The Github results were based on the assumed gender of submitters based on things like having female anime character avatars and "female sounding" usernames. I would posit that methodology as highly flawed.
Women are more likely to get hired to solve the unequal distribution of female/male engineers. This makes it more likely to have less qualified female engineers, and more qualified male engineers, since hiring requirements for men are more strict. A higher average of under qualified females compared to under qualified male engineers will lead to more code getting reject from the female counterpart.
Not necessarily. If the pool of talent is exactly the same(let's say 1000 women and 1000 men) and your interviewing process is perfect at determining engineer quality, but you are hiring 80 women and 20 men, then you'll hire the top 2% of men and the top 8% of women. Again, just assuming that the talent is exactly the same, the women you've hired will still be worse.
In reality, the pool of female talent is very shallow. Again assuming the talent levels are exactly the same, there's just not enough highly-skilled women to go around, e.g. if you're trying to hire the top 1% of engineers, and you have a pool of 1 million men and 100,000 women, then you're looking at 10000 men and 1000 women as your potential hires. If the top 1000 women are already taken, then you're forced to relax your interview requirements for women. In fact, if the pool is completely free for your usage, but you're looking to hire 10000 men and 10000 women, you'll be hiring the top 1% of male talent and down to the top 10% of female talent, which should lead to a difference in output.
This doesn't mean that "sexism" isn't real or that it's not a problem women face in the workplace. The question however, is how to measure the effectiveness of "anti-sexism" practices employed, if you cannot measure the effect of "sexism" itself, as it is entangled with all the other, confounding factors. You can't optimize performance if you can't measure the effects your solutions have on it.
the onus shouldn't be on multinationals to hire 50/50 gender, but on smaller companies that need lower skills and can accommodate junior roles.
or maybe multinationals should be honest about hiring juniors, and think that would be good PR for them rather than just let people use their staff as a statistic.
Not at all.
If the skill requirement is lower for women to get into the same position as men, is it not rational that, on average, those women will perform poorer than the men?
Imagine they did the same thing for university entrance: men need all prerequisite courses and a 3.8gpa, whereas women need only 75% of prereqs and a 2.9gpa.
It only makes sense that the women are going to score lower in exams/flunk out, since their initial starting point was, on average, further behind.
There are certainly good and bad developers no matter what gender.
If you have stricter hiring policies for men and e.g. only the top 10% of male candidates land a job while the less strict hiring procedures for females get the top 40% females a job, then you created an imbalance in skills between the genders in your company (and only there), and this is what might have happened here
No, there's no biased assumption that women are worse developers, just one about relative numbers and relative skill distribution.
If the number of qualified female developers is less than the number of qualified male developers, then any company which manages to achieve hiring parity has either lowered the qualifications it demands from women developers, or increased the incentives it offers to them relative to other firms (enabling it to capture a higher share of high-skill female developers). Since the latter approach is obviously discriminatory and quite possibly illegal (not to mention a blatant violation of the principle of equal pay for equal work), the former is more likely.
The result of the former is that the median female developer at that company will have lower skill than the median male developer at the same company, even if women in the population at large have the same or better development skill.
The incentive strategy could enable a firm to secure the best female talent, but besides the ethical and legal implications it would also make things much worse for all the other firms out there: the best female talent being off the market, those firms would be forced to either settle for lesser-skilled women or accept a disproportionate share of male developers, neither of which would be appealing.
Naturally, increasing the supply of women developers to match that of male developers (or reducing the supply of male developers …) would improve the situation, but that's a horse of a different colour.
It's more like... if 1 good female developer gets a CS degree for every 10 good male developers, you would expect the workforce to be 10% female. However, if you artificially force 50% female representation you'll have to start drawing upon bad female developers since the rate of good developers is currently only 10%. You first have to increase the output of good female developers from schools before you can scale up their representation in the workforce, not the other way around.
"(...) However, Facebook employees now speculate that Parikh’s findings mean female engineers might not be rising in the ranks as fast as male counterparts who joined the company at the same time, or perhaps that female engineers are leaving the company more often before being promoted."
> Code rejected due to rank
> Women are of lower rank than men
> We need to promote more women faster
Facebook is about to get a taste of their own medicine.
Facebook took heat for not having a 50/50 male/female workforce, when that is not the ratio of male to female devs available. They hire more females, based on the fact they are female. The code quality of some of these hires was not good, because they were not hired on merit, and they've created a code quality problem.
You're making a lot of assumptions there. Do you have any evidence that they lowered the bar? Also, aren't they hiring kids straight out of college in droves? How do you go about properly assessing "merit"? Isn't that a bit of an unsolved problem in hiring?
To me the whole thing just raises a bunch of questions that don't have any satisfying answers.
I suppose that if a company is going to try to artificially hire a greater number of a single group then naturally the bar must be lowered.
Take this hypotherital:
Bob's factory employs machinists to make parts. Of all machinists, 20% belong to group A and 80% belong to group B. When hiring, Bob's factory selects a number of applicants who pass the test. If we assume that group A and group B have the same average skill level then naturally the employees in the factory will be 20% group A and 80% group B.
If Bob wanted to hire eaqual amounts of Group A and Group B, then he would need to hire as much of group a as possible and as little of group b as possible. If bob could only hire a certain number of group b, then he would naturally only pick the best few.
Thus, artificially eaquallizing the work force in bobs factory must neccecarily lead to a skill gap.
This assumes that a job opening has a perfect distribution of applicants. Let's take another hypothetical;
Bob's factory posts machinist ads on Craigslist. CL turns out to send Bob 90% males and 10% females. Bob decides to try to increase the number of females he hires, and advertises on a second site. Turns out this site is viewed by a higher percentage of females, and thus his number of female applicants increases, and the bar for hiring stays the same.
> If we assume that group A and group B have the same average skill level
This assumption, on which your argument rests, is invalid.
I'm an engineering college student who's had conversations on this subject with admissions officers at my school. The gender ratio of accepted students is much more balanced than the gender ratio of applicants. Every year, the admissions department fields complaints from upset male applicants who were rejected; they feel that they were done a disservice because "the bar was higher" for them than for the female applicants. They point to the different acceptance rates for men and women as proof of this "lower bar" for women.
In reality, according to my school's admissions officers, the average female applicant is significantly stronger than the average male applicant. There exists, apparently, a huge volume of unqualified male applicants, who inflate that "80%" figure despite having no real chance at earning a position. We're talking SAT scores hundreds of points below the typical admitted student, no actual engineering experience, things like that. (I have heard that large companies like Texas Instruments also have this problem, which is why they place restrictions on which specific degrees applicants can have - to filter out that chaff.)
These men remain in the applicant pool where women would've washed out, the admissions officers at my school hypothesize, because it is easier for women to "give up" on STEM fields at the first sign of trouble due to social / societal pressures. Incompetent men, however, can think they're good at STEM more or less forever.
A more reasonable conclusion to draw, instead of "Facebook's male engineers are more skilled than their female engineers", would be the one that is presented in the article by the people who have the data. Namely, that this statistic arises from the confounding effect of seniority: Facebook's more senior engineers tend to be male, since the drive to hire more women has been more recent, and more senior engineers have code accepted more often.
You can still lower the bar below "hiring straight out of college in droves".
Hiring more self taught and bootcamp people is often advocated as a way to increase diversity, and a lot of roles don't require a college degree (plus a lot of self taught and bootcamp folks are very good). However, on a large enough dataset, they will be weaker candidates. So if you use those sources to improve diversity, you'll end up with a correlation between diverse hires and more junior devs. Which screws up with every other metrics you'll look at.
I'm not sure you're going to correct the perceived gender/racial/whatever imbalances if you start hiring self-taught programmers and bootcamp grads in droves, just looking at the typical demographics of the people that self-select into those cohorts.
If the bar is the same, but there are fewer female applicants than male applicants, then there will necessarily be fewer females accepted (assuming standards are applied equally).
