How Duolingo achieved a 50:50 gender ratio for new software engineer hires
making.duolingo.com
making.duolingo.com
This isn't justified anywhere, it's just taken as an assumption. It would be good if these implicit positions were justified rather than imposed. The Ys getting discriminated against, will understandly be angry about it, especially since only a small percentage of the Ys actually benefitted from historical discrimination against Xs, yet all of them are now eligible to be reverse-discriminated against.
Also Duolingo has ~100 employees total. There's no way that the industry's demographics are somehow limiting their available talent pool. Having a policy like this sounds like a great way for a small company to attract highly talented women.
However if they don't assume that, then they are effectively worsening the 18:82 ratio for other organisations who will be logically and mathematically unable to reach a 50:50 ratio without leaving (82-18)/82 of Y unemployed, so they would be virtue signalling.
So either they assume that, or they are virtue signalling, or they are trying to make (82-18)/82 of Y unemployed. Q.E.D.
I was trying to Assume Good Faith in my first comment but sure, call me out for not including the "asshole options".
edit: yeah sure, downvote me for pointing out that badly-thought out ideological policies don't work mathematically
> This strategy implicitly assumes
That is the real objective assumption that any implementation of this strategy makes (or the other "asshole" choices).
Good intentions are worthless if they can't work mathematically.
EDIT: it's clear that you believe ideology is what's motivating this, despite your claim that you don't care about motivation.
This article is hitting the front page because some people think it's impressive. I don't think it's impressive nor do I think it's solving any real issues. I pointed out in very neutral terms why it's not solving any real issues, but some people some to feel offended by my pointing out a "trivial point".
"Suppose there was overwhelming evidence that 95% of women were terrible at technology and 5% of women were awesome at technology. There are roughly 7 billion people on the planet, roughly 3.5 billion women, roughly 1.5 billion women who work outside the house for a wage. In this scenario, where only 5% of women love technology, there are 75 million working women who are awesome at technology. According to the Bureau Of Labor Statics, the USA had 1,256,200 software developers in 2016. The BLS also tracks some other minor categories, such as Web Developer, which have about 150,000 jobs. Lump all the sub-categories together, and let’s say there are 2 million such jobs in the USA. Let’s be wildly generous and double the number for the EU, and triple it for Asia. That gives 12 million software developer jobs in all of the advanced and developing economies. So even with exaggerated assumptions about women’s inherent weakness in technology, we still end up with a scenario where every single programming job in the world can be filled by a woman who will be awesome at the job. There is no need for men, at all, in the tech industry."
http://www.smashcompany.com/business/why-are-women-being-pus...
More to point, women are being pushed out of tech. Women constituted 35% of all software engineers in 1990, but only 26% in 2015 (see above url). Given the rough figure of about 2 million jobs, that 9% difference suggests there are 180,000 missing women from engineering jobs right now. Initiatives like Duolingo help reverse the trend, and therefore should be praised.
Why is the same level of angst not applied to education, where <20% of primary school teachers are male? Have the public schools been pushing men out, or are men just not interested? Or maybe it's some of both? Do we even know?
It's no surprise that the efforts to achieve gender balance in a highly-paid industry gets a lot more attention than any gender-imbalance in a lowly-paid industry.
Put another way: men weren't too upset about being pushed out of teaching positions because they were able to make better money elsewhere.
If you wanted to talk about gender parity in software development then you would need to use 0.7100086% of women (who are employed) are (expected to be) good at tech, 99.2899914% of women (who are employed) are bad at tech.
There about 387,508 female programmers (.212 mul 473000) + (1536000 mul .187) and about 1,621,492 males ((1536000+473000) - 387508) [2] according to more recent statistics. From the same report it looks like 46.9% of the workforce is female with a total number of 153,337,000 workers for 2017. If we apply the 46.9% number we get 71915053 females. If we further apply our 0.7100086% multiplier we get 510,603 females who we'd predict to be software developers. This number is higher than but close to our 387,508 figure from earlier. This was off by 0.76x
If we apply the same logic to males we will find that of the 81,421,947 (153,337,000 mul .531) working males after we apply our 0.7100086% from earlier we will find that the number we get extremely under-predicts the number of software developers at a predicted 578,102 males. This is off by a 2.8x
I have a few questions that I have not been able to answer. They are as follows:
- Why is my percentage of humans who go into software more accurate for women?
