Mentoring and diversity for Python [LWN.net]

Mentoring and diversity for Python

Posted Jun 21, 2018 7:54 UTC (Thu) by raegis (guest, #19594)
Parent article: Mentoring and diversity for Python

A quote from one of the provided links (emphasis added): "CODE documentary exposes the dearth of American female and minority software engineers and explores the reasons for this gender gap."

Notice how the writer forgets he or she mentioned "minority" by the end of the sentence. There is a lot of talk about diversity for "women and minorities" when people really mean just women. This meaningless "minority" name- dropping is everywhere, and getting sort of tiring. (I'm an African-American male, by the way.)

Aggressively recruiting women is awesome. Pretending to recruit "minorites" when you're just really concerned about female representation masks a different problem which requires different outreach. Note that I have no problem with women recruiting women only, and I strongly encourage it. I'm just suggesting more honest language, so that organizations which promote diversity do not overlook non-female "minorities".

to post comments

Mentoring and diversity for Python

Posted Jun 21, 2018 11:46 UTC (Thu) by vstinner (subscriber, #42675) [Link] (12 responses)

Currently in CPython, some core developers are trying to mentor more women than men (positive action). There is no clear path how to increase diversity in term of minorities. On GitHub or by emails, it's not always easy to guess if the author of a PR/email is a woman or part of a minority.

Right now, CPython has 2 woman core developers for a total of 171 core developers... or 90 if you only count core devs who are on GitHub... or 34 if you only count core developers working on pull requests (review/merge). We are discussing to move "inactive" core developers in a new "emeritus core devs" list to better track the current number of "active" core developers.

By the way, my notes about CPython core developers (especially how to get more of them!): http://pythondev.readthedocs.io/core_devs.html

Mentoring and diversity for Python

Posted Jun 21, 2018 18:51 UTC (Thu) by Ranguvar (subscriber, #56734) [Link] (11 responses)

> On GitHub or by emails, it's not always easy to guess if the author of a PR/email is a woman or part of a minority.

Perhaps that is how it should be? So everyone is treated equally?

I see merit to introducing more developers who think differently into a project.
There are downsides as well though, so any top-down approach should be careful.

I don't see any benefits towards selecting for specific race, sex, gender, or skin color, however.
In fact, we should be moving **away** from doing so, and towards concentrating on ideas.

> "This is wrong", Wijaya said.

The differences between the career and hobby choices men and women make _increase_ with greater choice, they don't decrease.
The research on that is quite clear, but here is one specific example.

https://www.theatlantic.com/science/archive/2018/02/the-m...
http://journals.sagepub.com/doi/10.1177/0956797617741719

Mentoring and diversity for Python

Posted Jun 22, 2018 22:42 UTC (Fri) by rwm (guest, #104883) [Link] (6 responses)

I sometimes lament how the Internet has panned out. Profile pictures abound, they add no value when browsing a Github repo. Real names everywhere, in Facebook, Linkedin, along with your personal and professional life stories.

I had hoped we'd move to being increasingly anonymous, devoid of skin color, gender, existing in communities where "...nobody knows you're a dog". I wanted hiring to be the same. I wanted judgement based on patchset, not identity intersection.

But instead we seem to be going in the opposite direction, everyone's talking about gender and skin color, and how we can get more of some genders and colors into ${project|company}.

A well meaning re-architecture of society is being attempted at the moment, but it is one that frequently confounds equal opportunity with equal outcome. I'm not going to say they're right or wrong, but we do have to be very careful - for the tools being employed to bring about systematic change and fighting the alleged gender and racial discrimination are frequently themselves gender and racially discriminating.

Mentoring and diversity for Python

Posted Jun 23, 2018 4:54 UTC (Sat) by raven667 (subscriber, #5198) [Link] (3 responses)

I understand your position but if you think logically about how the ethics of this approach work in practice you will find that people are not, have never been, and probably will not ever be truly color-blind, so members of the dominant social group, in this case white hetero males, can freely choose to be anonymous or use their real names and faces without any particular repercussions but members of non-dominant groups are basically required to be anonymous to be treated fairly, deny the basic truth of who they are and how they live their lives, for the comfort of those who would treat them poorly. One would need to maintain near-Intelligence-mole levels of opsec to stay anonymous, because most people work on teams, go to conferences, engage in team building (see LinuxCon, etc.) where they could never truly hide these basic facts about themselves so even if you could maintain that kind of anonymity you would always be at a disadvantage to those people with a more conventional reputation, given an equivalent skill level, so this doesn't actually solve the negative effects of racism and sexism, in and out-group discrimination, in practice.