Until the number of women in our field are on par with men, efforts to create an artificial balance will mean lowering the bar.
its even more troubling then that. When faced with stiffer ability to get interviews (let alone offers), we practice a ton more to increase our hard skills. more practice does tend to lead to being more proficient at something. my female friends generally got 5-10 interviews a week. me, not so much
They wanted more women. Dropped the hiring bar for female devs. Got more mediocre devs. Those got bad reviews. Then they blamed it on sexism, which reduced the women willing to go into tech even more.
-- edit --
I don't know if it went that way or not, it's just an explanation on how their problem would map to the cobra effect.
I believe they meant this in the context of popular gender gap solutions.
More specifically, the situation where pressure is put on teams to weigh gender as more important than skill in the hiring process to bring teams to gender-parity. The result of this is female employees submitting worse code. This isn't because women are inherently any worse at writing software than men are, it's simply that the employer has a smaller pool of talent to pick from.
> Historian Michael Vann argues that the cobra example from British India cannot be proven, but that the rats in Vietnam case can be proven, so the term could be changed to the "rat effect".
Can't read the linked article. Did the study look further into the gender bias assumptions beyond the simple female vs male statistic (e.g by anonymizing code submissions)?
Exactly. Maybe the code needed to be rejected. Many developers have no problem pointing out issues in code. Anyone's code. You can see it daily on stackoverflow. Just because it happens to be a female from time to time makes it newsworthy.
It means it's worth looking into... which is what they're doing. They also discovered the code is more often from more junior developers, which leads to an investigation of whether or not women are having a harder time being promoted or are leaving the company more often.
> Facebook, alarmed by this data, commissioned a second study by Jay Parikh, its head of infrastructure, to investigate any potential issues. Parikh’s findings suggested that the code rejections were due to engineering rank, not gender. However, Facebook employees now speculate that Parikh’s findings mean female engineers might not be rising in the ranks as fast as male counterparts who joined the company at the same time, or perhaps that female engineers are leaving the company more often before being promoted.
I'd love to see a distribution of level, age and years in industry, salary by age. If Fb has put more effort into hiring female engineers directly from college, then they will probably skew junior. The answer is probably not that simple and as you mentioned, all women are probably being challenged more than their male counterparts.
The general tone of this recent article on HN [1] I think was quite negative towards the OP, more so than the median. I have seen similar behavior in code and architecture reviews, interviews, etc for women at companies I have worked. Everything literally turns into a qual. Everything they do is judged to much higher standard. Or even non-sensical standard. It is still the case, that when I come across a woman in dev, she is most likely a freak'n bad ass and has a really thick skin. To me, this shows that we have not yet achieved equality, otherwise there would be more _average_ programmers who are women.
There were two studies, one that was unofficial (just an engineer looking at commits) that found the quoted result. It looked at gender adjusted by tenure. This study didn't have the resources to do that.
Then there was a second cover-your-ass study by Facebook HR that found that if you adjust gender by seniority/rank the bias went away. This raised the question of why female engineers were at lower ranks after the same amount of time on staff. It might be explained by prior work experience, or starting rank, but it might not be. From the articles on this it looks like they (fb) don't actually want to answer this question.
I guarantee you that no one in Facebook feels comfortable suggesting that they investigate another possible (not likely, but possible) cause: the code is just worse, on average.
And there could be many reasons for that, beyond just the simplistic "girls are bad at code." Maybe Facebook has a lower bar for hiring female engineers due to affirmative action.
No, the issue with the article is that its author didn't take into account (as someone claiming to be from Facebook said in a different thread) "skill level" when making their comparisons. (This claim aligns with the statement in the blockquote at the bottom of the article.)
Maybe on average women get less code into the code base, but it's unfair to everyone if you don't remove the influence that seniority has on your dataset before pointing at gender.
Women face many problems in this industry as it is. If I could, I'd ask The Verge not to make up any new ones.
That part doesn't make much sense either because gender representation isn't a legitimate concept. Individuals have different strengths which may help them succeed if their pursuits use those strengths. This results in uneven gender distributions within industries and there's nothing wrong with that.
Unless you think there is some sort of overt discrimination happening that's stopping females becoming senior engineers, in which case you'll need to define it.
Can you provide evidence for the specific claim that women, in particular, are statistically-significantly less strong at the software engineering industry, in particular?
(I buy the argument that there are statistically-significant differences in some strengths that apply in some industries, yes. I'm curious to see how this applies to software engineering in particular.)
I don't recall the exact source but I believe that overall, women score lower in tests for mathematical and logical aptitude, and higher in tests for verbal and social aptitude.
This correlates with the low numbers of women that choose to enter the industry, despite increasing incentives. It also correlates with female domination of fields such as nursing.
OK. I'd argue that software engineering and especially senior software engineering (architecture, designing APIs and interfaces, reading and writing documentation, team management) is better correlated with verbal and social aptitude than with mathematical and logical aptitude. Being able to get a tricky sorting algorithm right is the sort of thing a junior engineer is more likely to do than a senior engineer; when the senior engineer does it, they also tend to need to explain to others why the algorithm is right.
Of course this is just my belief, as a member of the industry, and not a scientifically-researched conclusion.
So, again, I'm curious to see evidence of why women's aptitude is poorly correlated with the specific industry of software engineering. (I would even buy, based on your argument, that women's aptitude is poorly correlated with academic computer science, but that's a very different field.)
It's true communication skills are more important once the senior level is reached. Perhaps the work involved in moving through the junior ranks is not appealing.
It's clear women on the whole are not drawn to STEM. It's not clear why but there does appear to be a biological component. It's certainly not primarily socially constructed.
That's my observation anyway. I haven't studied the issue in great detail.
No one should be driven toward any field they aren't interested in, nor should they be actively discouraged from anything they are interested in.
Today, different families and schools push varying expectations onto children. This cultural aspect is layered on top of biological differences in body and intelligence.
If we remove cultural impediments and allow biology to express itself more fully, the uneven distribution of gender in different fields increases. In other words, the relative evenness of the distribution we currently have is socially constructed.
So I'm not convinced by the argument that women (and indeed many men) are dissuaded because they hold negative stereotypes of people and working culture in STEM fields. However, I'm willing to keep an open mind on this, so if this is true then more effort should be made to welcome in talented outsiders.
> It's clear women on the whole are not drawn to STEM. It's not clear why but there does appear to be a biological component. It's certainly not primarily socially constructed.
I don't think I believe any of these three claims.
For these I don't think asking for supporting evidence would help - these aren't claims where you can easily point to a research paper, but these are interpretations of a large amount of data (some rigorous, some anecdotal).
But it's interesting that we have such different interpretations, worldviews if you will, of the data we've seen. I think women are drawn just fine to STEM as a whole (if by "drawn" you mean "innate interest"), I think they're somewhat structurally discouraged by socially-constructed barriers to STEM as whole and, importantly, quite a bit more discouraged to software engineering in particular. And I don't think there's anything biological about it (unless you count the fact that many people with wombs plan to bear children and nobody without wombs does, and long parental leave is a serious career hit, but I'd still argue that 90% of that is a socially-constructed barrier and not a biological one).
I doubt that hyperdunc was directly referring to the idea that:
> women, in particular, are statistically-significantly less strong at the software engineering industry
I think that what was they indicated was that because individual strengths may vary, because there likely isn't a strong balance of genders in the workplace. It would be obvious that due to different strengths of individuals that there would could be an apparent bias.
Let's assume for a moment, that engineering requires intelligence, and to be a top engineer you need to be in the top X% percentile of the population.
As it turns out the distribution of IQ in men and women is different[0][1][2][3], with men having more variation (more extremely dumb men and more extremely smart men), and an ever so slightly higher median (men have slightly higher IQ on average). These small differences lead to huge difference in representation of exceptionally smart people, because of how normal distribution works. These differences manifest themselves in tasks requiring high intelligence, such as chess, go, engineering and software development.
I say this as a scientific statement, with zero sexism implied. I believe that people should not be afraid to state scientific truths, no matter how controversial they may seem.
As it turns out women have barely had any time at all being afforded the same privileges as men[0], and that's just the United States, there remain cultures that are extremely oppressive to women. It also turns out we still tell women from a young age that "they can't".[1] It also turns out there are a ton of biases pushing women out of STEM[2][3][4][5].
