They’re a good two cents. 🙂

Thanks Hugo

I hope the pool of potential speakers is not as skewed as you suggest. I’m aware of at least one strong female speaker who was rejected this year, so I’m confident they could’ve done better. And the same occurs for other conferences. I think data has strong female representation (compared to other IT disciplines), and a lot higher than 22% could be achievable with a change in method. I know there are a number of options being considered, and I will point some of those people at your suggestions.

I appreciate that if the skew in the potential speakers is higher than the skew in the chosen speakers, the range in quality between the demographics could seem to be a factor. But I think the quality in diverse communities is generally higher than in less diverse ones, because they have to fight harder than us privileged. So I don’t think it really is a concern.

Rob

A submission pool with low diversity means that, indeed, doing blind selections will result in a similar low diversity in the accepted sessions. But I am not sure if your suggestion to do non-blind picks would be better, There is a list of names in our community that are considered to be “good”. Many on the selection committee will pick any session that has one of those names as the speaker. And when I think of who are on that list, I see lots of white men, and very few who are not.
How about we strip the name but leave the diversity information? Would that help? Well, I am sure that *some* on the selection committee would try to include more people with a diverse background in their process. But, call me cynical if you wish, I am not at all convinced that there would also be people on that same committee that would actually favour non-diverse submissions. Both conscious and subconscious bias are a realty that we have to admit to when we really want to solve this issue.

Now I do not want to nag only. As I was reading your post I had an idea for a new selection process that might perhaps yield better results. It is a lot more time consuming, and probably only usable for large conferences. And I don’t *KNOW* that it will be better. It’s just a thought. (Also note that the numbers below are just an example and should probably be tweaked).

1. Give the people on the program committee only the abstracts of one of the minority groups (e.g. non-whites). Go through a full selection process. Use this to fill the first 25 slots.
2. Next, give the PC only the abstracts of the next minority group (e.g. women, of course excluding those that were already chosen, but still including non-white women that were not selected in the first round), and repeat the process to fill another 25 slots.
3. Repeat this once more for e.g. speakers not originating from USA or Europe, though here (given the reality that travel restrictions and cost will always impact this group) perhaps a lower number of sessions is warranted, let’s say 20.
4. Finally, a last iteration of the process to give 10 speaking slots to groups that actually are a small minority in the real world: transgenders, people with disablities, perhaps more groups I am overlooking now.
5. At this point, there are 80 slots assigned to diverse speakers, still based on a solid process that tries to filter for quality and good content. The remaining 50 slots (stealing the 130 assumption from your post) can now be selected from one last iteration of the standard selection process that includes ALL abstracts that have not been selected yet: all those from a diverse speaker that were skipped before, and all those from a non-diverse speaker.
6. Session selection is not made public until the full process has been completed, and nobody is told during which iteration they were selected. A woman of colour could have gotten her slot in one of the first two priority rounds, or in the final round, and because nobody knows there’s low chance of her thinking (or being accused) that she was ‘only selected to meet a target’.
7. During each of the selection phases, the people selecting do not know which phase they are in and the abstracts they get are anonymized. They just get a bunch of abstracts and the request to pick the best ones.
8. The process is documented so that speakers know, when submitting already, that answering the optional diversity questions will increase their chances. I hope this will increase the percentage of speakers who provide this information.
9. To prevent abuse (e.g. a white man indicating he is a non-white woman on his submissions), anyone found to be dishonest in their diversity questions will be banned for life from that conference.

I’m sure there are lots of issues and problems with this idea too. But even so, I still want to put this suggestion in the pool of ideas to consider when thinking about how conferences can work towards higher diversity in their speaker pool