In practice what works is to make note of when someone is in one or more protected categories, that have been demonstrated to experience discrimination, and take conscious deliberate (one might say affirmative) effort to undo the effects, by rolling out the welcome mat through mentoring programs to generate a pipeline of qualified people and then when it's time to include more people on the team deliberately choosing from the pool of qualified candidates the one who increases the diversity of the team.

As more people get used to working closely with peers who are different then themselves in these fundamental ways, then the base level of racism and sexism will lower and there will be no advantage in hiding your identity. Its never the person who is discriminated against's job to "fix" the discrimination, it is the responsibility of the moral agent who is choosing to discriminate, rather than discipline their negative tendencies, to learn to be better people. Hiding all the non-dominant people just makes that harder, there is no easy fix.

Mentoring and diversity for Python

Posted Jul 2, 2018 6:06 UTC (Mon) by Wol (subscriber, #4433) [Link] (2 responses)

> you will find that people are not, have never been, and probably will not ever be truly color-blind, so members of the dominant social group, in this case white hetero males, can freely choose to be anonymous or use their real names and faces without any particular repercussions but members of non-dominant groups are basically required to be anonymous to be treated fairly, deny the basic truth of who they are and how they live their lives, for the comfort of those who would treat them poorly.

How true.

The recruiters need to confront their own prejudices (which means they need to be allowed to talk about it!!!), and then the only questions that should be asked are "Can this person do the job? Will they add value to the team?".

To give a couple of examples when I was recruiting, we had a far-Eastern guy apply. Can't remember whether he was white expat or Chinese, what I do remember is he was brilliant but the reaction of everybody on the recruitment board was "yes, he's technically the best candidate we've had, but do we really want to work with him?!". For a different job, we had a black guy turn up, but as I put it, he was "English first, black second", a very nice guy and he got the job. Compare that with a black lad I was at school with - "nobody likes me because I'm black". No - nobody likes you because you keep playing the race card.

The problem with being a member of the minority is that if your face doesn't fit, it doesn't help you get the job. Quite how you square that circle I don't know, but if you can do your first-round selection on technical merit, and your second-round on added-value (bearing in mind that diversity is almost always a value-add!) then hopefully things will improve. Except that this will probably then bring on claims that you are discriminating against the majority! :-(

Cheers,
Wol

Mentoring and diversity for Python

Posted Jul 2, 2018 10:40 UTC (Mon) by farnz (subscriber, #17727) [Link]

The other "exciting" part of the problem is the presence of a minority of genuine bigots who know how to disguise their bigotry as real concerns about someone's ability to add value to the team. So, you get an awesome candidate who seems perfect to all but one team member, who then raises concerns like "will this candidate be a team player? Will they be able to cope with the workload we face?" meaning "I want to sow doubt about this person's ability, because if we hire them, I'll have to hide my distaste for the targets of my bigotry".

And this is deeply difficult because it's carefully disguised to be exactly the same as a non-bigot raising real concerns; distinguishing bigotry, ignorance[1] and real issues is a challenge that I don't think anyone has a solution to.

[1] Ignorance in this sense being a complete lack of knowledge, such that you raise concerns that can be addressed by educating you - e.g. someone being concerned that a candidate with a First from the University of Cambridge is perhaps underqualified compared to someone with a degree from Harvard.

Mentoring and diversity for Python

Posted Jul 2, 2018 12:04 UTC (Mon) by fb (guest, #53265) [Link]

> Compare that with a black lad I was at school with - "nobody likes me because I'm black". No - nobody likes you because you keep playing the race card.

Living in Europe at "the wrong end of the minority stick", I have sympathy for that kid at your school.

It can be tough to drop or ignore the subject of race/origin/background ("race card") when it is *constantly* brought back to you. I think part of the trouble is that people not "at the wrong end of it" don't get how often one gets that card waved at you.

FWIW Ever thought at how bad teenagers are at regulating emotions and dealing well with conflict? How well do you think *you* would have dealt with racism at school age had it been towards you? Not just one aggression but all the years of it.

[....]

/me recently visiting the office where most of my team works:

Coworker: "Don't be too sensitive. It was just a joke."
Me: "I've heard that same 'joke' four times today."

Mentoring and diversity for Python

Posted Jun 23, 2018 8:18 UTC (Sat) by mpr22 (subscriber, #60784) [Link]

The thing is, the Internet has shown us what kind of communities we get when everyone is anonymous: 4chan.