- Why is my percentage of humans who go into software off by 3x for men?
- There are many higher paying and lower effort jobs that are male dominated that men could choose to go into? Why would this skew be so great for software engineering?
- Are there other careers who have a similar factor of inaccuracy except flipped (3x more women then predicted, .7x the predicted number of men?)
- Does this error factor show dependence on salary? Hours worked? Location of jobs?
[0] - https://en.wikipedia.org/wiki/Software_engineering_demograph... [1] - http://www.dlt.ri.gov/lmi/pdf/usadj.pdf [2] - https://www.bls.gov/cps/cpsaat11.htm
Why are software engineering jobs being looked at in a vacuum? And why is capability to do the job the only metric being looked at here?
We could do the same thing with men and teaching children (the distribution there is even more skewed than software engineering, only the opposite direction) and say that there's no need for female teachers. We won't because that also simplifies a very complex equation to a handful of variables. It appears that most men don't want to be teachers because the incentives don't align with what most men want. Teaching is a field that doesn't pay very well, but requires a lot of social interaction, compassion, empathy, and generally is suited to more maternal people (this may be considered controversial, but on average women are stronger in these traits, according to psychological studies). Software engineering is the opposite.
Full disclosure: I'm a male software engineer that lives with a female preschool teacher. She was on the path for medical school (a field also somewhat dominated by women currently but pays much better than teaching) but decided that becoming a teacher would be more rewarding for her.
We've discussed our career choices and I realized I could never do what she does because it requires more patience and compassion than I want to exert on a daily basis, I wouldn't feel as much gratification from it as she does, and it pays a fraction of what I currently make. I realize this is an anecdote, but it does represent the average distributions of male and female careers and the incentives for each gender.
Could it possibly be that in general men and women want different things out of a career? And maybe software engineering offers more of benefits that are valuable to men, in which case it would be fine that there are more male engineers, just like it's also fine that there are more female teachers.
I'm using in general and on average a lot here because I don't want to imply that there aren't compassionate men or women who want high paying jobs - they're just less common than their opposites, on average.
And note that I'm not saying that these preferences are due to biology - whether it's that or sociological factors doesn't really matter here. Trying to artificially equalize the distribution between males and females in a given industry doesn't make sense either way.
A few years ago I watched a documentary, «The gender equality paradox» [1]. It discussed the strange fact that it is far more likely to find make masons and female nurses in Norway, despite the fact that everything in that society pushes towards gender equality.
In that documentary, I heard an interesting fact: the proportions between make and female engineers are much farther from 50:50 in Norway (where men are more common) than in India and other less-developed countries, where the two sets are more evenly balanced.
The documentary provided this possible explanation: let's suppose there is a biological reason why it is less likely for women to be interested in engineering topics, and men in works requiring social skills. One could imagine that in those countries where there are plenty of jobs and possibilities, everybody is free to pick the job they like most. On the other side, in less developed countries there isn't as much choice, so people are more willing to accept things different from what they would have chosen otherwise.
It is clear this is just a theory, but I find the basic fact of the existence of this Norway/India difference intriguing.
Women constituted 35% of all software engineers in 1990, but only 26% in 2015
It's one thing to value merit over politics -- I'd expect nothing less from my peers in the bay. However, it's an entirely different matter to value merit over long term strategic objectives, and from what I can tell, an alarming number of commenters here are bereft of any understanding of the long term implications of not striving to increase the potential engineering talent pool.