So when you cite evidence that says men are smarter than women conducted by men standing on the shoulders of a society built for men, you can't be shocked when people question it.
I understand that there are biases against women in our society, and I don't argue with that.
I argue that there are biological differences between men and women, and these differences are causing different representation in STEM fields.
I would also say that today, we're living in the most equal opportunity society than ever before. Let the free market sort itself out. If you try to artificially increase the proportion of women in STEM fields, you will decrease the quality of engineers. I'm sorry, but that's how it is.
> So when you cite evidence that says men are smarter than women conducted by men standing on the shoulders of a society built for men, you can't be shocked when people question it.
So, did they use a flawed methodology? Were these studies sexist? Could you point in which way these studies are sexist? Do you disagree that males often have higher variance in different traits in many species? Do you disagree that men have higher variance in IQ?
> I argue that there are biological differences between men and women, and these differences are causing different representation in STEM fields.
I argue that you don't actually have any reason to believe that the differences are biological instead of social. Certainly, while extremely interesting (I mean that. Not sarcasm.), none of your links demonstrate it. Your links claim to show a difference, but they do not claim to explain the cause of said difference.
> I would also say that today, we're living in the most equal opportunity society than ever before.
You could say all kinds of things and more. But, and I'm not agreeing here that it is actually true, because I'm not fully convinced that it is, even if it _is_ true, being better-than isn't the same as being good.
> Let the free market sort itself out.
Only a properly regulated market ever sorts itself out. Otherwise you end up with natural monopolies, because barriers to entry are historically compounded. This has always been true of marketplaces.
> So, did they use a flawed methodology? Were these studies sexist?
Well, one flaw is that your conclusions don't follow from the studies.
> I argue that you don't actually have any reason to believe that the differences are biological instead of social.
I could provide studies on how IQ development is set by genetics and very early childhood. So yes, I do have a reason to believe IQ is biological.
> Only a properly regulated market ever sorts itself out.
This is true, but "we must hire women otherwise people think we're sexist" is not a properly regulated market.
> Well, one flaw is that your conclusions don't follow from the studies.
But my conclusions do follow from studies. If you take top 2% of people by IQ from a random population sample, you expect to have more men. The same would be true if you took the bottom 2%, but that doesn't interest anyone.
Well I could provide a tortoise that speaks seven different languages. Saying what you could provide is fairly bad form.
> and very early childhood.
Wait, how early? Which part of childhood is the genetic part? Heck, what about the potential for non-uniform distribution of teratogens? Has that even been studied?
> but "we must hire women otherwise people think we're sexist" is not a properly regulated market
"We should prefer to choose an equally qualified woman because not only is she presently equally qualified, but she has achieved being equally qualified in an environment that in-many-small-ways-collectively-and-constantly tries to prevent it" is, though. There is no shortage of well-qualified individuals in the world. And if you don't agree, then we must first begin another conversation on what exactly you think qualifies someone to develop software.
> But my conclusions do follow from studies.
They certainly don't follow from the ones you linked, even though you said they would. I know that because I read the studies you linked. So maybe these other ones also don't support your conclusions any better?
> If you take top 2% of people by IQ from a random population sample, you expect to have more men.
Maybe. Now tell me the part of the study that says why. And further tell me the part that indicates a positive correlation with success in computer software production. And then tell me the part that indicates positive correlation with representation in the field. And then tell me the part that indicates...heck, maybe that indicates that g is even a useful measurement to begin with.
Because you said, and I quote, "[biological] differences are causing different representation in STEM fields", which is an unsupported conclusion.
Thank you! I think this analysis is simplistic and has a lot of room for questions, but it is an analysis and it is data-supported, and I respect that. (And I don't think that citing these facts is sexist in any way.)
One line of argument occurs to me: my understanding is that the breakdown of incoming freshmen at top colleges (by which I mean the sorts of colleges Facebook tries very hard to recruit from) is very gender-balanced, much more so than would be implied by the normal distributions you cite. This raises two questions:
1. Are these colleges selecting students for some reason other than general intelligence (whether this is "political correctness" or "intelligence isn't everything" is irrelevant to this question), and if so, can we measure the IQ of men and women at these colleges and find a significant difference? Even if men and women as a whole have different IQ distributions, I would initially assume men and women at top colleges to have comparable IQ distributions (because that's what college admissions should select for), but I can believe that this is not true.
2. How much is raw general intelligence useful for software engineering compared to the pedigree of having attended one of these top schools? I would assume that being in the 95th percentile of intelligence and having a degree from one of these schools is more likely to lead to success than being in the 98th percentile and not, but, again, I can very easily be convinced otherwise.
> 1. Are these colleges selecting students for some reason other than general intelligence, and if so, can we measure the IQ of men and women at these colleges and find a significant difference?
I'm open to being wrong, but my prediction based on the data I have is either a) the measured IQ distribution is going to be different, with equal representation of sexes; b) the representation of sexes is going to be different with equal IQ distribution;
> 2. How much is raw general intelligence useful for software engineering compared to the pedigree of having attended one of these top schools?
That's a really good question, and the answer again is that I don't know. I don't even know how to measure this.
As an anecdote, I didn't attend a top CS college, but I'm a high-ish ranked software developer.
I'd think that it would be fairly easy to measure, honestly: do a longitudinal study of new hires at several top companies, and come back 5 years later and see who's still in the industry and what their level of promotion is if so. You know both their educational history, their demographics, and their standardized test scores (a close-enough proxy for g), and you can see how well they predict success.
There will be biases (in the scientific sense) in the data because of how the interview process selects for successful candidates, which is going to be correlated with these three and probably many other things. But I think the study would still be sound.
(Also, it occurs to me that I've been making about one assumption that is likely to be untrue: that people who do better at top software engineering companies are actually better software engineers. It is entirely possible that software engineering companies are systematically failing to hire and promote the best software engineers, because their interview and review processes are broken. In fact, that is what the pervasive-sexism-in-tech hypothesis implies: that individuals are so sexist that they would prefer to hire a man over a more qualified women, possibly because they're subconsciously failing to realize that the man is actually less qualified. If we want to test this hypothesis seriously, we need to separate "actually a good dev" and "recognized as a good dev". I could easily believe that being in the top percentile of g is correlated with hiring and promotion, because it's easy to measure, but less tightly correlated with actually being good at your job.)
This argument founders against a pretty simple rebuttal. Unless you believe software development of the variety practiced at places like Facebook is so intellectually challenging that its demands outstrip molecular biology, mathematics, and indeed all of science save the fields of physics, computer science, and engineering --- despite the fact that virtually all of science now requires deep facility with computers! --- then that argument has to account for the fact that every one of those fields has much better parity between men and women than software development.
Of course we all here understand that 80% of software development as practiced commercially is not especially intellectually demanding, being as it is essentially a series of variations of transforming database rows into text markup and back again.
> Unless you think there is some sort of overt discrimination happening that's stopping females becoming senior engineers, in which case you'll need to define it.
I think it's pretty likely that the environmental conditions that hinder women from ever becoming engineers in the first place don't all suddenly go away for the ones who manage to do it anyway.
Or are we talking about first principles here? Because I also think that a lot of maybe-not-intentional-but-pretty-much-everywhere-nonetheless discrimination hinders young girls from having early lifestyles that lead to becoming women engineers. I don't really see how one could reasonably argue otherwise. And if discrimination prevents LARGE_% of A, and B requires A, then discrimination has almost certainly also prevented SOME_% of B. That's just basic middle school math.
"maybe-not-intentional-but-pretty-much-everywhere-nonetheless discrimination" is so vague it can be immediately dismissed.
If you're talking about culture or traditional gender roles, then I put forward that these are not primarily a result of discrimination but mostly evolved over quite a long time based on biological differences.
If we are going to deconstruct aspects of culture because we consider them unjust, we'd better be honest about why things are and what we might lose if we try to force certain outcomes.
A recent examination of a Scandinavian country purported to be globally the least discriminatory (Denmark or Norway - I don't recall which) of gender distributions in different fields showed increased unevenness.
