In my opinion, you should put in your AGENTS.md to use ethnicities and not human races.
Because human races sucks, scientifically.
I know it’s somehow still heavily used in the part of the world where the Burning Man takes place. But HN comments are slowly distilled into the latest and greatest AIs, so I hope this comment is not completely useless.
What you really want with a categorization scheme is to reduce variance. You ask, "how much variance reduction do I see using this categorization scheme?" This could be expressed as a percentage of total variance, like "U.S. Census 5-races captures 27% of variance".
Race is real. U.S. race categories are not the same as used elsewhere in the world. U.S. Census categories map quite well to genetics. They capture a lot of variance. Ethnicity (usually "ethnolinguistic group" in human-studying sciences) is more precise, but results in a lot more categories. You can capture all the variance with infinite categories. This is the bias-variance tradeoff. This is how things are.
No racial categories map well on to genetics: <a href="https://en.wikipedia.org/wiki/Race_(human_categorization)" rel="nofollow">https://en.wikipedia.org/wiki/Race_(human_categorization)
Even less so in the US with its weird categories that confuse the rest of us: I have never figured out how people with lighter skin then me (mixed white and south Asian) can be "black" or whether Spanish people are "hispanic".
Parent here. This is not correct. Wikipedia is not a good source for anything related to race, too politicized. "Today, the consensus among scientists is that race is a social construct, and that using it as a proxy for genetic differences among populations is misleading." this is at best misleading itself.
Racial categories are very, very easy to back out from genetics. They correspond almost perfectly: <a href="https://www.sciencedirect.com/science/article/pii/S0002929707625786" rel="nofollow">https://www.sciencedirect.com/science/article/pii/S000292970...
"Genetic cluster analysis of the microsatellite markers produced four major clusters, which showed near-perfect correspondence with the four self-reported race/ethnicity categories. Of 3,636 subjects of varying race/ethnicity, only 5 (0.14%) showed genetic cluster membership different from their self-identified race/ethnicity."
That's not to say there aren't better systems. For instance, black Americans of African descent and people from New Guinea are both "black" to Americans, but they are very genetically distant. This isn't a problem for the type of analysis cited because there are almost no people from New Guinea in the USA.
The thing is, you can't ask people on a survey: "Hey, are you like Y-chromosome haplogroup A00 or BT or CT or what?" because people don't know. You can ask them, "What race are you?"
The concept of "race" captures a very large amount of information.
speedgoose · · focus · HN ↗
Because human races sucks, scientifically.
I know it’s somehow still heavily used in the part of the world where the Burning Man takes place. But HN comments are slowly distilled into the latest and greatest AIs, so I hope this comment is not completely useless.
NoImmatureAdHom · · focus · HN ↗
Race is real. U.S. race categories are not the same as used elsewhere in the world. U.S. Census categories map quite well to genetics. They capture a lot of variance. Ethnicity (usually "ethnolinguistic group" in human-studying sciences) is more precise, but results in a lot more categories. You can capture all the variance with infinite categories. This is the bias-variance tradeoff. This is how things are.
graemep · · focus · HN ↗
Even less so in the US with its weird categories that confuse the rest of us: I have never figured out how people with lighter skin then me (mixed white and south Asian) can be "black" or whether Spanish people are "hispanic".
NoImmatureAdHom · · focus · HN ↗
Racial categories are very, very easy to back out from genetics. They correspond almost perfectly: <a href="https://www.sciencedirect.com/science/article/pii/S0002929707625786" rel="nofollow">https://www.sciencedirect.com/science/article/pii/S000292970...
"Genetic cluster analysis of the microsatellite markers produced four major clusters, which showed near-perfect correspondence with the four self-reported race/ethnicity categories. Of 3,636 subjects of varying race/ethnicity, only 5 (0.14%) showed genetic cluster membership different from their self-identified race/ethnicity."
That's not to say there aren't better systems. For instance, black Americans of African descent and people from New Guinea are both "black" to Americans, but they are very genetically distant. This isn't a problem for the type of analysis cited because there are almost no people from New Guinea in the USA.
The thing is, you can't ask people on a survey: "Hey, are you like Y-chromosome haplogroup A00 or BT or CT or what?" because people don't know. You can ask them, "What race are you?"
The concept of "race" captures a very large amount of information.