a million people gets 105, that 105 would be an IQ of about 180.

    So I suppose that, strictly speaking, you could plausibly find people who were actually thirty standard deviations above the mean in performance… but that doesn’t mean that their performance would necessarily be noticeably better than people only six to nine standard deviations above the mean.

  27. Sinij says:

    We have naturally occurring 200+ IQ individuals. While they are not known to relate well with the comparatively dumb ‘the rest’ of humanity, they have not attempted to conquer or domesticate it either.

  28. It is already done. Border Collie is the smartest dog and is bred for working/herding livestock, obedience etc. Apperantly those qualities correlate very high with intelligence.

    One Border Collie called Chaser has a vocabulary of more than 1000 words.

  29. L o p says:

    Dogs like us because they don’t understand us ;)..

  30. whyteablog says:

    Greg, someone gives you a $5 million budget to go make better humans, by any means you can imagine. What’s your program?

  31. James K says:

    30 standard deviations higher IQ? 100 + 30 * 18 = 640. The prospect is scary. Even if these “people” liked us and had no antisocial tendencies, to them we would be scarcely better than farm animals.

    The prospect of super-intelligent dogs is more interesting; but dogs that are more intelligent than humans will soon tire of their mundane existence, and covet the position of pack leader. As a sci-fi plot it would make an interesting twist on “Planet of the Apes”.

  32. We don’t even have enough data to validate IQ tests in above 3 SD range, let alone +30 SD.
    Blank slaters will make fun on us for writing this +30SD.

  33. Thiago Ribeiro says:

    “Dogs actually like us, which is not at all guaranteed for supermen.”
    Stupid dogs like us (or at least lack the means to successfully overthrow us or offer detailed criticism of our civilization). There is no good reason to believe that more intelligent dogs would like us.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s