To make a specific point, without prejudice to the main points of the blog post:
>One of the open secrets of working in technology is that technical interviews are completely worthless as a predictor for whether someone is a good hire or not.
[...]
>> We looked at tens of thousands of interviews, and everyone who had done the interviews and what what they scored the candidate, and how that person ultimately performed in their job. We found zero relationship.
> It’s amazing to think that Google found zero relationship between an interviewer saying “Hire” and whether the candidate was actually a good hire or not.
The fact [the scores of someone who gets hired] are not predictive of their eventual performance, does not mean that hiring process is completely random, or worthless.
In order to tell that, you would need to look at the cohort of people who were 'no hire' and compare them to the 'hire' cohort; it's not enough to just look within the 'hire' cohort.
In other words, the hiring process is (1) designed to be a Classifier, typically tuned with an emphasis on having a low false positive rate.
Hoping that (2) the scores of the small subset of candidates who are classed 'hire', also act well as a regression to the candidate's eventual performance, is a big additional ask.
Just because there is evidence that (2) is untrue, or particularly noisy, does not necessarily mean that (1) is broken.
I don't see a great way of solving this, without a large company like Google also randomly hiring 'no hires' from different stages in their hiring process (and subsequently measuring their performance) in order to quantify the performance of subsequent steps in the hiring process - not aware of anyone who has done that; it would be really interesting to read how a company who does that gets on, but an expensive experiment.
>One of the open secrets of working in technology is that technical interviews are completely worthless as a predictor for whether someone is a good hire or not.
This is wrong. Apparently google has internally done much research and their interview techniques have strong collreation to success on the job.
I'm a big fan of Dare's writing, but on this one, I think he missed the mark wildly, for exactly the reason you cite: Google's research says almost the exact opposite of what he's claiming in this post.
With that said, the large variation in ratios of disadvantaged minority employment at big tech companies does seem meaningful (and he cites that wide variation). I think intern and HR pipeline variation is likely very relevant to this difference.
The problem is with his implication that throwing darts to pick from a pool of moderately qualified people would yield a reasonable result.
Both "they work" and "they don't work" are unsupportably broad. Google has substantially reworked their interviewing process on at least one occasion, producing a big change in results. Story here: http://www.nytimes.com/2013/06/20/business/in-head-hunting-b...
Meanwhile it's crucial and nigh-impossible to identify how much success is based on interview performance as opposed to selection effects in who interviews, and how interviewers see different candidates.
> The fact [the scores of someone who gets hired] are not predictive of their eventual performance, does not mean that hiring process is completely random, or worthless.
It does not automatically mean this, but it is a strong indicator. If the output of the classifier is not correlated with a given output sample it is reasonable to assume, in the absence of other data, that it is uncorrelated for other samples as well.
I do accept it is evidence - but I don't think I would say a strong indicator, and it doesn't justify the strong conclusions made that the process is completely arbitrary.
- It's imbalanced classification: there are going to be more rejections than acceptances; so you have less statistical power, to find if there is a real relationship, when you just analyse the acceptances. Even apart from that, class imbalance can unintuitively affect post-hoc inferences.
- Scores are in context of the other candidates at a given level, and stage in the interview process, so are fundamentally already residual in some sense.
- I'm willing to bet, from hiring I've seen, that scores are distributed somewhat about the 'hire/no-hire' bar, and that extremes are rare - the process is going to be tuned, implicitly in the organisation, to make that 'hire/no-hire' decision, rather than predict performance.
> it is reasonable to assume, in the absence of other data, that it is uncorrelated for other samples as well
I think that's a very strong statement: it's not a random sample, its explicitly concentrated at one end of the distribution.
I do accept it is evidence, just not near enough to justify the conclusion the whole process is arbitrary.
No, it's not a strong indicator that the process is random or worthless. That makes the assumption that the people rejected by the process would have scored just as well. It could be very good at determining who has the technical skills to work there. It might be that their performance scoring of hires is measuring something else- like how much their manager likes them.
It could, but if technical skill is uncorrelated with performance scoring it will be really hard to determine how many people who make it through the process actually have the technical skill and how many don't.