We in the US are in a situation where:
1) talent is scarce
2) companies, startups more specifically, are now less willing to hire international talent for fear of visa/etc. risks in the current age,
3) politics are disincentivizing efforts to broaden the talent pool via obvious means -- such as by reintroducing and motivating women to participate.
The end result of this isn't going to be job security for existing participants in the tech industry here in the US. It's going to be the flight of entrepreneurs to talent centers where talent can be acquired more reasonably, and right now, that's starting to look less and less like the United States given the above.
Shoot, even YC is broadening beyond these geographical and political boundaries through tools like Startup School.
If your goal's to keep our tech industry in the US from becoming a laggard, it's in your interest to encourage the system to strategically start motivating and representing the population space more accurately. Otherwise, the talent drought will just drive ideas overseas.
(I apologize to any overseas readers reading my comment. I hope you understand.)
TL;DR: very few people here are thinking about this from a long term strategic angle; it seems like everyone's just concerned about their own jobs.
...it's a strategic play. Changing hiring needs influences post-secondary institutions which influences secondary and primary schooling decisions at the policy level, or such is what's envisioned.
Same as I've said above. I also made a point about people not understanding the strategic play, which you've evidently validated.
Can I get a source for this?
https://www.bostonglobe.com/business/2016/02/19/the-war-for-...
https://m.huffpost.com/us/entry/3900575
And a seemingly contrarian source which actually supports the argument I made:
https://techcrunch.com/2015/11/12/the-tech-talent-shortage-i...
> Racial And Gender Inequality
> Opportunity inequality has become so pervasive that minority tech employees are banding together with the explicit purpose of increasing their numbers in the space. If you’re a tech leader, you must fight against this inequality to benefit from talented people of all stripes.
> Racial minorities are woefully underrepresented in the tech industry. And it’s not just a matter of race, either. Many don’t remember it, but women were once much more prevalent in tech. However, the proportion of female computer science students has halved since its heyday in the mid-1980s. And the women who do enter engineering are leaving at an astonishing rate.
> These problems aren’t self-correcting; their resolution requires dedication from leadership. We’re all obligated to take the first step to reverse this worrisome, decades-long minority flight by acknowledging the issue and magnifying diverse voices.
> If our industry has any hope of becoming more inclusive — which is a key step to solving our so-called “talent shortage” — we must be understanding, empathetic and open to others’ viewpoints.
Would you like me to go on? I thought the talent scarcity problem was obvious, but I've got no qualms with more sourcing.
https://medium.com/@simon.sarris/why-is-computer-science-enr...
You can do all the clubs and scholarships you want, but the proof is in the hiring
Or will you hire a female even if she sucks at the job because your team is only made of males? Are you filling a sticker collection or building a team?
All companies cherry pick from the best Universities. The fact they've had to look elsewhere means they're not getting the best from the best Universities (based on rankings). They're getting the best female graduates possible. I assume these are good enough but not the best.
Male hires are probably still from the best universities. This creates an imbalance in competence. The female hires are good enough, the male hires are exceptional. Male hires move up the ranks on virtue of their competence. Other male hires leave after a couple promotion cycles expire.
Now we have situation where male hires are predominantly in coveted positions in the company and female hires are mostly junior mid hires.
The equality problem has now shifted up. What do you do then? How far is a company willing to go with an equality experiment before shareholders start noticing the erosion of the companies competitive advantage and start protesting.
I appluad Duolingo for it's boldness. Am doubtful on its sustainability without either the rest of the industries following suit while pre university educational centres attempt to rebalance.
Is there any objective evidence that people who graduate from the "best universities" become the most competent engineers?
Regardless whether this is the right thing to do companies fall back to this approach time and time again.
I don't know why people think turning women into "butts in seats", as opposed to genuine respect and admiration of skills and abilities is going to make this problem better.
Correct. But the goal here I suspect is to get enough corporations onboard such that schools (k-12 and undergraduate/graduate) are now motivated to select for and encourage balanced talent pool participation for fear of missing out on "hip" companies.
