5 ms·

This is why I find movements like HBD so disgusting, despite their "scientific" origins. They start with a conclusion, the justification of race, and seek evide

This is why I find movements like HBD so disgusting, despite their "scientific" origins. They start with a conclusion, the justification of race, and seek evidence to prove it. Of course, they find such evidence. Anyone looking to prove any conclusion can find evidence - even scientifically strong evidence. Heck, even psychics and supernaturalists have scientific evidence on their side! [1] No matter the evidence though -- Categories aren't facts. > “If you call a tail a leg, how many legs does a dog have?” And the answer: “Four – because a tail isn’t a leg regardless of what you call it.” Categories are useful _caches_ of information. We define "limbs involved in the movement of the center of mass of a body" as _legs_ because the concept is useful as a cached "leg". If the concept stops being valuable, we will eventually purge it from the cache. The danger lies in mistaking the cached concept for the original value. [2] Race doesn't predict anything; nowadays, the cached value, crime/success/happiness, is substituted for race, and the results are a reinforcing cycle. [3] More relevant to HN, categories are why "adversarial" effects on many neural networks are so strong. NNs are training to categorize to exacting strategies - strategies not at all in mind by their creators. The apocryphal tank NN meme - that dataset bias led algorithms to detecting sunny skies instead of tanks -- comes to mind. [4] The story is made up, the moral somewhat muddled in the telling, but the surprise survives. What we reward and what we desire are separate things, and in the difference, the whole world lies. [1]: http://slatestarcodex.com/2014/04/28/the-control-group-is-out-of-control/ http://slatestarcodex.com/2014/04/28/the-control-group-is-ou... [2]: http://slatestarcodex.com/2014/05/02/causal-models-at-work/ http://slatestarcodex.com/2014/05/02/causal-models-at-work/ [3]: https://www.vox.com/policy-and-politics/2018/3/27/15695060/sam-harris-charles-murray-race-iq-forbidden-knowledge-podcast-bell-curve https://www.vox.com/policy-and-politics/2018/3/27/15695060/s... [4]: https://www.gwern.net/Tanks https://www.gwern.net/Tanks

lopmotr 8y ago
Race absolutely does predict things but it predicts group properties accurately and individual's only probabilistically. That's bound to happen no matter how you categorize people. Every category has to have a different mean height, weight, etc. from the others because those are continuous variables. It turns out that for some useful properties, race is a better classification than random groups. Of course, it is a human invention, just like planets and fish, but it's one with predictive power, just like those other examples.
- aetherson 8y ago
  "Better classification than a random group" is a fairly low bar.
  - lopmotr 8y ago
    I'm trying to counter the claim that race doesn't predict anything. More than nothing is something. How much? I haven't quantified it but neither has the OP.
    - andrewprock 8y ago
      It's not bad at predicting what race people will think someone is. Outside of that, it's of marginal utility at best. At worst, it's used for systematic oppression on a global scale.
      - alanbernstein 8y ago
        If the goal is to counter that oppression, how can it help to dispute or ignore the categories?
bordercases 8y ago
You're saying a lot of things which are true without mapping them to explicit implications; this is creating a smokescreen by assuming that your 'technical nuance' makes your conclusion self-evident. As far as I can tell, what you're trying to imply is that the definition of terms are driven by their use, and the use of the term race is disgusting, because it makes people feel like they can't change their situation or inflate the ego of other people with all the trappings that go along with it, which is bad, so we shouldn't use it and race doesn't exist. Yet so-called "human biodiversity" doesn't go away just because the way some decide to use the information is bad. Genetic variation exists, we can count it, and we know it's unevenly distributed. But even through this uneven distribution we can assemble models that track the migration of genes through human beings and notice pools of genes in certain populations. Consider k-clustering: it's a way of bootstrapping features, approximate categories. The process is parameterized so that you can create distinctions as great or as fine as you want. Bioinformaticians do this kind of analysis all the time and can assign each person their own label if you like. But it isn't as useful and so we get haplogroups which track variation of genotype much more robustly and compactly. Geneticists and evolutionary anthropologists didn't invent their fields to oppress people. A lot of them were just trying to do science to see where the genes are. You're shooting the science when it's policy makers who are responsible for the "disgusting" use of such information. And if someone has maladaptations, that's no one's fault in particular except the random accidents of Nature; the moral impetus is made more urgent by the acknowledgement of our genes rather than their denial.
spacehome 8y ago
That's a strange argument to make on this article. Sure, races are an attempt at categorization of something much more complicated than the categories can encode. As such, there's differing perspectives about how many racial categories there are, how to draw the boundaries, and no matter what, there will be people who don't fit into any of your categories. None of this detracts from the fact that racial categorizations are useful and predictive for the overwhelming majority of the Earth's population, just as the questionable status of Pluto doesn't interfere with research on extra-solar planets. By HBD, I assume you're referring to differences in intelligence between races, but please correct me if you're referring to something else. It's always a bad idea to start with the conclusion you want and work backwards. However, in the US, amongst the present-day population, there is OVERWHELMING evidence that there are average IQ differences between races. The open question is what causes these differences in the averages. We know IQ is partly genetic, and we know races differ in allele frequency of many genes. It's possible, though exceedingly unlikely, that genetic differences between races contribute precisely zero to average IQ differences. And there seems to be some circumstantial evidence that genetic differences between races do account for some of the average differences. This is a question that is at least amenable to study, by correlating genes with IQ, and matching this against allele frequency in groups. HBD is not unscientific, and the comparison to psychics is unwarranted. HBD is an attempt to explain an observation (IQ deltas) with a hypothesis.
- jonnybgood 8y ago
  What races are you referring to? Irish, German, English, Spanish? What’s their average IQ? And what genes play a factor in IQ?
  - bordercases 8y ago
    > what genes play a factor in IQ? All of them play a little bit (majority of quantitative differences) and some of them play a lot (when there are significant qualitative differences). https://www.youtube.com/watch?v=eVSbFAcBAyw https://www.youtube.com/watch?v=eVSbFAcBAyw
  - vixen99 8y ago
lopmotr 8y ago
> They start with a conclusion, the justification of race, and seek evidence to prove it. Of course, they find such evidence. Anyone looking to prove any conclusion can find evidence - even scientifically strong evidence. Can you point to scientifically strong evidence of anything which is also known to be false? I mean evidence which is still accepted as scientifically strong, not something that's been discredited. If not, then I think you've just argued that we can't use science to explain anything because it lets us choose any conclusion we want. I don't think you're trying to promote an anti-science ideology, so can you clarify that?
nailer 8y ago
Do you find movements that discuss commonly accepted neck-down differences between races disgusting too?

