Which isn't a problem if you have blind judging of submissions--just make sure that your submission pool is diverse, and pick the best papers without reference to race, orientation, or gender.
..and if the field itself has a diversity problem, then what do you do? This is not a conference-level problem; it is a problem that starts much earlier in life and which will only be fixed by solving it earlier in life. Go to middle schools and figure out why girls who were doing well in math and science in elementary school suddenly lost interest in those subjects, and once you have worked your way from there to having more diversity in technical professions, we can talk about whatever diversity problems are left at conferences.
It may not be a conference level problem, but diversity at the conference level can go a long way to showing girls in middle school that STEM careers are viable for them. We know that a lot of the reason girls drop out along the way just is the perception of STEM as a male dominated field (and paralleling that, in countries like China where it's not considered strictly male, we see high female participation).
"diversity at the conference level can go a long way to showing girls in middle school that STEM careers are viable for them"
When was the last time you saw middle school students wandering around at a conference?
"We know that a lot of the reason girls drop out along the way just is the perception of STEM as a male dominated field"
Yes, clearly that's part of the problem. So why didn't the organizer of this conference go out of her way to invite middle school girls to see all the women she managed to get into the conference?
When I was an undergrad, the EE department had a problem: the policy of doubling female enrollment each year had to be revised to having female enrollment at all. Part of the solution was to print new admissions pamphlets that showed equal numbers of men and women, and equal numbers of white, Asian, and black people (none of these proportions even remotely reflected the reality of the department) smiling while working on their breadboard projects (also somewhat disconnected from reality). This is forgivable, of course, for the following reasons:
1. It is an advertisement. Advertisements always paint a rosier picture.
2. Nobody received any sort of career boost from being featured in the pictures.
3. The pamphlets were sent to high schools, which is exactly who the department needed to target to meet the goal of increased female enrollment.
Compare that to the conference:
1. Conferences are not advertisements for a field or job (usually)
2. Being invited to speak at a conference is a career-booster
3. The demographics at a conference have no impact on middle or high school girls' attitudes about math and science.
You continue to try to locate the problem in middle school while removing any means, direct or indirect, of addressing it. As you observe, speaking at a conference is beneficial to one's career; fostering female participation in conferences advances the careers (and visibility) of females in the field generally. You don't need to bus in a bunch of twelve year old girls and point at the female speakers. You just need the field itself not to look so totally male. In medicine, rising female participation outside of nursing has demonstrably accelerated the rate of female participation since the 1970s. The same can happen in the rest of the STEM field. It's obviously not a turnkey solution, but is valuable as part of a general attempt to diversify the field.
