>Oh for goodness sake. The point was, as an employer or potential partner, you want someone who will be effective in that role.

That's how I understood it too.

I was pointing to the inhumanity of treating people as mere means to an end ("not effective") especially in the case of pregnant women (which translates to women should either not get pregnant because its bad for business or be unemployable when they consider to do so).

You say it's just a "statistical argument". That's what I responded to, too: that's it's not OK - at least to me - to do mere statistical arguments when moral judgements are involved. Or rather, it's ok to do them as science, but this is not that, this is how actual managers and employers think and treat employees.

An overly narrow view. A predictably potential rapist who was subsequently convicted would become permanently unavailable and and could bring down a huge liability on an employer. An employee who gives birth (or who becomes a father) is engaging in normal human activity; you treat that the same way you treat any other indisposition like an unexpected health condition, and hedge against it.

Really, ask yourself how you'd feel if a prospective employer or business partner said 'Joe, what assurance can you give me that you're not going to knock your wife up and go all googly-eyed over some squalling infant a year from now? Can you commit to tying a knot in it?'

Statistically, there are equally-qualified candidates with none of these problems. A moral-free algorithm would surely choose young unmarried technologists over any other kind. I'll repeat: the trouble with discrimination is, it works so well.