In other words, given the agency by their culture, women tend to avoid STEM fields.
I think it's fair to say that everyone should be given the opportunities and challenges they can rise to, but no one should be coerced into any role for the sake of an asinine idea like "gender representation".
As long as Facebook continues to increase the percentage of women it hires, it will have a disproportionately high number of junior female engineers. Intentionally failing to control for that fact is great if you want to profit from the problem of under-representation of women in engineering roles, not so much if you actually want to solve the problem.
> As long as Facebook continues to increase the percentage of women it hires, it will have a disproportionately high number of junior female engineers.
I don't understand why this follows. The standard argument I hear for why women are under-represented (relative to the null hypothesis) is the "pipeline," that at various stages in the process women drop out and it is hard to retain them. In particular, there's the argument that women put their career on hold for their family, which implies that there's a significant number of under-employed senior women.
I'd be very interested to see the number of years of experience vs. level assigned by Facebook, grouped by gender, which would answer whether Facebook's gender balance is really the result of a bunch of 0-years-experience hires. It seems very possible that causality here is the other way around: because women get their code rejected more often, therefore they do not get promoted as much.
The engineer-lead study did control for that fact. The company study used promotion level instead of seniority which one would probably expect to have similar bias impact as code review.
It needs to match the proportion of qualified women to qualified men, because to do otherwise is to either fail to employ qualified people or to employ unqualified people, both of which are bad for Facebook as a company (and arguably for the world at large).
I have not yet seen any concrete evidence to dispute the null hypothesis, that while gender is correlated with a great many things it is not correlated with the ability to be successful at Facebook. Therefore I accept the null hypothesis and think the right number is ~50%. I'm open to believing that the number should be different, possibly very different, given evidence.
* the data itself hasn't been released so no one outside of Facebook has in any way analyzed it, we're relying on their analysis and reportage of that analysis to be correct
* the rejection rate is not just higher for female engineers, it's a lot higher: 35%
* Facebook is confident that the effect is not due to gender bias (scrutinizing female workers' code more closely) but that it's due to a confounding effect of engineering rank (females tend to be lower rank, and lower ranks tend to have more rejections through some mechanism)
In other words, we definitely need more study in the area before I would be confident drawing any important conclusions. It is a fascinating step forward though.
The study that found the 35% controlled for seniority (at Facebook). It would be most correct to use overall experience including prior to FB, but it is probably less correct to use promotion level when promotion decisions would very likely suffer from the same biases.
Creating a throwaway since I've taken part of this internal discussion at Facebook. Once normalized for skill level (i.e. junior engineer, engineer, senior engineer) the level of rejection between genders is the same.
They aren't for an individual, they are for a data set.
Meaning that more senior people have a higher skill in the aggregate, and given the context they're talking about at least dozens of data points. One exception doesn't discount the trend.
The study that showed the bias accounted for seniority at Facebook (though not previous experience). That seems more valid overall than "rank" which, if these biases do exist, one would expect to also be influenced by them.
And can you share insights into what percentage of female engineers there see promotion (as compared to males), and if it's at the same velocity? Because, you know... spin.
Does this mean employees eventually shown more details? The WSJ article I read[1] cast both analyses as flawed in that the employee's study may not have accounted for an engineer’s previous experience before joining FB, while the official one apparently failed to offer enough data to conclusively rule out gender bias and employees' request for more detailed data was denied.
If anything, they should check the rejected code and verify that it was not worthy of rejection, that is, it really was likely the gender of the submitter that caused the rejection.
If higher rank of coder would make rejection less likely, it would also not bear well on the code reviews.
I've heard that the opposite of this suggested bias ought to be consciously corrected.
If the stereotype for a senior engineer is a balding white male with glasses, and you have two prospective engineers with otherwise identical resumes and work histories: one balding white male with glasses and even a pocket protector on his white shirt, and one black woman with a wild T-shirt, which would you hire? The answer I'd heard before was that you should hire the non-conformist, because they've risen to identical positions and accomplishments despite the bias against them at each step, while the stereotypical engineer was probably assumed to belong in the positions they attained. So you should choose the stuttering surgeon and the messy lawyer when given the opportunity.
(No idea where I heard it or what the phenomenon is called, though - anyone remember this?)
With someone like Sheryl Sandberg heading up Facebook's operations, it would be hard to believe that there weren't mandates to achieve gender parity starting from the top.
That Facebook hasn't achieved gender parity yet would suggest to me that the world hasn't yet caught up with supplying the demand for female employees at various levels of the corporate ladder without it impacting their business operations.
I'm very torn on these kinds of studies. In my field we see a lot of studies looking at gender and racial disparities in healthcare offered (e.g. what kinds of _______ therapy are prescribed to various racial groups). While such disparities are probably important in some cases, the "why" is generally not addressed in the study outcome, and rather the results are open to all sorts of interpretations based on our biases.
And then of course any disparities, properly normalized for alternative factors (or not), have to be corrected. Very messy science. It's also easy and politically expedient science - these days we're obsessed with inequality, and if you just comb around you're bound to find some statistically significant disparities and that gets everyone's attention. If researchers were trying to investigate population-level thoughtcrime, this is what it might look like.
This is not a throwaway account. Anyone can look at both my comment history and profile to determine identity.
You, on the other hand, represent yourself under the guise of anonymity, yet with purported knowledge of Facebook internal processes and discussions. The ramification is that, we as a group, must now take the irrationality of "your" identity and match it up with a rational "truth" presented which is "the level of rejection between genders is the same" (at Facebook).
The act of anonymity is likely rationalized as required to avoid judgement. Alternately, the act may be rationalized as an act by the company to avoid judgement, which means you aren't you, but someone posting on behalf of the company. Third, you may be an irrational actor who is posting this to troll us, which is to say an irrational desire to spread irrationality further into our group.
In all cases, this comment is highly irrational. If it's a troll, it's an objective of the troll to seek to spread further irrationality (additional unknown work). If it's a real human, who works at Facebook, posting this information is irrational because we have no way of knowing the context of the conversation, i.e. why did such a topic come up to begin with, who discussed it, who wrote the tests for it, and who conducted and analyzed the test results? Third and finally, if it is the company itself posting it, why is the company posting it? From an irrational standpoint, keeping this information "unofficial" makes it more likely to spread as a "truth" coming from the company, which means they are acting more like a troll with a purpose. (Which is not to say all trolls do not have purpose.)
If the company was ever to make this a public statement, I'm sure more work, or irrationality would come of it, given the questions I raise above about whether testing for this is done in a logical and fair manner.
Side note: I don't believe we should allow anonymous accounts to post because of the work required to untangle the logic. This is currently not possible to do with the HN software, so it's a moot point, and remains a source of dissonance for us.
Edit: Hello Facebook. Streisand effect, anyone? Getting about 1 negative downvote every 2 minutes. Fascinating.
Please don't go off like this here. Throwaway accounts are standard and Hacker News users are perfectly capable of incorporating that datum into their interpretation. We detached this thread from https://news.ycombinator.com/item?id=14255769 and marked it off-topic.
I'll go off anyway I like. That's what the downvote mechanism is for here, after all. Hell, the main article got flagged and pulled. Something is clearly going on. Read the article. It's rational, mostly.
I will point out that large entities like Facebook may misuse this forum to their liking, and that anonymous posts who directly contradict a posting here, in this group, indicates a motive that benefits the entity doing the posting. It would be different if it weren't about something highly controversial that has no means by which a truth can be determined.
FACT: Facebook will and can do anything they like to their products to increase revenue, products of which billions of people use daily to do things they think are trustworthy, when in fact they are provably not. They can also do whatever it is they are capable of doing as a gorilla entity to silence those of us who question their motives in product development.
In these types of cases, we are actually NOT perfectly capable of "incorporating datum" into our interpretations, whatever the hell that means from your perspective.
Frankly, I find dogmatic replies such as yours highly annoying and boring. We, the members of this group called HN, are mostly culpable for making this fucking mess of this thing we call the Internet, with it's mostly viral models and broken security. We've leveraged capital to "trick" users into using our products via viral growth models, all the while to obtain higher and higher payouts. That I come in here to try to make a small dent in the effort to combat this and am colloquially "smacked down" for it is an indication my gut is right about my assertions. People are greedy and will continue to do so as long as they have the means.
