Diversity at conferences

Continuing my slight diversion from the core topics of this blog – I’ll return soon, I promise. Today, I want to share some thoughts on conferences.

In particular I want to suggest two things so lets cut to the chase, state them and then I’ll explain myself.

Firstly, we need to recognise a conflict between inclusion & diversity and community. Having any community requires a differential between those inside, and those outside the community. You can welcome, even seek out, diverse people to join the community but if you prioritise diverse outsiders above insiders then the community is weakened.

Second, I think its time – for the sake of inclusion and diversity – to use random selection as part of conference selection.

What has changed

I think I know a thing or two about conferences. In the last decade I’ve spoken at more conferences than is good for me – some years more than a dozen, and thats not counting local group meeting. I’ve been intensively involved with three conferences, years on the programme committee of each, and I’ve chaired a few.

The pandemic obviously halted everything, but conferences have bounced back. There are now online conferences and physical ones. Online I find sterile, few have any feeling of community, they are transactional. But they exist and succeed in knowledge transfer. That makes physical conferences even more valuable for their community side.

But the second change pushes against community: the rise of online mass submission systems.

Many conferences use these now. As a speaker I can register, select the conference I want to submit to and do it in a few clicks. The system will then say something like “People who submitted to this conference also submitted to …” and they will encourage me to submit to more.

Except, I don’t know anything about those other conferences. I’m not a member of that community. Sure I can submit but am I really interested in going? It is a transaction. But its cheap to submit so …

Now the conference committee have to review my proposal together with all the other people who submitted on a whim. So the first problem: these systems make more work for conference organisers while weakening the community.

So the conference committee now have lots more submissions. Which is good because most conference want to broader diversity and become more inclusive: equal representation of the sexes, more under represented minorities and so on. I can’t really object to that – especially ‘cos, while I may present as a white-middle-aged-man I am also in one of those minority groups (neurodiverse).

The real solution to diversity at conferences is to tackle the source: we need more people in under represented groups submitting. This also means that conference organisers need to be actively monitoring submissions and working to increase minority submissions.

So two more problems: organisers need to monitor submissions for diversity, most don’t. Then they need to work – more work again – to increase representation. Few have the resources to do either properly.

When it comes to selecting submissions many conferences will tell you they select the best. So when comes to selecting organisers are looking for quality and diversity. Which is itself a change, conferences have always looked for quality but many traditionally privilege their own community: those who had attended before, perhaps regularly, and those who had spoken before.

The problem is: community and diversity are opposites. They only align for a very few people. Staying with your community reduces diversity but increasing diversity reduces community.

Imagine you run a hard core programming conference. Traditionally you have white-male programmers turning up. They are your community and the speakers are drawn from that community or people on the edges, who are probably also white-male programmers. Sure there are a few non-white, non-male, non-programmers but are there enough for a whole conference?

Now pause for a moment. It might look like your programme and audience lacks diversity what about the diversity you aren’t seeing? What about neurodiversity? Sexual orientation and gender that are invisible? Or maybe age profiles, skewed older, or younger? Social profile, income level, education level, religion?

Because your population is biased towards one way you have to go outside the community for diversity but by definition that reduces the community aspect. What is the balance between continuity and disruption?

Then there is a side problem of: what do you want your programme to look like? Male-female balance is easy, but what about ethnic group? neurodiversity? Religion? And remember, if we are talking about IT conferences we are skewed towards the upper income levels already. And talking about IT … the IT industry has an age bias. While I see conferences looking for “young upcoming talent” there is a blind spot on older workers.

Do you want your conference to reflect the industry (or society) as it is?

Or as you want it to be?

How many vegetarians should be on the programme? And should you correct for historical in balances?

And remember, the more diverse you make the programme the weaker you make the community. What happens next year? Who comes back?

And remember, because you get more “random” submissions from people on mass submission systems it is harder to form a community of people who come back year-after-year.

O, Quality

Of course you want high quality speakers no matter what their background. But, if your conference is any good there will be more good speakers who want to speak than you have space for. (You could make space but that would damage the economics of the conference.)

If you are selecting on merit you might find you have 30 submissions for 4 speaking slots – the kind of ratio I have seen.

Some conferences will ask for more an more details on the submission page. However, this runs against diversity and inclusion. You are increasing the barrier to submission, you are also filtering for those who are good at writing submissions and have the time required. Dyslexics (more words) and some parents (more time) will be put off.

Conferences which anonymise submissions are fair except you can no longer use past experience or watch old recordings. So you are filtering for people who write good, mainstream, submissions. (Really alternative submissions find it hard to meet criteria.)

Take away the rubbish and good but not very good, you might be down to 10 to 3. If you have multiple reviewers grading them 1 (worst) to 5 (best), you might find yourself choosing between one scoring 4.6 and one scoring 4.65. Is that fair?

What if you then look at their name to see if they are male or female? Or their profile picture to check skin colour? None of this is reliable let alone fare.

What is a fair submission systems?

For years I equated fair with merit. I wouldn’t do it now. I went along with blind reviews for a while but I think they are flawed. Even if you don’t sneak a peak do they give you a diverse set? And what about community?

I’ve given this a lot of thought in the last few years and while I don’t have all the answers I do have some ideas I think are worth trying.

My suggestions

Organisers must monitor submissions for minority criteria they consider important. So sex, ethnicity and age are monitored, but perhaps not religion, handedness and neurodiversity. If nothing else this will move the arguments over criteria to the front of the process rather than arguing about them during selection.

Perhaps controversially, while I don’t want to create entirely curated speaker programmes I think it has a role. I’d say about 40% of speakers should be invited outside of a call-for-papers.

Some of these would be for community continuity. Some would be specially invited because the conferences wants them or their topic (a little of this always happens behind the scenes, if only for the keynotes who are usually invited.) The rest would be from minority groups the organizers wanted represented at the conference.

There would be open submission, probably blind, for the other 60% and they would be reviewed by a panel. But they wouldn’t be selected on merit. (I still think I got it right when I created a 2 round submission process, first round blind, second round with speaker profiles.)

The conference would set a quality bar. Anyone below the bar would be rejected. So if you have 20 submissions for 3 slots on a 1 to 5 scale, the quality bar might be 4.1 so the 10 scoring less would be rejected. From the 10 scoring 4.1 or above 3 speakers would be selected randomly. So another 7 would be randomly rejected.

This seems fairer to me than a process that ignores community during selection and is biased towards those with good grammatical form-filling skills (which might be a privately educated bias too.)

At the end of the day there are just too many competing factors to balance. A random element might be the only way to remove hidden or overlooked biases. That said, in order to keep community and allow the organisers to create the conference they want there needs to be some pro-active selection.

I don’t claim to have all the answers. And maybe I’ve missed some vital point, but this is what I’m seeing right now. I’m interested to know what readers this, so at the risk of spam I’ve enabled comments for this post. Please let me know.