lopmotr 8y ago

Race absolutely does predict things but it predicts group properties accurately and individual's only probabilistically. That's bound to happen no matter how you categorize people. Every category has to have a different mean height, weight, etc. from the others because those are continuous variables. It turns out that for some useful properties, race is a better classification than random groups. Of course, it is a human invention, just like planets and fish, but it's one with predictive power, just like those other examples.

aetherson 8y ago
"Better classification than a random group" is a fairly low bar.
- lopmotr 8y ago
  I'm trying to counter the claim that race doesn't predict anything. More than nothing is something. How much? I haven't quantified it but neither has the OP.
  - andrewprock 8y ago
    It's not bad at predicting what race people will think someone is. Outside of that, it's of marginal utility at best. At worst, it's used for systematic oppression on a global scale.
    - alanbernstein 8y ago
      If the goal is to counter that oppression, how can it help to dispute or ignore the categories?

bordercases 8y ago

You're saying a lot of things which are true without mapping them to explicit implications; this is creating a smokescreen by assuming that your 'technical nuance' makes your conclusion self-evident. As far as I can tell, what you're trying to imply is that the definition of terms are driven by their use, and the use of the term race is disgusting, because it makes people feel like they can't change their situation or inflate the ego of other people with all the trappings that go along with it, which is bad, so we shouldn't use it and race doesn't exist. Yet so-called "human biodiversity" doesn't go away just because the way some decide to use the information is bad. Genetic variation exists, we can count it, and we know it's unevenly distributed. But even through this uneven distribution we can assemble models that track the migration of genes through human beings and notice pools of genes in certain populations. Consider k-clustering: it's a way of bootstrapping features, approximate categories. The process is parameterized so that you can create distinctions as great or as fine as you want. Bioinformaticians do this kind of analysis all the time and can assign each person their own label if you like. But it isn't as useful and so we get haplogroups which track variation of genotype much more robustly and compactly. Geneticists and evolutionary anthropologists didn't invent their fields to oppress people. A lot of them were just trying to do science to see where the genes are. You're shooting the science when it's policy makers who are responsible for the "disgusting" use of such information. And if someone has maladaptations, that's no one's fault in particular except the random accidents of Nature; the moral impetus is made more urgent by the acknowledgement of our genes rather than their denial.

spacehome 8y ago

That's a strange argument to make on this article. Sure, races are an attempt at categorization of something much more complicated than the categories can encode. As such, there's differing perspectives about how many racial categories there are, how to draw the boundaries, and no matter what, there will be people who don't fit into any of your categories. None of this detracts from the fact that racial categorizations are useful and predictive for the overwhelming majority of the Earth's population, just as the questionable status of Pluto doesn't interfere with research on extra-solar planets. By HBD, I assume you're referring to differences in intelligence between races, but please correct me if you're referring to something else. It's always a bad idea to start with the conclusion you want and work backwards. However, in the US, amongst the present-day population, there is OVERWHELMING evidence that there are average IQ differences between races. The open question is what causes these differences in the averages. We know IQ is partly genetic, and we know races differ in allele frequency of many genes. It's possible, though exceedingly unlikely, that genetic differences between races contribute precisely zero to average IQ differences. And there seems to be some circumstantial evidence that genetic differences between races do account for some of the average differences. This is a question that is at least amenable to study, by correlating genes with IQ, and matching this against allele frequency in groups. HBD is not unscientific, and the comparison to psychics is unwarranted. HBD is an attempt to explain an observation (IQ deltas) with a hypothesis.

jonnybgood 8y ago
What races are you referring to? Irish, German, English, Spanish? What’s their average IQ? And what genes play a factor in IQ?
- bordercases 8y ago
  > what genes play a factor in IQ? All of them play a little bit (majority of quantitative differences) and some of them play a lot (when there are significant qualitative differences). https://www.youtube.com/watch?v=eVSbFAcBAyw https://www.youtube.com/watch?v=eVSbFAcBAyw
- vixen99 8y ago

lopmotr 8y ago

> They start with a conclusion, the justification of race, and seek evidence to prove it. Of course, they find such evidence. Anyone looking to prove any conclusion can find evidence - even scientifically strong evidence. Can you point to scientifically strong evidence of anything which is also known to be false? I mean evidence which is still accepted as scientifically strong, not something that's been discredited. If not, then I think you've just argued that we can't use science to explain anything because it lets us choose any conclusion we want. I don't think you're trying to promote an anti-science ideology, so can you clarify that?

nailer 8y ago

Do you find movements that discuss commonly accepted neck-down differences between races disgusting too?