So, again, I will go off how I like, when I like, and those who don't like it can do what they will to silence me. I choose to have free will, and deny those who deny that choice for myself.
> -1 points by kordless 12 minutes ago
BTW, Facebook, actively monitoring this article is an asinine approach to brand management. Get a life.
Sorry, I should've been more specific: baseless accusations of shillage, astroturfing, etc. are off-topic on Hacker News and we ask users to email us at hn@ycombinator.com when they have concerns. It matters a lot to us and we're more than happy to investigate.
> BTW, Facebook, actively monitoring this article is an asinine approach to brand management. Get a life.
This is an example of such a baseless accusation—please stop.
I've seen this before. I've also seen plenty code written from developers at higher levels that really should have been reviewed with far more scrutiny - but because of politics, was practically rubber stamped with hardly a comment. IOW, developers at higher levels don't necessarily write better code, often it is worse.
I think this could be improved by a system of code review anonymization that sends code reviews out - company wide - without identifying author info.
Several other sources for this story were flagged out yesterday, even though this is very clearly a topic of interest on hacker news. This one is also currently showing up as flagged. Curious who doesn't want to grapple with an evidence-based look at bias in a tech leader.
Given that the engineer-lead study found this impact when controlling for seniority at FB while the suit-lead one didn't for rank, it certainly raises the question of why women have a lower rank for a given seniority. Charitably maybe men have more experience coming in, but if that was truly the case they could have used that as their input variable to avoid the obvious follow up concerns.
I think to understand gender discrimination we need to understand what hazing is all about. A basic definition may be "An unpleasant experience to gate access". At first, I thought it was about building bonds but what I believe hazing actually does is solidify the boundary between in-group and out-group via a specific and unique obstacle.
The enphasis of this boundary creates a space for privileged information amongst the network. those internal to the network have access to resources others do not, most often in the form of people and the information in their head. The actual quality of information amongst groups is probably roughly the same, but with in-network information, there is more willingness to be open about this information. This is because information can be infinitely duplicated. Giving out information reduces your competitive advantage just as teaching someone everything you know would as well.
Take away the concept of groups and you remove overt hazing. However, individuals exist along with their gated information. This information includes their deepest secrets and fears, their de-facto model of the organization, as well as the secrets to their strengths. Regarding this last one, if you have a high enough gated privilege, he may just give you a link to the resource he used. If you don't, he may just try and explain it himself. And if you have low privilege, he'll be too busy.
Most organizational skills are NOT generic skills learned in school. They are often tips about organizational management, who to ask about what as well as the hidden mental models that keep the legacy code running.
Take this potential acenario: If a women asks a man how the fuck the shitty legacy code works, there is a threat. The woman who is fresh from school has some ivory tower ideas and the man is threatened because he knew he took some shortcuts. This manifests itself as arrogance when explaining, perhaps even just telling the woman to just figure it out as some sort of hazing test. One she is jaded enough by organizational existence will no longer be a threat.
So there is a constant dynamic of men needing to orove themselves to other men and women. When women are involved, the need is higher, creating more arrogance and a higher power distance. Many woman know this, which is why they are more likely to say: "I heard you were the expert in this area, could you give me a few pointers? I'm SOOO bad at this stuff. " while other men might "What the fuck is this shit?!"
Men basically have 2 choices to extract information. Playing the mentee, or playing a high status individual that the other person has to prove themselves to. Women often only have the first because culturally they are socially punished for aggressive behavior.
One root cultural stem is that a man mist be in some eay superior to a woman to "deserve" her. Gender roles have the man playing the hero. This imperative prevents men seeking help from women but also systematically prevents women from reaching high status without overt and obvious superiority in an aspect.
This last part is very important. The issue is NOT about actual skill. Out society is set up fairly well in that it is still mostly a meritocracy. The best rise to the top, and this includes women.
The issue is that your ability is determined by your access to experience and attitudes. (No longer knowledge, e.g. Blacks being banned from libraries or similar) This can create a headwind or tailwind for you in different directions.
Also, in places where measuring skill is flawed and has huge variance, women often have to doubly prove themselves because their "scores" are artificially deflated while an extroverted tall white males might be inflated. From an organizational standpoint this is in some ways even "good" because the org may bot be sexist, but it sure as hell wants its ICs to fall in line. That entails a manager or leader that can utilize aggression and status.
122 comments
[ 2.9 ms ] story [ 135 ms ] threadThe Verge is using the WSJ as the source.
* facebook got criticism for not hiring enough women as coders.
* facebook hired more women (probably lots of junior coders)
* they do what junior coders do
* someone turned the code review rejections into a spreadsheet and disregarded seniority (experience mainly), and made the issue about gender.
Or, to put it another way: when contributors are anonymous and genderless, work by women is accepted at the same rate as men?
In reality, the pool of female talent is very shallow. Again assuming the talent levels are exactly the same, there's just not enough highly-skilled women to go around, e.g. if you're trying to hire the top 1% of engineers, and you have a pool of 1 million men and 100,000 women, then you're looking at 10000 men and 1000 women as your potential hires. If the top 1000 women are already taken, then you're forced to relax your interview requirements for women. In fact, if the pool is completely free for your usage, but you're looking to hire 10000 men and 10000 women, you'll be hiring the top 1% of male talent and down to the top 10% of female talent, which should lead to a difference in output.
This doesn't mean that "sexism" isn't real or that it's not a problem women face in the workplace. The question however, is how to measure the effectiveness of "anti-sexism" practices employed, if you cannot measure the effect of "sexism" itself, as it is entangled with all the other, confounding factors. You can't optimize performance if you can't measure the effects your solutions have on it.
or maybe multinationals should be honest about hiring juniors, and think that would be good PR for them rather than just let people use their staff as a statistic.
Imagine they did the same thing for university entrance: men need all prerequisite courses and a 3.8gpa, whereas women need only 75% of prereqs and a 2.9gpa. It only makes sense that the women are going to score lower in exams/flunk out, since their initial starting point was, on average, further behind.
There are certainly good and bad developers no matter what gender.
If you have stricter hiring policies for men and e.g. only the top 10% of male candidates land a job while the less strict hiring procedures for females get the top 40% females a job, then you created an imbalance in skills between the genders in your company (and only there), and this is what might have happened here
If the number of qualified female developers is less than the number of qualified male developers, then any company which manages to achieve hiring parity has either lowered the qualifications it demands from women developers, or increased the incentives it offers to them relative to other firms (enabling it to capture a higher share of high-skill female developers). Since the latter approach is obviously discriminatory and quite possibly illegal (not to mention a blatant violation of the principle of equal pay for equal work), the former is more likely.
The result of the former is that the median female developer at that company will have lower skill than the median male developer at the same company, even if women in the population at large have the same or better development skill.
The incentive strategy could enable a firm to secure the best female talent, but besides the ethical and legal implications it would also make things much worse for all the other firms out there: the best female talent being off the market, those firms would be forced to either settle for lesser-skilled women or accept a disproportionate share of male developers, neither of which would be appealing.
Naturally, increasing the supply of women developers to match that of male developers (or reducing the supply of male developers …) would improve the situation, but that's a horse of a different colour.
> Code rejected due to rank
> Women are of lower rank than men
> We need to promote more women faster
Facebook is about to get a taste of their own medicine.
If FB female code is of objectively lower quality promotions should make things worse. If it is not - it will be fixed.
To me the whole thing just raises a bunch of questions that don't have any satisfying answers.
Take this hypotherital:
Bob's factory employs machinists to make parts. Of all machinists, 20% belong to group A and 80% belong to group B. When hiring, Bob's factory selects a number of applicants who pass the test. If we assume that group A and group B have the same average skill level then naturally the employees in the factory will be 20% group A and 80% group B.
If Bob wanted to hire eaqual amounts of Group A and Group B, then he would need to hire as much of group a as possible and as little of group b as possible. If bob could only hire a certain number of group b, then he would naturally only pick the best few.