It is possible. Check out the difference between a concurrent and predictive validity study. Also, range restriction corrections can approximate the unrestricted population variance and validity coefficient.
> it's not enough to just look within the 'hire' cohort.
Enough for what? Why not? They specifically have a "score", not just a hire / no hire decision. Sure it would be better, but you don't make any convincing argument why their analysis is not valuable.
> designed to be a Classifier, typically tuned with an emphasis on having a low false positive rate.
classifier of what? Why is classifier capitalized? I don't think so.
Let me throw out some made up numbers as an example.
Company X hires 100 people out of 500 applicants.
If only 30% of them perform well, you might argue the interview process is rubbish.
However, if you can test and show that of the "un-hired" 400 only 5% would perform well, then clearly the interview process is better than picking at random.
This is why just looking at hired performance is not indicative of whether the process works - because the task was not just to predict good performers but to predict a better rate of good performers than a random sample of applicants.
That points to some aspect (not race) of African-Americans that cause them not to excel. Is it broken families or a culture of not "acting white"? (acting white: https://en.wikipedia.org/wiki/Acting_white)
Those things needs to be dealt with on a community level, a band aid and a few initiatives by tech companies won't solve anything properly.
What do you mean? Controlling for the fact that children of African immigrants do very well should eliminate systemic racism, as they have been exposed to that their whole life, and would be exposed to all the "systemic racism" their African-American cohort members would have been exposed to as well.
Recent immigrants and their children will speak differently, dress differently, act differently and probably live in different places than multi-generation American black families. This creates a difference wherein the immigrant is perceived as not displaying the "bad" aspects of "black culture" (i.e., the immigrant's English is unlikely to be AAVE, which is a gigantic leg up right off the bat in terms of how they'll be perceived by white peers, and may even be considered pleasing; the immigrant will probably dress more "white", etc. etc.). Which gets right back to... you guessed it, systemic racism labeling "black culture" as an inherently bad or inferior thing.
The problem being discussed is prejudice. Whether that prejudice comes from some nebulous idea of "race" or "culture" or something else is besides the point.
The hiring data from high tech companies is not broken down by culture, but by race.
They're saying that racial equality is the important bit, and if black people aren't equally represented then there's prejudice.
According to that logic, it doesn't matter what culture the person has, as long as they have African ancestry. It doesn't matter if they're part of a high achieving immigrant community, or a low achieving black american community.
Not necessarily. It's possible, for example, that schools may have higher expectations and better supports for immigrant children. Immigrants may have economic or family educational advantages (they were the ones privileged enough to emigrate); they may reside in ethnic enclaves or white-majority neighborhoods that do not draw police suspicion and predatory economic practices; or, like the known hiring bias toward white-sounding names, human resources people may have an unconscious bias in favor of African-sounding names over recognizably African-American ones.
You realize that you bring up _nothing_ that is a factor of race, correct?
All the points you raise are valid, but have to do with culture rather than ethnicity. Nothing you said discriminates against people with an ethnicity from Africa than an ethnicity from Northern Europe.
They are all aspects of institutional racism in the US; they discriminate against African-Americans, while African immigrant are somewhat shielded from their effects. So, if your argument is that what we term racism should better be called culturism or something, ok I guess so. Still is a real problem.
See my comment to a thread above. Basically, according to tech hiring data, they _don't_ break things down by culture, just by race.
That's why it's such an important distinction. Is hiring 20 highly achieving black african immigrants who grew up in white enclaves just as desirable as hiring 20 low achieving black people who spent generations here being discriminated against?
According to these statistics, both are equally good and desirable.
Maybe we are going in circles. I perceived your original comment to place the blame on African-American culture (nonsense about "broken families" e.g.). Yes, hiring Africans into predominantly European-American companies produces one sort of diversity there, and it's a good one, but hiring African-Americans and Latino-Americans produces a different (also desirable) diversity, which does arguably more good in our traditionally less-privileged communities by getting more parents employed in good jobs and seeing real opportunities ahead for the kids in school to strive for.
Even in this comment I'd debate the term "low-achieving." There are many high-achieving African-Americans, and the automatic association of race/culture with low achievement is a problem.
