How my company is lowering the bar to increase diversity
medium.com
medium.com
This does not pass the smell test. Sure, a certain type of "diversity activist" might say something like that, but for a company of any sort of quality to just accept it and implement it, that's unrealistic.
More realistically, what they may have implemented is something like accepting that people without CS degrees from top universities may still be perfectly capable software developers, which is true (being one, I'd like to think so, at least), but if that's the case, the article is grossly misrepresenting it.
Unfortunately it's hard to find sources for this except for internal videos.
A quote from one of the trainers that gets at the same idea:
"In sourcing diverse candidates, it is imperative to avoid criteria that are inherently biased, like using school selectivity and previous company as a proxy for performance or as a signal that someone is a strong candidate.
Instead of focusing on previous companies or schools, which limits the talent pool, you should focus on relevant skills. For example, if Google is typically used as a signal that someone is a good candidate, push yourselves and your clients to articulate why that's the case. Is it the experience working at a large company that's useful?"
Here's an attempt at a charitable characterization: Lots of heuristics for quickly evaluating programming skill may inadvertently toss out qualified applicants or give an undeserved boost to unqualified applicants who tick off superficial traits. Instead of leaning so heavily on heuristics that might disproportionately favor white men, let's try to create hiring systems that fairly evaluate people on their merits and give people full opportunity to demonstrate their talents. We should also try to make people aware that humans can make unconsciously-biased decisions based on some deeply-hardwired evolutionary heuristics around race and gender, so they have the awareness and mental tools to recognize when unconscious bias is influencing their decisions.
For example:
>Stop considering some colleges as better than other colleges because that’s racist
No one says this, because it does not follow. One might say that it is important to recruit outside of the Ivies/Stanford, because you will miss qualified candidates who are otherwise pushed out of the industry despite their talent.
Or:
>Stop looking for people with relevant industry experience because it is sexist
No one says that, ever. I think the original kernel of advice may have been, don't only post listings for people with 15 years of experience. Your crud app doesn't require that, so give someone a little greener a chance.
I could go on, but this post is basically reactionary garbage.
*edit for formatting
For school I've been told that we should look if they have a degree or not, and not consider which institution. This explicitly considers all colleges the same.
Every point in the post is an actual suggestion made on how to improve diversity. Obviously not all the suggestions have been implemented, but I'm afraid the committee for improving hiring might do just that.
I'm not saying this wasn't said, but I've literally never encountered this before today. Perhaps your company is approaching diversity the wrong way for the wrong reasons.
To the point about the rankings of the colleges, it may matter less than you think. If you look at salary numbers over the lifetime of graduates, the ranking of their undergraduate has only a small effect for STEM majors. If you consider salary to be at least somewhat tethered to performance, this suggests that in the long run, the institution doesn't matter as much as the individual. There's some research around this, but I don't recall the authors at the moment. And, perhaps I'm wrong. I can imagine the differences are greater for recent grads.
Again, if they're all "actual suggestions," that seem strange to me. I never see or hear those suggestions, and they seem to run counter to the way people who are writing thoughtfully about tech diversity approach the issue.
For example, someone might argue that actively seeking out recent graduates of coding bootcamps when hiring for junior positions will help find a more diverse set of candidates. And there's truth here: bootcamps tend to have better gender and somewhat-better racial balance than university CS departments or existing tech shops, and bootcamp graduates on average seem to be pretty good (there are selection and maturity and self-motivation effects there which raise quality compared to the typical randomly-chosen bunch os CVs).
But it's technically correct, so long as you don't mind completely misleading connotations, to describe that approach as "to eliminate racism and sexism, hire people who don't have CS degrees and don't have industry experience" and imply it's "lowering the bar".
And anecdotally, when people make claims like the ones in the OP article, my experience is that it's almost always the case that someone is carefully choosing how they describe things in order to be technically truthful while maliciously misrepresenting the situation in a way that suits their personal axe-grinding.
I think it's pretty disparaging to say women don't have the time for writing some code outside of work because of childcare commitments (for everyone involved - including the father of the child who's immediately and implicitly branded as off-scene or not pulling his weight).
"Cultural fit" in my experience has been equated to "is this person someone I can see five days a week, 8 hours a day without driving me potty". That seems fairly legitimate to me - and completely inclusive of everyone.
I think it's still a useful metric, but would now not take a lack of portfolio as an outright negative, without other factors considered.
I think these are legitimate questions, since one of the best metrics for success in software is the level of interest you hold in it.