Thus, artificially eaquallizing the work force in bobs factory must neccecarily lead to a skill gap.
Bob's factory posts machinist ads on Craigslist. CL turns out to send Bob 90% males and 10% females. Bob decides to try to increase the number of females he hires, and advertises on a second site. Turns out this site is viewed by a higher percentage of females, and thus his number of female applicants increases, and the bar for hiring stays the same.
This assumption, on which your argument rests, is invalid.
I'm an engineering college student who's had conversations on this subject with admissions officers at my school. The gender ratio of accepted students is much more balanced than the gender ratio of applicants. Every year, the admissions department fields complaints from upset male applicants who were rejected; they feel that they were done a disservice because "the bar was higher" for them than for the female applicants. They point to the different acceptance rates for men and women as proof of this "lower bar" for women.
In reality, according to my school's admissions officers, the average female applicant is significantly stronger than the average male applicant. There exists, apparently, a huge volume of unqualified male applicants, who inflate that "80%" figure despite having no real chance at earning a position. We're talking SAT scores hundreds of points below the typical admitted student, no actual engineering experience, things like that. (I have heard that large companies like Texas Instruments also have this problem, which is why they place restrictions on which specific degrees applicants can have - to filter out that chaff.)
These men remain in the applicant pool where women would've washed out, the admissions officers at my school hypothesize, because it is easier for women to "give up" on STEM fields at the first sign of trouble due to social / societal pressures. Incompetent men, however, can think they're good at STEM more or less forever.
A more reasonable conclusion to draw, instead of "Facebook's male engineers are more skilled than their female engineers", would be the one that is presented in the article by the people who have the data. Namely, that this statistic arises from the confounding effect of seniority: Facebook's more senior engineers tend to be male, since the drive to hire more women has been more recent, and more senior engineers have code accepted more often.
Hiring more self taught and bootcamp people is often advocated as a way to increase diversity, and a lot of roles don't require a college degree (plus a lot of self taught and bootcamp folks are very good). However, on a large enough dataset, they will be weaker candidates. So if you use those sources to improve diversity, you'll end up with a correlation between diverse hires and more junior devs. Which screws up with every other metrics you'll look at.
Until the number of women in our field are on par with men, efforts to create an artificial balance will mean lowering the bar.
-- edit --
I don't know if it went that way or not, it's just an explanation on how their problem would map to the cobra effect.
More specifically, the situation where pressure is put on teams to weigh gender as more important than skill in the hiring process to bring teams to gender-parity. The result of this is female employees submitting worse code. This isn't because women are inherently any worse at writing software than men are, it's simply that the employer has a smaller pool of talent to pick from.
This guy won't win any prizes for marketing.
The general tone of this recent article on HN [1] I think was quite negative towards the OP, more so than the median. I have seen similar behavior in code and architecture reviews, interviews, etc for women at companies I have worked. Everything literally turns into a qual. Everything they do is judged to much higher standard. Or even non-sensical standard. It is still the case, that when I come across a woman in dev, she is most likely a freak'n bad ass and has a really thick skin. To me, this shows that we have not yet achieved equality, otherwise there would be more _average_ programmers who are women.
[1] https://news.ycombinator.com/item?id=14227892
Then there was a second cover-your-ass study by Facebook HR that found that if you adjust gender by seniority/rank the bias went away. This raised the question of why female engineers were at lower ranks after the same amount of time on staff. It might be explained by prior work experience, or starting rank, but it might not be. From the articles on this it looks like they (fb) don't actually want to answer this question.
And there could be many reasons for that, beyond just the simplistic "girls are bad at code." Maybe Facebook has a lower bar for hiring female engineers due to affirmative action.
Maybe this is a "Facebook rejects..." story. Maybe it's a story about how bad Facebook is.
But if you keep reading past the introduction it sure sounds a lot more like a "male engineers reject..." story.
And then if you keep reading some more it sure sounds like a "code is rejected proportionally to engineer seniority..." story, which is not a story.
And then at the very very end, you finally get to the only part of this terrible article that makes any sense:
> the current representation of senior female engineers both at Facebook and across the industry is nowhere near where it needs to be.
Facebook talking about itself is the only honestly presented thing on the page.
So is the story really that the n is too small for this story?
Maybe on average women get less code into the code base, but it's unfair to everyone if you don't remove the influence that seniority has on your dataset before pointing at gender.
Women face many problems in this industry as it is. If I could, I'd ask The Verge not to make up any new ones.
Unless you think there is some sort of overt discrimination happening that's stopping females becoming senior engineers, in which case you'll need to define it.
(I buy the argument that there are statistically-significant differences in some strengths that apply in some industries, yes. I'm curious to see how this applies to software engineering in particular.)
This correlates with the low numbers of women that choose to enter the industry, despite increasing incentives. It also correlates with female domination of fields such as nursing.
Of course this is just my belief, as a member of the industry, and not a scientifically-researched conclusion.
So, again, I'm curious to see evidence of why women's aptitude is poorly correlated with the specific industry of software engineering. (I would even buy, based on your argument, that women's aptitude is poorly correlated with academic computer science, but that's a very different field.)
It's clear women on the whole are not drawn to STEM. It's not clear why but there does appear to be a biological component. It's certainly not primarily socially constructed.
That's my observation anyway. I haven't studied the issue in great detail.
It's also clear that girls on the whole are not driven toward STEM by their adult mentors, and outside the bubble are actively discouraged from it.
> It's certainly not primarily socially constructed.
Certain, are you? Based on what?
Today, different families and schools push varying expectations onto children. This cultural aspect is layered on top of biological differences in body and intelligence.
If we remove cultural impediments and allow biology to express itself more fully, the uneven distribution of gender in different fields increases. In other words, the relative evenness of the distribution we currently have is socially constructed.
So I'm not convinced by the argument that women (and indeed many men) are dissuaded because they hold negative stereotypes of people and working culture in STEM fields. However, I'm willing to keep an open mind on this, so if this is true then more effort should be made to welcome in talented outsiders.
I don't think I believe any of these three claims.
For these I don't think asking for supporting evidence would help - these aren't claims where you can easily point to a research paper, but these are interpretations of a large amount of data (some rigorous, some anecdotal).
But it's interesting that we have such different interpretations, worldviews if you will, of the data we've seen. I think women are drawn just fine to STEM as a whole (if by "drawn" you mean "innate interest"), I think they're somewhat structurally discouraged by socially-constructed barriers to STEM as whole and, importantly, quite a bit more discouraged to software engineering in particular. And I don't think there's anything biological about it (unless you count the fact that many people with wombs plan to bear children and nobody without wombs does, and long parental leave is a serious career hit, but I'd still argue that 90% of that is a socially-constructed barrier and not a biological one).
> women, in particular, are statistically-significantly less strong at the software engineering industry
I think that what was they indicated was that because individual strengths may vary, because there likely isn't a strong balance of genders in the workplace. It would be obvious that due to different strengths of individuals that there would could be an apparent bias.
As it turns out the distribution of IQ in men and women is different[0][1][2][3], with men having more variation (more extremely dumb men and more extremely smart men), and an ever so slightly higher median (men have slightly higher IQ on average). These small differences lead to huge difference in representation of exceptionally smart people, because of how normal distribution works. These differences manifest themselves in tasks requiring high intelligence, such as chess, go, engineering and software development.
I say this as a scientific statement, with zero sexism implied. I believe that people should not be afraid to state scientific truths, no matter how controversial they may seem.
[0] http://www.sciencedirect.com/science/article/pii/S0160289606...
[1] http://www.sciencedirect.com/science/article/pii/S0191886911...
[2] http://www.sciencedirect.com/science/article/pii/S0160289606...
[3] http://www.sciencedirect.com/science/article/pii/S0160289604...
So when you cite evidence that says men are smarter than women conducted by men standing on the shoulders of a society built for men, you can't be shocked when people question it.
[0]https://en.wikipedia.org/wiki/History_of_women_in_the_United...
[1] http://science.sciencemag.org/content/355/6323/389
[2] http://www.uchastings.edu/news/articles/2015/01/double-jeopa...
