Towards improving community diversity

I think it would be good to hear from women here: What is it that you think would help women programmers most? What is it that seems to hold women back? Are there some positive changes that seemed to have made a difference? (If so, we should know. If it works, that is great.) Are there some steps the community took but that did not seem to be effective? (If so, we should know. Less of that then…)

7 Likes

I know you’re coming from a good place, but I’m not terribly comfortable explicitly asking underrepresented groups to do additional work to improve their own representation. I’d rather just create an open space that is welcoming and affirming and allows for participation in a manner that’s (hopefully) comfortable to all.

We had a very productive BOF session in the 2020 JuliaCon that highlighted some of the things that have worked and some places where we can do better, and there’s ongoing dialog as a result of that.

17 Likes

Right you are. On the other hand, wouldn’t it be nice to get some feedback?

I can’t speak for women but I find diversity talk interesting. I don’t know how many middle eastern people are here but probably not many as a percentage. So I guess I “qualify” to talk about diversity. I think to try to “understand” the problem, it helps to understand how we got here first. Julia was started by a few people at MIT so its initial audience was the group of friends and colleagues of the Julia founders at MIT and the collaborating institutes. I presume those were predominantly male. The Julia core team is predominantly male. This is one source of bias. And I don’t mean bias here in a bad way. I just mean it in a self-reflective way. I think bias happens and it’s natural.

The next group of people who probably joined the Julia community back then (according to my pure speculation, correct me if I am wrong) were people in tech and academia who were frustrated enough with existing technology that they decided to spend the extra time playing around with this new language. This probably included the JuMP folks, Chris of SciML, programming language and compiler enthusiasts and other Julia pioneers who were early adopters of the language. Those again seem to be predominantly male. Imo this is hugely influenced by the male to female ratio in the applied mathematics and computing schools and departments in those universities that had early exposure to Julia. This is another source of bias.

So when evaluating our diversity stats, to be fair we have to “correct” for the second point above by seeing how far off we are from the same diversity metrics in the fields predominant in the Julia community in the schools where most people in the Julia community come from. I am sure there is a statistically “correct” way to quantify this that isn’t too self-beating but also doesn’t deceivingly pat us on the back. This would be an interesting statistics project. Correcting for the first point is more difficult but maybe possible. Assuming that males tend to have more male than female friends, and people of a certain ethnicity tend to have more friends of the same ethnicity than any other, the “exposure” of a certain sub-group to Julia is probably going to follow some noisy S-shaped, pandemic-style curve in every sub-group. This is the “natural” flow of things. Understanding and quantifying how much “natural” growth is going to happen without much interference lets us predict how much worse the diversity numbers will get in the next years if nothing is done. But more importantly, it points at some good actions that can be taken to at least stop the numbers from getting worse or maybe even improve them.

Given the above, I think it’s somewhat obvious what needs to be done:

  1. Find fields, schools, universities and countries with more minority students and academics than average and promote Julia there. When talking about different countries, the language and culture barriers become an issue here as well.
  2. Give special opportunities to minorities in the community or those who want to join the community. Although that comes at a cost for the majority. Since opportunties are generally scarce, giving advantage to one group can disadvantage other groups. Finding the balance here is “not easy”, to say the least.
  3. Have more role models. This is really not something that we can just make happen. Whether someone reaches a “role model status” or not depends on a lot of factors, e.g. career choices, personality traits, level of Julia knowledge and experience, culture, etc. I think we already have a number of female “role models” in the Julia community so this may be one less thing that needs to be worked on :slight_smile:

Anyways, this post is getting too long for my liking and isn’t offering much beyond paraphrasing the “known” so I will stop here.

12 Likes

On second thoughts, my post above displays insensitivity and is exclusionary. I’m sorry and
I take it back.

I think the proper way of getting feedback is through an anonymous
survey. Asking for a posting here (under more or less identifiable handles) would
exclude people who are not comfortable providing these disclosures.

Perhaps @joaoui1 would like to manage this survey and post the results here?

4 Likes

There once was a similar topic:
https://discourse.julialang.org/t/creating-a-world-of-programming-language-to-be-inclusive-as-possible
it wasn’t started by me but splitted like this one.

After it was closed I asked myself what I did wrong and why it was closed. I am still not sure about the close.

This topic is already a problem again it seems to me:

The problem I see here is that we want to talk about being inclusive or lack of diverstity but some people (not necessary @mbauman the quote is just an example) are extremely exclusive on the way this topic should be talked about and what opinions are valid in such a discussion and what is not.