They are, and the fact that you even have to defend asking them is a non-trivial indicator of how bat-shit crazy the thinking has gotten on this topic.
And for that matter, you shouldn't penalize anyone for not writing a ton of code in their free time, there are other things in life, and more demands on your time the older you get. And anecdotally, some of the better developers I know write 0 lines of code in their off hours, they are consummate professionals but leave it at work.
Do basketball teams sit around worrying that they ought to not penalize players who don't practice free throws on their personal time?
I would also argue that a person who is more deliberate about their use of fewer hours will do better than someone who is equally smart but practices less deliberately for more hours[1].
I think that imagining two fully equal candidates only differentiated by their volume of personal projects isn't a particularly useful exercise.
[1]http://projects.ict.usc.edu/itw/gel/EricssonDeliberatePracti...
But that's precisely the point - two candidates will not be equal if one spends 20 hours a week on personal coding projects and the other does other things (however interesting or beneficial for society at large they may be).
At the end of the day the world is not fair. The question is what outcome you (as a business owner) are looking to advance. If your primary goal is to make the best product - you will have to hire the best people. If your goal is to advance women's careers - you will have to optimize for a different set of objectives.
> I completely agree. I’ve interviewed people from MIT who weren’t familiar with a hash table and the best software developer I’ve worked with never went to college. The CEO of Twitter, Jack Dorsey, didn’t go to a top CS school. However, it’s still the case that college and experience is an imperfect signal of ability.
In other words: Here's all of these great reasons I've found not to trust someone's schooling as a good indicator of their ability, but I'm going to use it anyway.
OP is what's wrong with tech hiring.
There also exists an athlete in the most prestigious wrestling program in the country who always loses because he never makes weight, and when he does make weight, he gets defeated in seconds.
Therefore I can confidently conclude that 'number of limbs' has no correlation whatsoever to wrestling ability."
Hmmm...
The top of a company's funnel could also be made more diverse by starting with "OK, there are 7+ billion people on the planet..." - that doesn't make it a good idea.
>And perhaps there aren't enough resources to improve said process and handle a greater volume of potentially unqualified applicants.
That's how you handle a great volume of applicants - you look for indicators of skill. None of them are necessary or sufficient; they just increase the likelihood that this will be a good hire a little bit. Giving every single person who is able to use the "resume" template in Microsoft Word an in-person interview would catch those false negatives you're missing, but it would be ridiculously inefficient.
Or am I to assume that when you want to hire a plumber you canvass your neighborhood door-to-door so you dont miss out on someone who might be qualified but just drummed out of the plumbing industry for institutional reasons?
Although I do like how you dismissed points 2-8 with a simple handwaving of "elite males are more likely to score higher on these points, so they're no good."
Not to mention the fact that that argument does virtually nothing to contracdict point #8. If youre someone who truly believes they're a Great Programmer Who's Not Being Given a Chance Due To White Male Supremacy, a github profile is the greatest arrow in your quiver that you could possibly dream of.
But if your company is taking on the diversity challenge then yes, start with the fact that there is 7+ billion people, (or at least the ones with Github profiles, to your point). And yes, canvas all available channels so you don't miss out on people "drummed out... for institutional reasons (there's an entire industry that does this). Then start over in defining your signals and indicators --in this case some sweet spot between the people they hope(!) exists and the false positives they wish they had missed.
If you're not willing to take these measures, simply sit back, do work and try to make money (or in this guy's case leave the company since he's not in charge). Otherwise, let's give retirement to diversity in this year's tech hall of fame.
But this doesn't smell right. Feels like a real-world version of this video:
If your company thinks it has a diversity problem, it's probably related to "fit". My suggestion is to continue to filter people out based on education and experience equally and give a boost in the "fit" category to categories of people that are minorities in your company. I think that's perfectly fair to offset internal biases.
EDIT: I get it some people are arguing that, by filtering on education/experience, there won't be any minorities left to be "boosted" in the "fit" category. If it is THAT bad, the company can try to expand who it's reaching out to but I understand that's a bigger problem (e.g. company in SV cannot afford relocation costs and thus has to look for local candidates only will not help with the diversity issue).
[points 2,3,4,5,6,7,9]
In addition, pretend like not ruling people out before you've spoken to them is equivalent to not reviewing their work, or giving them a "diversity" grade boost.