[3] http://www.pnas.org/content/111/12/4403.full
[4] http://www.pnas.org/content/109/41/16474.full
I argue that there are biological differences between men and women, and these differences are causing different representation in STEM fields.
I would also say that today, we're living in the most equal opportunity society than ever before. Let the free market sort itself out. If you try to artificially increase the proportion of women in STEM fields, you will decrease the quality of engineers. I'm sorry, but that's how it is.
> So when you cite evidence that says men are smarter than women conducted by men standing on the shoulders of a society built for men, you can't be shocked when people question it.
So, did they use a flawed methodology? Were these studies sexist? Could you point in which way these studies are sexist? Do you disagree that males often have higher variance in different traits in many species? Do you disagree that men have higher variance in IQ?
I argue that you don't actually have any reason to believe that the differences are biological instead of social. Certainly, while extremely interesting (I mean that. Not sarcasm.), none of your links demonstrate it. Your links claim to show a difference, but they do not claim to explain the cause of said difference.
> I would also say that today, we're living in the most equal opportunity society than ever before.
You could say all kinds of things and more. But, and I'm not agreeing here that it is actually true, because I'm not fully convinced that it is, even if it _is_ true, being better-than isn't the same as being good.
> Let the free market sort itself out.
Only a properly regulated market ever sorts itself out. Otherwise you end up with natural monopolies, because barriers to entry are historically compounded. This has always been true of marketplaces.
> So, did they use a flawed methodology? Were these studies sexist?
Well, one flaw is that your conclusions don't follow from the studies.
I could provide studies on how IQ development is set by genetics and very early childhood. So yes, I do have a reason to believe IQ is biological.
> Only a properly regulated market ever sorts itself out.
This is true, but "we must hire women otherwise people think we're sexist" is not a properly regulated market.
> Well, one flaw is that your conclusions don't follow from the studies.
But my conclusions do follow from studies. If you take top 2% of people by IQ from a random population sample, you expect to have more men. The same would be true if you took the bottom 2%, but that doesn't interest anyone.
Well I could provide a tortoise that speaks seven different languages. Saying what you could provide is fairly bad form.
> and very early childhood.
Wait, how early? Which part of childhood is the genetic part? Heck, what about the potential for non-uniform distribution of teratogens? Has that even been studied?
> but "we must hire women otherwise people think we're sexist" is not a properly regulated market
"We should prefer to choose an equally qualified woman because not only is she presently equally qualified, but she has achieved being equally qualified in an environment that in-many-small-ways-collectively-and-constantly tries to prevent it" is, though. There is no shortage of well-qualified individuals in the world. And if you don't agree, then we must first begin another conversation on what exactly you think qualifies someone to develop software.
> But my conclusions do follow from studies.
They certainly don't follow from the ones you linked, even though you said they would. I know that because I read the studies you linked. So maybe these other ones also don't support your conclusions any better?
> If you take top 2% of people by IQ from a random population sample, you expect to have more men.
Maybe. Now tell me the part of the study that says why. And further tell me the part that indicates a positive correlation with success in computer software production. And then tell me the part that indicates positive correlation with representation in the field. And then tell me the part that indicates...heck, maybe that indicates that g is even a useful measurement to begin with.
Because you said, and I quote, "[biological] differences are causing different representation in STEM fields", which is an unsupported conclusion.
One line of argument occurs to me: my understanding is that the breakdown of incoming freshmen at top colleges (by which I mean the sorts of colleges Facebook tries very hard to recruit from) is very gender-balanced, much more so than would be implied by the normal distributions you cite. This raises two questions:
1. Are these colleges selecting students for some reason other than general intelligence (whether this is "political correctness" or "intelligence isn't everything" is irrelevant to this question), and if so, can we measure the IQ of men and women at these colleges and find a significant difference? Even if men and women as a whole have different IQ distributions, I would initially assume men and women at top colleges to have comparable IQ distributions (because that's what college admissions should select for), but I can believe that this is not true.
2. How much is raw general intelligence useful for software engineering compared to the pedigree of having attended one of these top schools? I would assume that being in the 95th percentile of intelligence and having a degree from one of these schools is more likely to lead to success than being in the 98th percentile and not, but, again, I can very easily be convinced otherwise.
I'm open to being wrong, but my prediction based on the data I have is either a) the measured IQ distribution is going to be different, with equal representation of sexes; b) the representation of sexes is going to be different with equal IQ distribution;
> 2. How much is raw general intelligence useful for software engineering compared to the pedigree of having attended one of these top schools?
That's a really good question, and the answer again is that I don't know. I don't even know how to measure this.
As an anecdote, I didn't attend a top CS college, but I'm a high-ish ranked software developer.
There will be biases (in the scientific sense) in the data because of how the interview process selects for successful candidates, which is going to be correlated with these three and probably many other things. But I think the study would still be sound.
(Also, it occurs to me that I've been making about one assumption that is likely to be untrue: that people who do better at top software engineering companies are actually better software engineers. It is entirely possible that software engineering companies are systematically failing to hire and promote the best software engineers, because their interview and review processes are broken. In fact, that is what the pervasive-sexism-in-tech hypothesis implies: that individuals are so sexist that they would prefer to hire a man over a more qualified women, possibly because they're subconsciously failing to realize that the man is actually less qualified. If we want to test this hypothesis seriously, we need to separate "actually a good dev" and "recognized as a good dev". I could easily believe that being in the top percentile of g is correlated with hiring and promotion, because it's easy to measure, but less tightly correlated with actually being good at your job.)
Of course we all here understand that 80% of software development as practiced commercially is not especially intellectually demanding, being as it is essentially a series of variations of transforming database rows into text markup and back again.
I think it's pretty likely that the environmental conditions that hinder women from ever becoming engineers in the first place don't all suddenly go away for the ones who manage to do it anyway.
Or are we talking about first principles here? Because I also think that a lot of maybe-not-intentional-but-pretty-much-everywhere-nonetheless discrimination hinders young girls from having early lifestyles that lead to becoming women engineers. I don't really see how one could reasonably argue otherwise. And if discrimination prevents LARGE_% of A, and B requires A, then discrimination has almost certainly also prevented SOME_% of B. That's just basic middle school math.
If you're talking about culture or traditional gender roles, then I put forward that these are not primarily a result of discrimination but mostly evolved over quite a long time based on biological differences.
If we are going to deconstruct aspects of culture because we consider them unjust, we'd better be honest about why things are and what we might lose if we try to force certain outcomes.
A recent examination of a Scandinavian country purported to be globally the least discriminatory (Denmark or Norway - I don't recall which) of gender distributions in different fields showed increased unevenness.
In other words, given the agency by their culture, women tend to avoid STEM fields.
I think it's fair to say that everyone should be given the opportunities and challenges they can rise to, but no one should be coerced into any role for the sake of an asinine idea like "gender representation".
As long as Facebook continues to increase the percentage of women it hires, it will have a disproportionately high number of junior female engineers. Intentionally failing to control for that fact is great if you want to profit from the problem of under-representation of women in engineering roles, not so much if you actually want to solve the problem.
I don't understand why this follows. The standard argument I hear for why women are under-represented (relative to the null hypothesis) is the "pipeline," that at various stages in the process women drop out and it is hard to retain them. In particular, there's the argument that women put their career on hold for their family, which implies that there's a significant number of under-employed senior women.
I'd be very interested to see the number of years of experience vs. level assigned by Facebook, grouped by gender, which would answer whether Facebook's gender balance is really the result of a bunch of 0-years-experience hires. It seems very possible that causality here is the other way around: because women get their code rejected more often, therefore they do not get promoted as much.
Where does it need to be and why?
I have not yet seen any concrete evidence to dispute the null hypothesis, that while gender is correlated with a great many things it is not correlated with the ability to be successful at Facebook. Therefore I accept the null hypothesis and think the right number is ~50%. I'm open to believing that the number should be different, possibly very different, given evidence.
