Slashdot Mirror
Facebook Rejects Female Engineers' Code More Often Than Male Counterparts, Analysis Finds (theverge.com)
According to The Wall Street Journal, female engineers who work at Facebook may face gender bias that prevents their code from being accepted at the same rate as male counterparts. "For Facebook, these revelations call into question the company's ongoing diversity efforts and its goal to build overarching online systems for people around the globe," reports The Verge. "The company's workforce is just 33 percent female, with women holding just 17 percent of technical roles and 27 percent of leadership positions." From the report: The findings come in two parts. An initial study by a former employee found that code written by female engineers was less likely to make it through Facebook's internal peer review system. This seemed to suggest that a female engineer's work was more heavily scrutinized. 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. Either possibility could result in the 35 percent higher code rejection rate for female engineers. When contacted by The Wall Street Journal, Facebook called the initial study "incomplete and inaccurate" and based on "incomplete data," but did not shy away from confirming Parikh's separate findings.
The problem for many authors, not just coders, is that both women and men rank them more harshly.
No matter how you slice it.
I used to experiment with this by swapping names on code submissions with female colleagues and watching code suddenly be treated differently.
The cutting critiques were the worst parts.
Is it fair?
No.
Does it happen?
Yes.
My advice is find some token replacement method for code submissions so that evaluators can't extrapolate gender.
-- Tigger warning: This post may contain tiggers! --
https://pmd.github.io/
Use an automated code review to baseline. Compilers care nothing about genitalia.
Maybe they were hired for their looks not their leet coding skills. Not like that would EVER happen.
This. Exactly this. Posting as AC for obvious reasons. I've worked in the Bay Area for over 12 years for different companies, and I've seen it many times. Mediocre female engineers (software, electrical or network) get hired based on diversity quotas and in many cases their looks as well. Managers find out after about a year or so that their hires weren't such a good fit to the team as they hoped, and they end up promoting the mediocre engineers to poor managers. Now the good engineers report to the mediocre ones and get frustrated, and eventually leave the company. And of course, the /. feminazi crew will downmod this, but the truth has to be said.
/. or CNN of course.
I also need to add that I have seen many, many good female engineers. It's just that the ones that get hired based on their looks or for other reasons than their engineering qualities are usually not a good fit for the company. The good ones are often very much appreciated, and I've seen many occasions where they are paid the most of the entire team. But you never hear that on
I just hope code review standards are not lowered in order to avoid emotional trauma
On the other hand, I'm sure you'd agree that raising the standards for men until the reject rate is equal for both genders will result in even better code quality overall right? See... we can frame it either way.
The more important issue is whether the review standards are currently being applied equally or not. They should be.
It -might- be that the pool of women isn't as good at coding as the pool of men... especially if they have been 'stuffing' the ranks with diversity hires. In which case the best solution is additional training for the people who need it of either gender; and culling those that can't be trained.. of either gender.
Or it might be that the people doing code reviews are being harder on code submitted by women for any number of reasons.
Or it might be both. It doesn't have to be just one or something else... maybe there's a woman doing code reviews who feels threatened by the other women so she's extra tough on them and rejects everything they commit... who knows?
For all I know, the women's solutions are better because of this, and the stats brought to light here are because men can't see that - because the thinking isn't the same.
Real life example of this happening: Female police officers.
For many years the most important criteria for evaluating a police officer was "number of arrests". Women just didn't measure up, and performed poorly.
Then "community policing" was adopted, and people realized that "making arrests" was actually a dumb way to measure police performance. Far more important was preventing the crimes from happening in the first place, and defusing potentially violent situations rather than escalating them. By these measures, women are, on average, better police officers than men.
The key to get more women into tech is start introducing programming in elementary school and hook them on coding before they get sidetracked with stupid tween shit.
Nope. As someone who coaches a programming/robotics program at an elementary school for 4th, 5th and 6th graders, I can say that it is very difficult to get girls interested even in elementary school. The few that participate are mostly there because their parents forced them.
I have tried hard to get more girls to sign up. I recruited a techno-mom to be an assistant coach and role model. We let them form an all-girl team (which they prefer). We tried cooperation oriented programming tasks, rather than competitions. We tried other girl-oriented stuff like 3D-printing dollhouse furniture. None of that made a difference. Half of them quit when there was a time conflict with the school play rehearsals. Zero boys have dropped out.
I feel very frustrated. If anyone has any ideas on how to get girls interested in tech, I would love to hear about it.
It would be interesting to study the patterns of the differences. There could indeed be a "style mismatch" between the way females tend to code versus males. Everyone has their personal preferences and as a reviewer, if a specimen doesn't match close enough to their preferences, they are more likely to reject it.
Anyhow, the devil's in the details, and we don't have those. Factors to be checked include things like duration at the company, education level, age, total coding experience, familiarity between the inspector and inspectee (including does inspector know the gender), reason(s) for the rejection, etc.
Table-ized A.I.
A few years ago my nieces were really interested in this: http://web.stanford.edu/class/...