I think we're mostly agreeing. The point I'm trying to make is that cultural diversity is arguably more important than racial diversity, but we're only measuring racial diversity where the ethnic background of the person is the only thing that matter.
For example, if a white couple were to adopt a black kid, and you hired that kid, that would boost "diversity numbers". What would be much more interesting to me would be to see what proportion of employees' parents had college degrees, were they the first ones in their families to get college degrees, the amount of money that their parents made when they were growing up, if their parents spoke the language as a native speaker, if they grew up in a traditional two parent family. That would be far more compelling and would result in more diversity.
For instance, there are plenty of fourth generation chinese immigrants to the US living in Chinatown that still don't speak English. If they tried to get help for coding or academics, they kind of out of luck because of the color of their skin. Similarly for white kids who grew up in a dying town in the rust belt, or an indian kid who's parents manage a hotel in the rural south. Because of their race, not because of their disadvantages, they're cut off from help.
And btw, there is not much nonsense about broken families. It can be debatable whether or not single parent households are good or not but black (including African immigrants, as they don't split them out) families are a lot more likely to be single parent. http://datacenter.kidscount.org/data/tables/107-children-in-...
You have to hit the whole pipeline, starting from k-12, and going all the way up through senior promotions. You have to think about which companies you do M&A work on, where you source in the world, and how you're going to engage with HBCUs and others.
Most importantly, you have to take action, and not just talk and write about it.
You have to change internal cultures. You have to recognize the biases and buzzwords such as "we don't want to lower the bar". You have to remind yourself that not too long ago, programming was something that very smart people, trained in many different things, engaged in as a side thing, not as a primary thing, and that's what's built our current industry.
meh, it's a lot of work, and will take many more years of concerted effort.
One of the key sources of this difference is whether you count "Information" degrees as CS. My number doesn't include information degrees and yours does.
People with CS degrees tend to be the ones building new software. Information degrees vary a lot, but tend to have decreased math requirements and be more focused on the application of technology. As a software engineer working in the bay area, I have seen hundreds of engineers with CS degrees and zero engineers with information degrees - in my experience there is a stark difference, and I would be surprised and interested if the data showed otherwise.
Just to add an anecdote, Most of the companies I've worked for screen out non CS degree's except for exceptional portfolios. Not sure how wide spread this is.
"I looked into it and it seems even after all these years the best way to get into Microsoft’s internship program is if your school is on the list of schools the company formally recruits from."
This isn't unusual for technology companies. It isn't even unusual for research labs[1]. So, basically, the company has delegated its hiring diversity to a university admissions department. Worse, many universities only accept people into their graduate program from certain other universities. I've never heard of a recruiter from any of these golden schools visiting anywhere near me[4]
Add to this ageism that is prevalent in technology and you have a the makings of lip service.
Why is ageism problematic? Many minorities, particularly enrolled members of Native American tribes, go to school later than their white or Asian[2] counterparts. Many go into the military or need the extra time to sort themselves out. They typically go to a community college first (different path from average), then go to a university[3].
The school I went to was on Microsoft's list to hire only support people from. IBM didn't have that hangup and got one of the best developers I've ever known and it worked out fine for them.
I would love to see these companies be a little wider with the hiring net, but I don't believe it will happen. They'll wring their hands and talk and release reports[5], but it won't change a thing because they don't get the basic problems.
1) in my youth I was refused for a summer internship because my high school and college were in the same area code....
2) seems Asian isn't considered much except in the reverse of requirements - see the lawsuits against the California university system by Asian students
3) I'll leave the discussion on scholarships and students lying to get scholarships directed at certain minorities for a different time
4) hell, when I was applying in 1987, I had to travel to get the applications, pay the fees with money I really could have used, and never get a reply. Someone told me it was because, like everyone on the reservation I was from, I had used a post office box for my address. That seemed a bit far fetched to me.
5) I'll believe something might change when I see "enrolled tribal members" listed on their reports and a list of schools that includes ones that actually care about recruiting in diverse areas.
I don't know that ageism in technology necessarily means that older students are less likely to be hired out of university. I would be interested in seeing data to support that.
As an anecdote, I have observed older students that attend the "right" school and achieve high grades are highly sought after by companies.
