Demystifier said:
Why would you add more men there? Why do you think that an optimal distribution of men and women is a uniform one?
Furthermore, is sex the only characteristic that should be uniform? For instance, suppose that someone told you that people with blood group A more often choose physics than people with blood group B. Would you then argue that there should be more blood group B people in physics?
Note that this thread is OT, and general politics is definitely OT, so while I am interested on the topic of equality vs efficiency, I will not address the first question for general jobs or tasks. For physics particularly, it is not about equality but about rightly pooling the intelligence reservoirs. So in principle yes, if we have a variable, as blood group, that happens to be statistically underrepresented, say five sigmas away, then an issue is happening.
Of course it could be that the issue is happening out of the reach of physicists. Say that some blood group is not being taught enough math due to whatever world geopolitics. But we can establish a baseline about having enough math, say the first year in university, or the number of graduates. If distributions change respect to this baseline, something is going wrong and we are leaking brain power. Note that the talk rightly concentrated in this "pipeline" problem. You are right that not only the male/female distribution, but also other ones (nationality, family income, etc) should be checked. Do we miss people because their fathers have a company and claim them to executive posts? Do we miss people because their families are poor and they need to move to stable work? Do we miss people because of their origin nationality?
Unfortunately, the finer you make the division, the most difficult to get enough statistics. We could check if the distribution of first letter is preserved along the pipeline, and in fact we could suspect that names with early position in the alphabet are favored. But to detect this bias we would surely need a lot more of sample size that the current number of graduating physicists (well, if the hypothesis is about first letter, and not about some letter, we can use distance to AAA instead of partitioning in 28 cases... hmm perhaps it was not a so good example of impossibility).