* the data itself hasn't been released so no one outside of Facebook has in any way analyzed it, we're relying on their analysis and reportage of that analysis to be correct
* the rejection rate is not just higher for female engineers, it's a lot higher: 35%
* Facebook is confident that the effect is not due to gender bias (scrutinizing female workers' code more closely) but that it's due to a confounding effect of engineering rank (females tend to be lower rank, and lower ranks tend to have more rejections through some mechanism)
In other words, we definitely need more study in the area before I would be confident drawing any important conclusions. It is a fascinating step forward though.
It is hardly surprising or novel that junior/less experienced developers will write code that has more problems and thus gets rejected more often.
https://www.google.com/amp/www.foxbusiness.com/features/2017...
Meaning that more senior people have a higher skill in the aggregate, and given the context they're talking about at least dozens of data points. One exception doesn't discount the trend.
https://www.wsj.com/articles/facebooks-female-engineers-clai...
If higher rank of coder would make rejection less likely, it would also not bear well on the code reviews.
If the stereotype for a senior engineer is a balding white male with glasses, and you have two prospective engineers with otherwise identical resumes and work histories: one balding white male with glasses and even a pocket protector on his white shirt, and one black woman with a wild T-shirt, which would you hire? The answer I'd heard before was that you should hire the non-conformist, because they've risen to identical positions and accomplishments despite the bias against them at each step, while the stereotypical engineer was probably assumed to belong in the positions they attained. So you should choose the stuttering surgeon and the messy lawyer when given the opportunity.
(No idea where I heard it or what the phenomenon is called, though - anyone remember this?)
Taleb - Surgeons Should Not Look Like Surgeons
That Facebook hasn't achieved gender parity yet would suggest to me that the world hasn't yet caught up with supplying the demand for female employees at various levels of the corporate ladder without it impacting their business operations.
And then of course any disparities, properly normalized for alternative factors (or not), have to be corrected. Very messy science. It's also easy and politically expedient science - these days we're obsessed with inequality, and if you just comb around you're bound to find some statistically significant disparities and that gets everyone's attention. If researchers were trying to investigate population-level thoughtcrime, this is what it might look like.
You, on the other hand, represent yourself under the guise of anonymity, yet with purported knowledge of Facebook internal processes and discussions. The ramification is that, we as a group, must now take the irrationality of "your" identity and match it up with a rational "truth" presented which is "the level of rejection between genders is the same" (at Facebook).
The act of anonymity is likely rationalized as required to avoid judgement. Alternately, the act may be rationalized as an act by the company to avoid judgement, which means you aren't you, but someone posting on behalf of the company. Third, you may be an irrational actor who is posting this to troll us, which is to say an irrational desire to spread irrationality further into our group.
In all cases, this comment is highly irrational. If it's a troll, it's an objective of the troll to seek to spread further irrationality (additional unknown work). If it's a real human, who works at Facebook, posting this information is irrational because we have no way of knowing the context of the conversation, i.e. why did such a topic come up to begin with, who discussed it, who wrote the tests for it, and who conducted and analyzed the test results? Third and finally, if it is the company itself posting it, why is the company posting it? From an irrational standpoint, keeping this information "unofficial" makes it more likely to spread as a "truth" coming from the company, which means they are acting more like a troll with a purpose. (Which is not to say all trolls do not have purpose.)
If the company was ever to make this a public statement, I'm sure more work, or irrationality would come of it, given the questions I raise above about whether testing for this is done in a logical and fair manner.
Side note: I don't believe we should allow anonymous accounts to post because of the work required to untangle the logic. This is currently not possible to do with the HN software, so it's a moot point, and remains a source of dissonance for us.
Edit: Hello Facebook. Streisand effect, anyone? Getting about 1 negative downvote every 2 minutes. Fascinating.
I will point out that large entities like Facebook may misuse this forum to their liking, and that anonymous posts who directly contradict a posting here, in this group, indicates a motive that benefits the entity doing the posting. It would be different if it weren't about something highly controversial that has no means by which a truth can be determined.
FACT: Facebook will and can do anything they like to their products to increase revenue, products of which billions of people use daily to do things they think are trustworthy, when in fact they are provably not. They can also do whatever it is they are capable of doing as a gorilla entity to silence those of us who question their motives in product development.
In these types of cases, we are actually NOT perfectly capable of "incorporating datum" into our interpretations, whatever the hell that means from your perspective.
Frankly, I find dogmatic replies such as yours highly annoying and boring. We, the members of this group called HN, are mostly culpable for making this fucking mess of this thing we call the Internet, with it's mostly viral models and broken security. We've leveraged capital to "trick" users into using our products via viral growth models, all the while to obtain higher and higher payouts. That I come in here to try to make a small dent in the effort to combat this and am colloquially "smacked down" for it is an indication my gut is right about my assertions. People are greedy and will continue to do so as long as they have the means.
So, again, I will go off how I like, when I like, and those who don't like it can do what they will to silence me. I choose to have free will, and deny those who deny that choice for myself.
> -1 points by kordless 12 minutes ago
BTW, Facebook, actively monitoring this article is an asinine approach to brand management. Get a life.
> BTW, Facebook, actively monitoring this article is an asinine approach to brand management. Get a life.
This is an example of such a baseless accusation—please stop.
I think this could be improved by a system of code review anonymization that sends code reviews out - company wide - without identifying author info.
Given that the engineer-lead study found this impact when controlling for seniority at FB while the suit-lead one didn't for rank, it certainly raises the question of why women have a lower rank for a given seniority. Charitably maybe men have more experience coming in, but if that was truly the case they could have used that as their input variable to avoid the obvious follow up concerns.
The enphasis of this boundary creates a space for privileged information amongst the network. those internal to the network have access to resources others do not, most often in the form of people and the information in their head. The actual quality of information amongst groups is probably roughly the same, but with in-network information, there is more willingness to be open about this information. This is because information can be infinitely duplicated. Giving out information reduces your competitive advantage just as teaching someone everything you know would as well.
Take away the concept of groups and you remove overt hazing. However, individuals exist along with their gated information. This information includes their deepest secrets and fears, their de-facto model of the organization, as well as the secrets to their strengths. Regarding this last one, if you have a high enough gated privilege, he may just give you a link to the resource he used. If you don't, he may just try and explain it himself. And if you have low privilege, he'll be too busy.
Most organizational skills are NOT generic skills learned in school. They are often tips about organizational management, who to ask about what as well as the hidden mental models that keep the legacy code running.
Take this potential acenario: If a women asks a man how the fuck the shitty legacy code works, there is a threat. The woman who is fresh from school has some ivory tower ideas and the man is threatened because he knew he took some shortcuts. This manifests itself as arrogance when explaining, perhaps even just telling the woman to just figure it out as some sort of hazing test. One she is jaded enough by organizational existence will no longer be a threat.
So there is a constant dynamic of men needing to orove themselves to other men and women. When women are involved, the need is higher, creating more arrogance and a higher power distance. Many woman know this, which is why they are more likely to say: "I heard you were the expert in this area, could you give me a few pointers? I'm SOOO bad at this stuff. " while other men might "What the fuck is this shit?!"
Men basically have 2 choices to extract information. Playing the mentee, or playing a high status individual that the other person has to prove themselves to. Women often only have the first because culturally they are socially punished for aggressive behavior.
One root cultural stem is that a man mist be in some eay superior to a woman to "deserve" her. Gender roles have the man playing the hero. This imperative prevents men seeking help from women but also systematically prevents women from reaching high status without overt and obvious superiority in an aspect.
This last part is very important. The issue is NOT about actual skill. Out society is set up fairly well in that it is still mostly a meritocracy. The best rise to the top, and this includes women.
The issue is that your ability is determined by your access to experience and attitudes. (No longer knowledge, e.g. Blacks being banned from libraries or similar) This can create a headwind or tailwind for you in different directions.
Also, in places where measuring skill is flawed and has huge variance, women often have to doubly prove themselves because their "scores" are artificially deflated while an extroverted tall white males might be inflated. From an organizational standpoint this is in some ways even "good" because the org may bot be sexist, but it sure as hell wants its ICs to fall in line. That entails a manager or leader that can utilize aggression and status.
So, to sum up, culturally women are often ...