Being willing, and taking satisfaction from beating your head against a computer to get it to do what you want it to do is something I just don't observe as being wide-spread in any population group. But we need as many of those people as we can find, wherever they happen to be and whatever they look like.
I guess some universities care about which companies recruit there (for marketing purposes, being able to say companies A,B,C recruit there), but does this really apply to k-12?
Also, if most "hip" companies chose to only recruit at universities with gender ratios exceeding some threshold (and by causality this happens before universities are "properly motivated" to encourage balanced gender ratios in CS), won't they be hurting themselves by competing for an arbitrarily chosen much smaller pool of talent?
I was replying to a comment that says:
> the goal here I suspect is to get enough corporations onboard such that schools (k-12 and undergraduate/graduate) are now motivated to select for and encourage balanced talent pool participation for fear of missing out on "hip" companies
And asking how will "fear of missing out" motivate k-12 schools to get more girls into those classes. Companies don't recruit in k-12 schools. How would duolingo and "other hip companies" recruiting exclusively from universities whose CS classes exceed a gender ratio threshold influence k-12 schools?
what about the fact that more men are dropping out of school and less are entering college? shouldnt we be encouraging those men to be engineers as well?
as far as encouraging undergaduate/graduate to have more 'balanced' classes, since there is a limited number of seats I can't see that going any other way than: you have a penis, thats a negative on your application.
I'm not here to entertain your Tu Quoque fallacies today. :)
I made no claims about your past actions or past claims.
I definitely think you could have left out the “holding the door for feminists” part though.
But how do you know they weren't hired for their skill set? Like, if you have two equally good apples in flavor, but you want to balance the profile of varieties, choose the one you're most lacking?
It's going to be much harder to get 50:50 excellent apples than 20:80 excellent apples.
How will you tell you won't be developing some bias over time just to make the hiring process an easier job for you?
There are an exhaustingly large set of companies that hire exclusively from the set of men, leaving competition for the smaller set of women much easier.
The entire discussion doesn't address the systemic changes which happen if nearly every female applicant is employed, of course; one obvious thing that would probably happen is that more females would consider applying to this sort of job, which most would agree is a good thing.
I don't know of any large companies that only hire from candidate pools of men. Does anyone know of an example?
Recently, 88.6% of developers surveyed identified themselves as men[0]. That ratio alone suggests that the vast majority of teams and companies have zero female developers. In my experience, women developers often cluster at companies (usually large companies) so that they are not isolated, which might make the numbers even more extreme.
[0] https://insights.stackoverflow.com/survey/2017#demographics
STEM is hard and it sucks.
Same reason women don’t have to mine coal. They have the luxury of being able to choose not to.
Men don’t have that luxury. We either earn or we live under a bridge eating food out of cans. Society puts an enormous pressure on men with the threat of homelessness and isolation, which doesn’t exist to anywhere near the same degree for women, because women have inherent value.
Men have no inherent value to society, their only value to society is realized through their work.
The result is men work harder and endure harder shittier jobs, including STEM.
If you want more women in STEM, then just eliminate all of their societal privileges, consider them inherently valueless like men, and put them under the same level of threat men are under of being isolated and going homeless.
In a randomized double-blind study (n = 127), science faculty from research-intensive universities rated the application materials of a student—who was randomly assigned either a male or female name—for a laboratory manager position. Faculty participants rated the male applicant as significantly more competent and hireable than the (identical) female applicant.
https://www.washingtonpost.com/news/morning-mix/wp/2015/04/1...
> And according to their latest study, published Monday in the Proceedings of the National Academy of Sciences, women are no longer at a disadvantage when applying for tenure-track positions in university science departments. In fact, the bias has now flipped: Female candidates are now twice as likely to be chosen as equally qualified men.
When I was hired out of college I didn’t know if it was due to my abilities, potential or because I was a cheaper hire than others.
I still showed up to work and got paid.
More or less "We realized we were looking solely at candidate performance when we were selecting a candidate pool. That wasn't producing the warm fuzzies, so we ignored our best indications of candidate performance, and instead hired from a pool that made us feel good about our arbitrary political goal while hopefully just stopping short of illegal discrimination.".