We did the image exercises. I started off doing most of the typing with their input, then they did some on their own. The cool thing with this library is you can go way off on a tangent. We made stripes of across some of the images for instance.
For what it's worth, my wife participated in an outreach program through her work to expose kids to programming. They sent a team to a school and each employee took a group of kids and did a different project. I suggested this one, which my wife customized a bit. It was by far the most popular project with the kids (I think they were 6th and 7th graders). Graphics are cool, especially the green-screen exercises.
Programming is a perfect, if often maddening, example of the creative problem-solving technique. The question, boiled down to its very essentials, is always this: how can I make this (expletive inserted or deleted as you wish) machine do what I want it to?
There are other, lesser, considerations, of course. The system will invariably impose all manner of restrictions: the memory space isn’t big enough, the language is limited, input-output devices won’t do what you would like them to. These become challenges, albeit frustrating ones.
Most of the restrictions, however, are in your own mind. It is very difficult to resist the trap of continuing to look for a solution in the wrong place. If an approach to a particular problem yields no results, the usual response is to work even harder on that same approach. This unfortunate fact seems to be true in all human endeavors, from child-raising to international relations. You, as programmer, must know about when something should be expected to pay off, and, if the time passes and there is no payoff, you must stop doing whatever you’re doing and try something else. You must look around at this point, try to see what you’re really doing, and do something else.
Often the problem is not where you think it is. You might stare at a formula or a loop for hours, wondering what’s wrong with it, when the problem is a duplicate variable or a problem with initialization much earlier in the program. It’s not easy to change directions. You will be convinced that you’re on the right track, that you’ve just about got it, that only a little more effort will solve the problem.
Forget it. It’s the will-o-the-wisp.
Now then, if you have made the necessary change in direction, you may find later that the program still doesn’t work because you have some vestiges left of your old way of thinking. Somewhere buried in the program is a critical juncture which depends on the old way. It will be much easier to find if you know it’s there It is often useful to stop completely and do absolutely no work on the project for a few days.
Now, it is important that you have stopped not out of laziness but to give your subconscious mind a chance to work. What usually happens is that the solution comes to you at some totally unexpected moment, usually while you are relaxed and not thinking of the problem at all. The butterfly lights on your shoulder when you stop chasing it.
There is also a tendency in this game to go for the needlessly complex solution. Computers can do such elaborate and complex calculations and handle such intricate routines in so short a time that one is tempted to keep adding details until finally there is no possible way to understand what has been done. It is too easy to overlook some critical detail if the path through the program is too difficult. It is something like sending a small child to the store for a loaf of bread. Your chances of success are greater if the route to the store is fairly short, passes no playgrounds or amusement arcades, and if you give the child the exact amount of money, than if you send the child on a five-mile jaunt through city streets with ten dollars in his pocket. The great ideas are all simple ones.
How can you keep from falling into this trap?
Awareness is part of the answer. Tell yourself that the danger is out there, and constantly ask yourself if there should be a simpler, easier way to do it. You might enjoy looking at all those lines of wonderfully obtuse, arcane code, loaded with complex algorithms and advanced programming features, all nestled among countless pairs of parentheses (these are especially good if four or five right-hand parentheses all come together at the end), but is it all really necessary?
The artist is ruthless: the most valuable piece of equipment in any art studio is the trashcan. Never mind that you spent hours working up some elaborate procedure. Never mind that it has become your baby, that you feel lik
I'm can't help you, but I'm just as frustrated.
My 12 years old niece loves, loves, Kerbal Space Program and plays it constantly. I introduced it to her when it first came out in 2015, when she was 10, and she's been having fun launching rockets, failing, redesigning, and trying again - she loves it. But she is absolutely terrified of her friends finding out, so much so that she'll play dumb about the game whenever her friends are around and a family member happens to mention it. I can't stress enough how scared she is for her friends to find out she likes a game about rocket science.
When I ask her why she says, pretty plainly, "Girls aren't supposed to be smart, no one likes smart girls - please don't tell anyone I like this game, I don't want people thinking I'm smart because then I won't have any friends!" She's 12! She's deliberately going out of her way to hinder herself and limit her choices in life because society has browbeat into her that she's not supposed to be smart or take an interest in science. The Ontario Science Centre is one of her favourite places to go, she begs me to take her whenever I can - and she totally gets the science, especially the Astronomy section, and could even be a scientist someday if she really wanted to - but she never tells her friends that she loves going there lest they think she's "smart" and shun her.
I've been trying to convince her otherwise, but it just doesn't work - peer pressure when you're a child is a horrific thing, it really messes you up - she's been telling me that girls aren't supposed to be smart for years now - and I think as adults many of us forget just how important social acceptance is to kids - how important it was to us when we were kids - and how that shapes one's perceptions far into the future.
Retention of young female lawyers is a major problem in London's City law firms - where I work - they get fed up after a few years and find something else. Quite why this is so is a topic of serious investigation: these people are expensive to recruit and train and something is pissing them off.
I note that you are drawing conclusions before the junior bias question is answered...?