In my opinion, if we talk about increasing the inclusiveness in this community we have to deal with all opinions on this matter, even if someone expresses something against inclusiveness or against more women or against some kinds of diversity, whatever.

If a discussion like this is not really open to all sides those who like to contribute to this matter would have to be extremely cautious on how they speak (like @PetrKryslUCSD above) . And this can not result in a good discussion if everybody has to avoid the holy inquisition.

We can’t improve on this matter if talking about is already censored.

4 Likes

Great post @mohamed82008. A few things to mention.

Or fields. I’ve noted before that some scientific fields, like biology, pharmaceutical sciences, and social sciences, tend to be a lot more diverse than the core fields that a lot of the Julia founders and package developers come from (mathematics, computer science, and physics). One effective way to increase diversity in our community may be to more actively recruit in these fields.

I think things like the NumFOCUS small development grants, and also making sure there’s continued funded student programs, helps this. Open source naturally selects for those who have the time and privilege to invest in something that either won’t have professional payoff or if it does, it may take years before there’s any clear advantage. It’s no surprise then that most people running open source communities come from places of privilege, since you need some kind of safe background in order for getting involved in open source to be sensible. Funded student programs is one way to break down those barriers, and it’s by breaking such barriers that we can begin to make major inroads into diversity of our community.

Indeed, this is great to have, but we do need to ensure that the burden of increasing diversity does not fall onto the underrepresented individuals of our community themselves. Too much diversity work in the hands of such individuals could detract from other professional activities and accidentally have the wrong effect. In that sense, we should not only attempt to foster more underrepresented models but simultaneously try to find ways to support them in that the work is shared by others as well. This means its important to have as many individuals as possible as part of such diversity initiatives! Exactly how to do this balance is hard, but it’s definitely the right question to ask.

15 Likes

There’s a whole literature on this (shameless self-plug, though I’ll note that men self-cite more than women). Unfortunately, it’s not super clear why this is the case.

Exactly what I’ve been thinking about. Of course, this is just one approach, but hopefully a fruitful one. Getting back to the OP, I think this is one of the reasons that the group of R users is so diverse, because it’s so widely used in biology and ecology.

5 Likes

I don’t like the concept of “diversity work”. Promoting diversity is just an education and marketing task primarily. What I mean by being a role model in this context is just doing whatever you are good at and having your work highlighted by the community and maybe even your character. It’s not much more than that. Anything beyond that is purely optional and so people can simply bow out when feeling overwhelmed, nothing shameful about that.

1 Like

Hi, I can create a google forms for the survey, but I’m extremely new to the Julia community, any idea where I should post it to get good answers?
Also maybe a more official effort would be better, made by people with more ability to change things and that could use Julia’s official channels to spread/promote the survey

3 Likes

Wow that discussion was painful, was gonna read through it, but after about 20 replies I couldn’t go on. Thankfully this discussion seems to be going towards a better direction

3 Likes

Talked to a friend and we ended up with the following 7 questions (better keep it short) partially inspired by @PetrKryslUCSD

  1. What is it that you think would help women programmers most?
  2. What turns you off from the Julia community?
  3. (for people in other communities) Are there some positive changes that seemed to have made a difference?
  4. (or people in other communities) Are there some steps the community took but that did not seem to be effective?
  5. Do you support the creation of safe spaces/women only spaced in Julia’s official communication channels? (i.e one channel in slack, one category in discourse…)
  6. Are you interested in being mentored by more experienced members/developers from the Julia community
  7. Are you interested in founding or being part of a R-ladies/Pyladies like group?
9 Likes

Should we extend the survey to other groups as well or just women for now?

We could ask what minority the person is part of

1 Like

Maybe its better to focus on one group at a time otherwise the survey could get too long/distracted from the goal.

1 Like

I would suggest to create a Google survey (https://support.google.com/docs/thread/6842711?hl=en), and post the link here. Perhaps in a separate thread with a clear subject label to draw attention.

1 Like

What do you think?
https://forms.gle/mto1ujy1SA5VFpCs8

1 Like

Might help to word this more neutrally, since some people might not be turned off from the community.

1 Like

image

Perhaps “Is there anything that would turn you off from the Julia community?”

image
I believe that these questions my apply not only to people in other communities, but in this one as well. There have been some steps taken to make discourse more welcoming etc. So: let’s find out what worked and what didn’t.

But you are right: the question could also be directing respondents to their experience in other communities as well.