> I don't know that ageism in technology necessarily means that older students are less likely to be hired out of university.
If a company is less likely to hire a 35 year old or 40 year old professions that got out of college at the expected time with work experience, then a 35 year old graduate is going to have a much harder time. I cannot think of a situation where that wouldn't be true.
I would bet it would even be worse because it deviates from the expected path (high school -> college [w/optional graduate work] -> work) at that time period in a person's life and that would be foreign and weird to many teams.
As to the "right" school, that's quite a bit of the problem.
You may be right, but I would be interested in seeing a study of ageism and non-traditional students. I am one. In my experience, and observation of other students like me, the situation is opposite to what you are saying.
That said, none of my non-traditional student friends have sought jobs at companies or teams staffed primarily by 20-25 year olds. My observations are also skewed more toward hardware jobs, which seem to value older employees.
One critical point I feel this piece does not touch on is how diversity of experience is great in the workplace. It's a huge benefit as a developer (and obviously other professions too) to work on teams with people who have lived very different lives. The Mirrortocracy[0] covers this well, but it should be clear to us all that being exposed to different views helps refine our ideas and process.
It's partly up to us to make diversity happen. We can prioritize offers that have diverse teams, and when we get a chance to hire to value diversity of experience.
I absolutely agree, though I always feel like culture and nationality are very underplayed in this discussion - I believe an American, an Eastern European and a Chinese national would bring a lot more diverse world views to the table than a female and a male American (or black and white).
> It's a huge benefit as a developer (and obviously other professions too) to work on teams with people who have lived very different lives.
I can see how that would be true for some types of development, but there are also many type for which I don't see offhand how it would matter.
For instance, if I were doing web design for my employer's shopping cart site I could see how having a diverse team could greatly help because the team would have people on it that are part of or identify with more of our customer demographics.
I as a white male mid-50s atheist tall fat guy could easily inadvertently come up with a design that might turn off non-whites, females, young people or elderly people, religious people, short people, or skinny people. Even if I don't end up doing something to offend people, I could simply miss opportunities that people of other backgrounds might see.
In fact, I've seen that kind of thing. I saw a company that was making CD-ROM caching software in the '90s find a nice cluster of sales when an employee who was also a Mormon pointed out that this software worked extremely well with the CD-ROM genealogy databases that were becoming quite popular among Mormons. Without that Mormon employee, they would have probably never noticed that Mormons could be a distinct market segment for this product that was worth specifically targeting.
What I actually work on, though, is backend stuff like processing orders and subscription billing, interfacing to payment processors, reporting sales tax in the US and VAT in Europe, analyzing A/B tests, and so on. Would having a diverse team working on this actually produce any different results?
Wouldn't "companies passing on highly qualified candidates due to a broken process" be a self-correcting problem, where other market players with non-broken hiring processes would be able to scoop the talent up and gain a competitive advantage?
That's the whole point of the article, really... if Apple can manage 8% black representation, why can't Microsoft, Google, Twitter, or any of the other top-tier companies break 2%? It puts the lie to whatever excuses they make. Apple is clearly not "lowering the bar".
Right. So eventually companies following Apple's lead will be rewarded, while the companies following a contrarian route will either adjust or disappear.
Over time, I imagine this will self-correct. For the last decade or so the popular thing has been to copy Google's interviewing/hiring process because everyone wanted to be Google(in terms of engineering notoriety, not product/industry). It seems like that is beginning to fade out of popularity a bit though.
>It seems like that is beginning to fade out of popularity a bit though.
I've been interviewing for last 2 months. And every single company regardless of their size/location/industry had the same exact process. Phone Screen->coding exercise->interview street quiz -> whiteboard,bigO onsite interview.
