You can quantify a lot of these things, and those things that cannot be quantified can be hashed with a pre-existing mapping that captures the relevant information in a way that doesn't reveal irrelevant genetic traits.
For example, high schools nationwide are constantly ranked and compared. If you have an identifier that simply says "rural tier 3 high school," it's unclear what the race of the applicant is.
A parent's income can be shown based on decile of Americans, without revealing the race of the applicant.
Geography can be given in a way that is nonspecific enough to prevent educated guesses about the applicant's heritage.
I contend that race doesn't tell you much more about how hard a person had to work, if you already have all of the preceding factors evaluated.
For example, high schools nationwide are constantly ranked and compared. If you have an identifier that simply says "rural tier 3 high school," it's unclear what the race of the applicant is.
A parent's income can be shown based on decile of Americans, without revealing the race of the applicant.
Geography can be given in a way that is nonspecific enough to prevent educated guesses about the applicant's heritage.
I contend that race doesn't tell you much more about how hard a person had to work, if you already have all of the preceding factors evaluated.