[points 1,8]
edit: and about "open source" contributions; ethnic minorities and women are far more likely to have worked in large corporations rather than in open source or startups, because those companies are held to a higher standard in their behavior towards minorities than the kind of self-assembling social clubs that this article (and site) are focused on.
edit 2: I know (very closely) a woman who has been in IT for 40 years. She was an English major, and has absolutely no involvement with open source. She doesn't know people in startups. Why would someone presume that she doesn't know the job?
So, suppose you had to fill an open position for a developer and you had two candidates: an MIT grad with years of experience at Apple and Google, and a homeless guy who never went to college and never worked in IT before.
You're honestly going to try to tell us that you have no idea who might be the better candidate?
And if hypothetically you could only interview one, you'd what? Be totally stymied? Flip a coin maybe?
There's no possible way you can actually believe what you're saying.
>In addition, pretend like not ruling people out before you've spoken to them is equivalent to not reviewing their work, or giving them a "diversity" grade boost. [points 1,8]
I don't recall him mentioning ruling people out before speaking to them, although once you get to a certain scale that has to happen at some point since if an individual manager gets 1,000 applicants for one position on his team he can't speak to them all. Or is math racist too?
The scale requirement becomes much smaller when talking about reviewing someone's work, because that is much more involved than a resume scan or even most phone screens.
Giving them a diversity grade boost is the definition of racism/sexism/*ism. Funny how youve managed to find yourself on the same side as someone who gives a buxom under qualified blonde a second interview because "we don't have any hot chicks in this office."
And yes, I registered this account because your comment was so breathtakingly ill-informed.
Open source software is probably the single most meritocratic resume-building opportunity I've ever seen in any industry.
If you don't like the fact that many minorities and women, far from being denied this opportunity, are simply declining to take advantage of it (as most developers do, by the way), then to paraphrase Richard Feynman, perhaps you should go find another universe where the conditions are more appealing to your sensibilities.
Or to put it another way, if you think that a submission on github containing a great new feature and some long overdue bugfixes for an open source project would get rejected because the submitter was black, then the part of your mind that desperately needs certain groups of people to be Irreparably Downtrodden is causing you to lose a great deal of your grip on reality.
The only test of skill is a test of skill. Blindly ask them both to make a thing, leave them alone for forty minutes to make it, and see who makes it better.
Let's flip all these changes around to show how it's not about "lowering the bar", it's about realizing that a one-size-fits-all "bar" isn't an effective hiring method in the long run.
1. Update your interview scoring methodology to account for non-traditional backgrounds. (Extra points for ethnicity, e.g., are a detrimental hack that shows how your system is broken.)
2. Don't discount candidates because of their degrees. (While a CS degree may be an indicator of qualification, its absence isn't a disqualifier.)
3. Don't discount candidates because of their educational institutions. (While a top college may be an indicator of quality, I'd rather hire the best candidate from a second- or third-tier school than a middling candidate from an Ivy.)
4. Expand your outbound candidate search to non-traditional channels. (See #3.)
5. Don't discount non-traditional industry experience. (A developer outside of the software industry may be just as good.)
6. Don't discount candidates because of their previous employers. (An ex-Googler may rightfully be an impressive candidate, but they also might be a low-performer.)
7. Don't conflate "years of experience" with "experience". (Very often, 15 years of experience is 1 year of experience repeated 15 times. No one has 15 years of experience in Node.js.)
8. Don't discount a lack of "extracurriculars" like OSS, volunteerism, or other factors.
9. Don't rely on referrals. (You can't staff a company entirely with "friends-of-friends".)
All of these statements are focused on broadening your perspective and getting away from statistics. Yes, a Stanford CS major with 15 years of experience at Google and an awesome GitHub profile is probably a good candidate. But a developer from Temple with 8 years of experience in the retail industry may be just as good.
Yes: it will take more effort to find those candidates, and your signal-to-noise ratio may go down, but your reward comes in the form of awesome people you never would have found before. Diversity will be a natural by-product, and a way to measure your progress.
"Diversity" shouldn't be apart of either of these scores. It shouldn't matter what race, religion or color you are.
I'm mostly surprised because "African-American" is a mouthful compared to just "blacks", so I'm suspicious that in common language it is somehow limited only to racists.
"Blacks" is also more accurate if you just want to refer to people by skin color (for whatever damn-fool reason).
Which means that instead of one HBCU being ranked on the US News list of best colleges for Computer Science [1] (albeit at the very bottom), there are zero.[2]
[1] - http://grad-schools.usnews.rankingsandreviews.com/best-gradu...
[2] - http://colleges.usnews.rankingsandreviews.com/best-colleges/...