I remember when take-home projects and then online in-your-own-time exercises started becoming popular, and it's unbelievable that people now seem to be doing them in addition to the whiteboard sessions that we all agreed were crappy and trying to figure out how to replace! We really seem to be in a strange spiral where people are complaining ever more loudly that it's hard to find developers, clamoring for trade schools and public education to create more supply, while simultaneously consistently making it harder and more soul-crushingly irritating for qualified people to actually get a job. What's going on here??
yea its hard to find a job these days without massive time investment. Coding exercises I've been doing are typically 4 hrs or more (example: https://mesosphere.com/careers/challenges/distributed-applic...), phone coding screen is 1-2 hrs, interviewstreet is 1 hr. So you are looking at 7-8 hrs for single application. Then comes 2-3 days time sink if you have to fly out to different city for onsite interview. I can't imagine people with kids/family/parents to take care of devoting this insane amout of time. Some of them like Amazon don't even respond to you with the results if you fail( assuming) their online quizzes.
I have like over a decade of experience and have 3 github code projects with over 1000 stars. But I have a feeling that companies are getting so many resumes that they don't even want to look at your resume before you have been screened. Its almost impossible to switch jobs these days without massive time investment, unless you are some sort of super star that can somehow short-circuit this process.
How is it disproven? He points out how Apple is above peer-average on diversity metric as well as significantly above peer-average on financial metric.
> This is where I’ll talk about the Big Lie from the title of my post. That lie is that there is some sort of pipeline problem preventing tech companies from hiring more black people. The reality is that tech companies shape the ethnic make up of their employees based on what schools & cities they choose to hire from and where they locate engineering offices.
I agree this is a substantial part of the problem, particularly in larger corporations where they do target specific cities/schools.
However, I'm not sure I can agree with the implication you should locate engineering offices based on the demographics as its the only way to fix the "cities where they locate their engineering offices". Demographic/diversity shouldn't be a criteria for office location.
You shouldn't have to move your entire organization just because the city you happen to be in is ~5% Black and the largest minorities happen to be Asian or Hispanic.
For instance at $Day_Job, the floor I'm on is 25% Asian, 5% Hispanic, 5% Black, 65% White. That is pretty much +/- 5% of the actual demographics.
I wouldn't argue we lack diversity because we just follow the local demographics. I'd say the problem is when you are so out of whack with the local demographics it becomes absurd. For instance, an office in Atlanta that is 90% white.
> tech companies shape the ethnic make up of their employees based on what schools & cities they choose to hire from
I wonder if even that is true for California tech companies. California cities feature a diversity which is absent from their local tech companies.
California's tech companies appear to be here for the VC money, not the local labor pool. California's schools, by and large, do not output a globally competitive workforce. The tech workforce is mostly brought in from somewhere else.
And California's government doesn't seem to care about the demographic mismatch.
Tbh, it probably has something to do with the fact we don't hire based on school name at all or even degree. We just care if you can perform X business function and have a couple years experience at it.
The OP pretty much only lists "Tech companies from SV" so it seems to me at least its a cancer specific to that group.
I'm always amused when people post racist comments and then claim "I'm not racist". Just own your racism.
In my article, I specifically talk about Atlanta which is a "black city" as you term it. Atlanta is home to Georgia Tech which is one of the best engineering schools in the country. So it is strange to imply you wouldn't be able to find CS graduates there.
Secondly, companies have no problem opening engineering offices in China or India to chase after demographics so I find it strange that opening an engineering office in Atlanta is such an absurd idea that one must trot out the must played out racist tropes about black people.
Georgia Tech only enrolls ~7.5% black students so it wouldn't help their quotas. Also companies aren't opening offices in Asia to chase after a demographic, they do it to cut costs.
> Black cities are backwards and shitty and don't have CS grads, so no tech companies are HQed in black cities, but it's got nothing to do with race. I'm not racist: I'd have lots of black friends if they weren't all living in poor neighborhoods and jail where I don't go. We avoid logic and good data and good analysis of data when it comes to race, this post is at least getting a little closer.
Eh? There are plenty of tech companies in cities with large numbers of African Americans. I'm just saying you shouldn't try to imply that 100% of Tech Companies should move to Atlanta (for instance) solely because there are a large number of black people there. You should be in Atlanta because it makes financial sense for the business you are operating.
I said it shouldn't be a criteria used to determine location. For instance, if you are doing eCommerce with a large import/export component, you are probably going to want to be HQ'd near a major port and you probably won't want to relocate any further than you have to (if you have to relocate at all) as a founder.
