In contrast, "See Color"'s attacks on old-school liberalism _land_. We open to a '90s-alike PSA invoking the "doesn't matter if you're black, white, or purple" trope (which has been cringe for as long as I (born 1987) can remember, but which I imagine sounded progressive the _first_ time someone said it), until Amethyst breaks character to object to the script—
-> AMETHYST: Woah, woah, woah. Hold up a minute here. Ugh, who wrote this? I think it kind of _does_ matter that I'm purple? I mean, I'm purple because I'm literally an alien.
+> AMETHYST: What the—woah, woah, woah. Hold up a minute here. Ugh, who wrote this? I think it kind of _does_ matter that I'm purple? I mean, I'm purple because I'm literally an alien.
>
> BLACK KID: Well I'm not an alien, but it definitely matters to me that I'm black.
>
How is the old-school liberal to reply to this? The function of saying "or purple" is to appeal to a _principle_ of equal treatment. Adding a fantasy race in there highlights the universality of our commitment to equality: purple magical alien gem superheroines might not exist, but if they _did_, they would be entitled to the same rights and dignity as everyone.
-But how should our principle of majestic equality be applied? [Categories summarize information](https://www.lesswrong.com/posts/esRZaPXSHgWzyB2NL/where-to-draw-the-boundaries)—[cluster-structure in the real world](https://www.lesswrong.com/posts/WBw8dDkAWohFjWQSk/the-cluster-structure-of-thingspace). As a matter of AI design, there would be no _functional_ reason to assign entities to different categories, if they didn't differ in _some_ decision-relevant ways. The reason it's pretty weird to reference Amethyst's skin color in a lesson about human racism, is because the challenges Amethyst might face as a gem in a world of humans—perhaps the perceptual skew of living thousands of years when most humans don't see a hundred—is going to depend on the ways in which gems and humans are actually different.
+But how should our principle of majestic equality be applied? [Categories summarize information](https://www.lesswrong.com/posts/esRZaPXSHgWzyB2NL/where-to-draw-the-boundaries)—[cluster-structure in the real world](https://www.lesswrong.com/posts/WBw8dDkAWohFjWQSk/the-cluster-structure-of-thingspace). As a matter of AI design, there would be no _functional_ reason to assign entities to different categories, if they didn't differ in _some_ decision-relevant ways. The reason it's pretty weird to reference Amethyst's skin color in a lesson about human racism, is because the challenges Amethyst might face as a gem in a world of humans—perhaps the perceptual skew of living thousands of years when most humans don't see a hundred—is going to depend on the ways in which gems and humans are actually different, which don't apply to humans of different races who are relevantly the same.
In this way, we see that old-school liberalism is effectively the position that race _shouldn't exist_ as a cognitively meaningful category. But is it that easy? If there's some sense in which race _does_ exist—even just as a social "type tag" based on superficial anatomy markers in humans who are [otherwise totally biologically the same](/2020/Apr/book-review-human-diversity/)—then verbally claiming to pretend that it doesn't, isn't a realistic or _honest_ strategy for remediating the harm done by [unfair conventions that culturally evolved around the presence of the category](/2020/Jan/book-review-the-origins-of-unfairness/).
projects / Ultimately_Untrue_Thought.git / blobdiff