But that's a losing strategy, in a game-theoretic sense, if your goal is to end sexism/discrimination/bias.
It is fine if your goal is upholding a moral code where you personally actively committing an act of discrimination is unacceptable, but even unintentionally; you personally passively upholding discrimination is fine; and other people actively committing discrimination and you failing to stop them is not a thing you're super morally culpable for. That's certainly not my moral code, and I think even among the crowds that believe in intrinsically evil actions regardless of context (e.g., Catholic moral theology), they wouldn't agree that passively upholding evil or failing to stop others from doing active evil is fine (e.g., "I have greatly sinned ... in what I have done, and in what I have failed to do"). If we're making a deontological argument, we should nail down what we think about passive wrong or allowing a wrong to continue, and if we're making a consequentialist argument, we're not worried about intrinsic wrongs along the way to a right.
It's a losing strategy because everyone who actively supports sexism, racism, and other forms of discrimination have plausible reasons why their discrimination is justifiable. Even the white-nationalist types these days hesitate to say that the white race is superior; they just say they want protections for the white race in white countries (whatever those are). And most of the discrimination in today's society doesn't come from people who are nowhere near as overt as white nationalists. The colleges say, "Oh, we're just trusting these test scores." The standardized test companies say, "Oh, we're just trusting past performance at college." And if anyone had previously introduced bias into the system, they've now successfully laundered the bias; there's a feedback system that keeps whatever biases were present when it was created, and you can quite genuinely say, "Oh, I'm just following this system, which on paper should be a perfectly objective system" and there's no proof that you're actively and intentionally discriminating. But you're upholding discrimination, all the same.
If you want to see this sort of bias-laundering in practice, my favorite recent example is the voting laws in North Carolina that were recently struck down by their Supreme Court:
http://www.chicagotribune.com/news/nationworld/politics/ct-n...
Every restriction, on the face of it, was defensible. Voter ID in the abstract is a good idea. Eliminating certain parts of early voting seems fine. But the courts looked at the emails behind this law, where legislators asked which particular voting mechanisms were used by specific demographic groups, and eliminated those mechanisms "with almost surgical precision". You couldn't prove from the text of the law that there was any intention at bias, which was the entire point; it wasn't supposed to look like a discriminatory law.
We don't have the benefit of seeing those discussions most of the time. So waiting until we have a proof of a wrong to fix that wrong is a losing strategy, one that is easily exploited by people who want to discriminate, and one that people who want to discriminate have demonstrated their willingness to exploit.