There are plenty of places on the West Coast where the largest minorities are Hispanic or Asian, they have easy access to factories in China & Mexico, and have small Black populations.
That doesn't mean they should relocate to Atlanta where they don't have physical access to the locations their merchandise enters the US.
The Bay Area is another area where this is an issue but it doesn't make sense to move (i.e. If you want access to Silicon Valley):
The area is less than 10% African American. You aren't going to have more than 5-10% African American employees unless you are biasing things intentionally. The largest minority? Latinos followed by Asians.
While there is more than adequate evidence of racism in tech hiring, it would be useful if articles like this included the expected percentage of X engineers if there were no bias in the hiring process.
E.g. if the expected percentage of black engineers if there were no bias in the hiring process was 2%, that seems like important information to include.
Something the author hints at is that the big discrepancy is geographic in nature, not racial. The geographic discrepancy manifests itself in not enough blacks getting hired.
Let me put it bluntly: I grew up in the rural south. I NEVER meet other programmers who have a background remotely similar to mine, ever.
When was the last time you worked with a coder who grew up in a place that wasn't a wealthy suburb or a nice city?
How many people who grew up in trailer parks work at Google or Apple?
The author is right: not enough black folks at these companies. What is sad is that this is a symptom of the real issue, which is socioeconomic and geographic in nature, and the companies trying to do something about it are only focusing on the symptoms.
This means that even companies like Apple will never fully address the underrepresented groups. America doesn't have a race problem, so much as a culture problem. Obviously culture is extremely aligned with race, but think about it this way: the geographical and socieconomic layers that are excluded by these tech companies are filled with people of all colors, and nobody gives a shit about the people that look too similar to the white folks from wealthy cities that have every advantage.
Given how much of the job is based on resourcefulness and 'sticktoitiveness' which wind up being completely unmeasured by the typical big-o-whiteboard-olympics.. I wonder if it wouldn't be an edge in hiring to fast-track anybody from a disadvantaged background. They've proven that they can break the mold and overcome adversity, kids from the suburbs->stanford route haven't, necessarily.
> Something the author hints at is that the big discrepancy is geographic in nature, not racial.
It's not an either/or proposition; it's both/and. According to the article, Google, Facebook, and Twitter are all ~2% black. The population of Silicon valley is just shy of 8% black. The national number is about 18%. Silicon valley is an unusally white area, and the staff demographics are skewed even relative to that baseline.
For what it's worth, I am originally from a small industrial town in rural Arkansas, and I am white. Getting hired in a tech firm was actually quite easy for me, given that I was willing to move away from where I grew up to an urban area with a thriving tech sector. I'd like to think that would still have been true even if I were black, but...
I live in Arkansas. I'm in the little bubble of prosperity surrounding Bentonville. I'm moving to Denver in 2 weeks because my job is requiring me to.
Some of the poorest people in the country (most of them white) surround this little bubble all through the Ozarks, extending east into Appalachia. You're from here, and you know that. Moving to an "urban area with a thriving tech sector" isn't exactly cheap. Getting into a good college is a nice precursor to that, and that is radically helped by having a decent public school system.
I don't think it's a stretch to say that these are very difficult obstacles for people, no matter there color.
In fact, I'll say it now: a black kid who attends school in Fairfax County, Virginia is going to have a better shot at a tech job down the road than a white kid who attends school in Ft. Smith, Arkansas. These are extremely different places, but I'm pointing this out because while race matters, waaaay too much emphasis is placed on it compared to the structural factors that impact people independent of ethnicity.
So I agree with most of what you're saying. I turned a state-funded boarding school education into a scholarship to an engineering university, and I'm acutely aware of how much luck is involved in that story, and of how much would have been difficult if not impossible if my family hadn't been at least middle class to start with. We do have a very real class problem in America, exacerbated by a deep urban/rural cultural divide, and it is a disgrace how little we're talking about it.
Unfortunately, we also have a very real racism problem, and I don't think it's helpful to say "the real problem is classism". The very real problems include both racism and classism. It's too hard to get into tech as a poor white man. It's too hard to get into tech as a middle class black man. For that matter, it's too hard to get into tech as a middle class white woman. And it's way, way too hard as any combination of those.