This doesn't really fit that narrative.
¹ https://www.smithsonianmag.com/smart-news/computer-programmi...
I think before you say that the highest performing schools "perpetuated the problem", you're going to need to substantiate that.
> https://www.smithsonianmag.com/smart-news/computer-programmi... .
I haven't thought about this topic for a while, but an analysis I heard goes a bit like this: Typist used to be a "woman's job" as well. My understanding is that the job was split. What you would call programmers today were typically called "architects" or similar. Lumping them all together you obviously get a serious skew toward women because considerably more technicians/programmers (where the ratio favours women, but not exclusively) were required for computers of the time (trailing off through the '80s as stored program computers became the hotness) than architects/designers (where the ratio favours men, but not exclusively).
I don't actually believe so given that the source link did so for me. The problem was substantiated in Duolingo's own metrics and reasoning for switching, the problem in this case being that their traditional college talent pools were overly biased towards men -- well, colleges select their own students. :)
> We secured the best new graduates for software engineering roles, but we noticed something: they were all male.
So they did decide to look pass candidate performance.
The problem is that they are bragging about simply hiring more females, anyone can do that. How the company will perform as result of that is what matters and they didn't address that at all.
The other rather interesting thing is that they seem to have introduced all of these measures at the same time; so they won't actually be able to analyze or compare the performance outcomes for a given year. The people who like the policies will say they're working great, but the year just didn't produce as many hires which turned out well (except perhaps through candidates from the schools which they did not remove from the list).
Not quite. Your quote is missing context: they hired the best candidates from the schools they visited. That’s not the set of all schools, though!
They tried to correct this by visiting schools whose CS demographics more closely match the general population.
> this meant going to some never-before-visited schools as well as cutting out a few schools that had been on our list before
They wouldn't have eliminated those schools if the new schools produced better candidates.
EDIT: Wait, I rushed and misread your comment. Why wouldn't they eliminate these schools if the new schools produced better candidates? I would imagine it was the opposite: remove schools because the new schools produce better candidates, and time is finite.
Judging from the number of "tech hiring process is broken" posts I see on HN, it seems pretty dubious to claim that existing indicators of candidate performance are actually measuring what they claim to be. So, if you look at the situation from that point of view, then there's no reason to elevate existing metrics to anything higher than "meh, maybe it's measuring something."
Those personality traits are part of performance as well; a combination of diligence and intelligence tends to predict both job and academic performance. You can be a genius, but you're not in that pool if you aren't willing to do the work.
Personally, I think the last reason provides an arbitrage opportunity that let’s small companies that focus on women compete for talent in a way that FANG can not.
None of this stops a candidate from applying to duolingo directly. There's no discrimination going on against humans, and discriminating against a school is totally fine so long as there's no suspect reason to do so (e.g. discriminating against Howard because they've got primarily African-descended students).
Not a lawyer. Would welcome a legal opinion.
I would highly recommend reading at least the Wikipedia article on "Disparate Impact."
They very clearly make the case that they are biased in favor of women -> over men. You might think that 50/50 means equality, but up front they state that there are only 18 percent women in the field. In order to achieve 50% women in their workforce they have to be biasing the hiring process. It seems strange to go out and write publicly about doing so.
It actually makes sense. Assume men and women are equally competent on average (probably true or at least close). If your applicants are 18% women and the applicants are randomly sampled from the talent pool and your interview is totally unbiased you should get 18% women hires.
If you convince a lot of (randomly sampled) women to apply so you have 50% female applications, you should still get 50% female hires. I don't think this is unfair on men as long as your outreach activities don't at all affect interview chances. In particular they shouldn't give an "inside track" to bypass HR.
Sounds like they've at least thought about that and made the first stage blind.
This is very similar to redlining and other implicitly discriminatory practices in the past, and the longer it goes on the more likely I suspect it will end up in court.