Mentoring and diversity for Python

Posted Jun 23, 2018 14:20 UTC (Sat) by rweikusat2 (subscriber, #117920) [Link]

Realistically, "nobody knows you're a dog" means "everybody assumes you must be a dog".

Mentoring and diversity for Python

Posted Jun 23, 2018 18:04 UTC (Sat) by dskoll (subscriber, #1630) [Link]

That Atlantic article was very interesting. Thanks for the link.

Mentoring and diversity for Python

Posted Jun 27, 2018 13:23 UTC (Wed) by nilsmeyer (guest, #122604) [Link] (2 responses)

I wonder if there is true merit in having a lot of people who "think differently" in a single project. It sounds like a source of conflict more than a source of more quality contributions (which should be the goal).

It's also interesting how often people underestimate the diversity of open source projects, by virtue of being open you'll usually have a lot of people from different backgrounds contributing. Of course this is usually limited to English speaking people with sufficient economic resources / free time to contribute.

Pointing out that certain groups are under-represented usually comes with an implied allegation of sexism, racism or another -ism. I think there should be some burden of proof for that.

Mentoring and diversity for Python

Posted Jun 27, 2018 15:10 UTC (Wed) by misc (subscriber, #73730) [Link] (1 responses)

I do not understand, you are the one suggesting there is something implied, shouldn't the burden of proof regarding the implied allegation be on you ? As otherwise, it would seems that you are making stuff nobody said, and then asking to prove while no one said anything. That sound really weird and I do not get the logic.

Also, yeah, that is really interesting how people misunderstand numbers and several studies regarding diversity instead of seeing how different are people from West coast and East coast, indeed.

Mentoring and diversity for Python

Posted Jun 27, 2018 16:01 UTC (Wed) by nilsmeyer (guest, #122604) [Link]

To give one example, people often say that a different CoC (code of conduct) is needed (it says so in the article), saying that the CoC adopted by the PSF is insufficient while the CoC of Pycon is better. The Pycon CoC lists a lot of "unacceptable behaviors". To me it follows if you need to spell out what is unacceptable or inappropriate these things are alleged to happen on a regular basis and a new CoC is needed to purge these destructive elements - the latter being conjecture on my part, perhaps there is a more innocent explanation and I'm being hypocritical here.

> Also, yeah, that is really interesting how people misunderstand numbers and several studies regarding diversity instead of seeing how different are people from West coast and East coast, indeed.

To which continent/country are you referring? Assuming you're thinking of the United States, you're kinda making my point for me since some US Americans often have somewhat warped views on that topic (categorizing people by somewhat arbitrary races). Of course when you look at a community that predominately communicates in English you'll find a lot of "white" people, it's an easy categorization to make but far too broad to assess a whole group.

Mentoring and diversity for Python

Posted Jun 22, 2018 17:16 UTC (Fri) by k8to (guest, #15413) [Link]

Maybe I'm emblematic of this problem.

I tend to think that "obviously" we should be able to have female participation with comparable numbers to male numbers, and that the lack of this shows a clear problem. I also feel like I've talked to a lot of people about this problem for the last decade and change, and think I have some idea about what the problem looks like specifically and how it plays out and what sorts of things are likely to help.

For participation from what we think of as racial groups, I have no idea what the numbers should be, and I don't really know what the biggest problems are and I don't know what would be likely to help.

I know there are some things that will help both scenarios. I've pushed for removing names from resumes in the hiring process, and encouraged dismantling talk of "meritocracies" which typically damage all out-groups. I have come to believe that broadening the mechanisms by which companies seek talent is necessary to avoid hiring mostly "people like us", including seeking talent from different physical regions and different schools, and that companies must be willing to hire junior staff to get access to groups before they give up on the industry entirely (never mind that the idea of hiring only people who you expect to come in with all the knowledge they will need is dumb in other respects.)

But I think this mostly comes from the targets for women being more obvious, and it having become a spotlighted issue recently drive this to a significant degree.

Or to be more awful, what is it like to be an african-american software engineer? I don't know. I've met maybe a handful in my life in the silicon valley bubble, and I'm not terribly enamored of turning the conversation into one of stupid, clumsy questions if I don't know them very well.

I guess I'm just saying you're probably right at a number of levels, and that I think there are a number of hurdles to getting to a more reasonable place. I spent time in a diversity and inclusion working group for around 6 months in a company that was pretty drastically diversity-challenged, and there were no serious discussions of minority outreach beyond a footnote of going to some job fair / career thing or other. We had east asian and south asian participationin the group, but in silicon valley software I don't know if these groups particularly think of themselves as minorities. (They don't get the same advancement opportunities, but representation is very high.)