I'm not going to reply to most of you post because I think it's getting unproductive and one part has more interesting to discuss than the rest combined.
However, I do wish you'd drop all the accusations of "moral crusader" and whatnot because you're basically inventing motivations for me and trying to divert the topic into mud-slinging.
However, it is important to remember that equality of opportunity is not the same thing as equality of outcome
I never claimed it was. However, I still dispute your point that the ONLY reason for the salary gap is because of years on the job. The reason I dispute it is there appear to be examples of unequal opportunity.
Some jobs require a given personality type or philosophy to be successful or even competent. And for various reasons the sexes have different distributions of these personality types which in itself is going to lead to statistical differences.
However, people are terrible at statistics: if there is a genuine statistical bias across genders (I'll accept that in this section for the sake of argument), then in the absence of any information, picking based on gender is the best you can do. Of course no one makes blind hires. If you have additional information on the quality of the person then the gender provides no further information.
I suspect you already know that.
The problem arises in that people are not good at statistics and make the assumption of independence and so after assessing quality using other measures then multiply in the effect of gender. Not everyone does this: I've had one conversation with someone who was making this mistake. Of course the person didn't know about statistical independence and marginals and conditionals.
You can see evidence of such errors in other things, such as the CIA's "Casio terrorist watch", which was frankly embarrassing. It's the bloody CIA: they ought to have one person on staff with a basic understanding of statistics!
Naturally people think they're being rational---actually they are when they make these decisions. It's rational to make decisions based on evidence. However their lack of understanding of statistics leads them down the wrong path.
Anyway, backing up from that, a genuine statistical difference also exacerbates other problems. Assuming men and women are equally sexist in a male dominated field, women will experience more sexism than men and vice-versa:
The interesting thing about that is that it requires only (a) a base level of sexism [it applies equally well to any -ism] and (b) an imbalance.
What things like that lead to is that for people in the minority, they will likely experience more bias against them than people in the majority. The interesting thing is I don't think that depends on any gender issues at all. It in one case depends on people misunderstanding proability and statistics (you'd have to provide a MOUND of evidence to convince me people do understand them) and in the other case, not even that.
So even without people being ethically or morally bad about something, you can still get biases cropping up pretty much as emergent behaviour.