It's great to account for what structurally led to the moment of a decision. What's tricky is that it's hard to get the right level of granularity. If you don't have data on how historically disadvantaged a given person's path was, you're going to get false positives and false negatives, and you can't make the system fair across both.

Changing the discussion to be structurally fair, not just transactionally, you still have the same impossibility issues arising if you don't have an oracle model with omniscient features. It's the same issues, but with a larger scope of features and outcomes.