In general, responding to "we have a huge racism issue" with "we actually have a huge classism issue" isn't wrong, but it isn't helpful either. It derails the conversation from a very real problem into arguing about who has it worse. And the worst part is that most of the changes that would help the racism issue would also help most of the other -isms. Blind hiring, more support for remote work, and less dependence on specific universities in hiring pipelines helps everybody. And even very targeted interventions help indirectly to the extent they erode the stereotype that only upper-middle-class 20-something white guys need apply.
> In general, responding to "we have a huge racism issue" with "we actually have a huge classism issue" isn't wrong, but it isn't helpful either.
What if the classism problem is a significant cause of the racism problem? Race and class correlate, so if someone has a problem with a particular class they are going to learn to associate certain races with that class, and their classism issue becomes a racism issue to the outside observer.
> These are extremely different places, but I'm pointing this out because while race matters, waaaay too much emphasis is placed on it compared to the structural factors that impact people independent of ethnicity.
I think that is one of the big impediments to actually solving many of the current problems that involve race.
For instance, take the issue of people getting shot by police. It is a fact that blacks are shot at a higher rate than whites in the United States. It is also a fact that blacks commit violent crimes at a higher rate than whites. Most people get hung up on one of those two facts and won't go farther.
The liberal side (I'm going to be doing some broad generalizations from here on...) gets hung up on blacks getting shot at a higher rate and attributes it to obvious racism on the part of police. They don't want to consider that the police may be shooting blacks at a higher rate for non-racist reasons.
The conservative side points to the higher crime rate among blacks, takes that as a satisfactory explanation for the police shooting data (and they are probably reasonably right, at least for the proximate cause of police shootings), and then gets hung up on that. They don't want to consider that there may be some factors that are disproportionately pushing blacks into conditions that tend to foster higher crime rates among people in those conditions regardless of race.
In particular, there is a pretty solid correlation between race and wealth in the US. Black people are disproportionately in poorer cites and states, where schools are worse, prospects are dimmer, and there is high population density. Those kind of areas tend to produce higher crime rates regardless of the race or ethnicity of the people that live in them.
(Well, to be accurate, that should be "higher crime rates for the kind of crimes that often gets you shot by police". White collar crime tends to be a lot safer during the arrest, because those arrests tend to come at the end of a long investigation and the arresting officers know your movements and can put a lot of planning into your apprehension. When you get arrested for robbing a liquor store or shooting a rival gang member, the arrest tends to be a more hastily organized affair, with a lot more improvising).
I've seen dozens of articles in newspapers and magazines that have looked at the police shooting data, and heard numerous stories about it on my local public radio station, but none that I've seen/heard have looked at how this relates to wealth levels. I realize that would be harder to do because that's generally not something that is directly in the data. But without that analysis being done, we are just going to be stuck at the "police are racists"/"blacks commit too much crime" level, which is not productive.
> The population of Silicon valley is just shy of 8% black. The national number is about 18%. Silicon valley is an unusally white area [...]
Unless you are counting Asians as "white", that is not correct. The nation as a whole is between 63% and 77% white (depending on whether or not "hispanic" and "white" are considered disjoint). In Silicon Valley, whites are only about 35%, and are not even the largest racial/ethnic group, having been surpassed by Asians.
It's also curious that everything that ever seems to be published on diversity - especially in technology - seems to focus either on the percentage of representation of black people or the percentage of representation of women. I guess this author focused on the percentage of representation of black people because he's black, and that impacts him, but just anecdotally based on the 25 years I've been doing this, I'd say that Indian people (men and women) are vastly overrepresented as a percentage of the tech workforce. Is that due to racial preferences for Indian people? Articles like this would seem to draw that conclusion - either the underrepresentation of black people is not due to racism or the overrepresentation of Indian people must be, correct? I also find it interesting that people are studying, say, the representation by gender and race but not, say, religion - how under/over represented are Jewish people or Muslim people in technology? Are there biases there?
There's a backstory to all this - Indian immigration is largely folks with wealthy backgrounds and high education. That's understandable without resorting to color. But Americans from similar schools and similar levels of wealth are not hired in Silicon Valley - we look for another reason. And it happens to correlate with skin color.
Just looking at stats and pretending that's the whole story is disingenuous.
A few very bad assumptions there:
1) not all employees have CS degrees from the US, many are foreign
2) the graduation rate by "race" or "gender" in 2015 is not what is was in 2005, 1995, 1985 or 1975, and the any industry has many people who graduated a long time ago. Comparing current year demographics to the industry as a whole seems really off.
3) Due to historical economic status which is correlated with race, more of those that started off less well off tend to gravitate towards more stable companies.
None of these things mean that negative racial discrimination or stereotyping don't exist, just that we shouldn't expect numbers for a particular industry population to match the aggregate stats on who graduated last year.
EDIT: I believe one big source of the difference between our numbers is whether you count "Information" degrees as a CS degree. My number only looks at "Computer science" degrees, whereas your data looks at "Computer sciences" degrees.
That's useful. Whether or not that's the right number is difficult to say, but it's good to have some sort of baseline assumption to frame the rest of the piece.
But that's not the number to look at for an article like this. The whole idea of pipeline problems is that you lose proportionality a little at a time, from early education to full time careers.
Ideally, you'd check the proportions starting from something like mathematical success or career intent in high school. Then you move on to college majors, graduation rates, and applicant pools. This is really important, because otherwise you end up focusing energy on non-leaky parts of the pipeline and achieving nothing.
This article focuses on interviewing, but the segment on black colleges and office siting is crucial. Ensuring unbiased interviews doesn't stop you from drawing interview candidates from disproportionately white colleges. Holding local code-ins won't save you from having offices in overwhelmingly white cities. "Lowering the bar", which companies are (according to this) hugely concerned about, still wouldn't solve the problem.
If you're studying the mismatch between 13% of the population and 1% of your workforce, the numbers you want are the ones that show where things went wrong.
I gaurantee you out of the people who can code or have cs degrees, blacks make up a much smaller percentage. You can't blame Apple for the deep societal issues we have that leads the average black person to be more likely to end up in prison than college
According to the numbers Dare sourced in replies in the thread, blacks have significantly higher CS college graduates per capita than whites/hispanics, and about on par with Asians. The number of blacks within just those who have CS college degrees is ~10%.
That number is misleading, because that data uses a "Computer sciences" category which likely lumps in "Information" degrees. The number I see in 2015 is ~2.5% (https://news.ycombinator.com/item?id=12101868)
To really understand if companies are biased or not, you also need to know the percent of applicants to these companies who are black. If only 2% of applicants to Google are black, I would expect only 2% of new hires at Google to be black.
The assumption that a white applicant and a black applicant should be roughly equal is a strong prior. I would need to see convincing data to counteract this assumption.
Absolutely, this is rather the point of the article: that diversity efforts focused on eliminating bias during selection are pointless unless you also ensure a diverse candidate pool to begin with.
I empathize with the author's frustration but there were many parts of the article which I found troubling:
> "I also quickly learned that hiring managers for “good internships” don’t come to black colleges looking for interns regardless of how good the students are."
I've never met anyone who didn't claim that their colleges is full of good students. There seems to be a heavy bias here - everyone has a rosy opinion of their own college, but clearly, companies can't go to every single one. If you feel that X college, whether it's a HBU, state-college, community-college, or code-camp, is great enough to be shortlisted over other alternatives, you really need to make a compelling case for this, using data. Just saying that Microsoft doesn't recruit from the college you used to attend, and therefore, they are being unfair/stupid, doesn't sound like a great argument.
> In one of the most reasonably-priced major cities of America, a ten-minute walk from one of the world’s top ten engineering schools, Google has an office. There were once dozens of SWEs and SREs in this office, humming along, cranking out GWT, nerd shit etc. Google decided to close Atlanta engineering
Again, this sounds exactly like the point above about colleges. There are ~30 major cities in America, and Google can't open significant dev-centers in every one of them. If you think that ATL would provide great value for Google along all axes (not just diversity), you need to make a compelling case for it. Just stating that Google doesn't have a major office in ATL, and therefore they are being unfair/stupid, isn't a great argument.
> There were also the Hiring Committee meetings that became contentious when I advocated for diverse candidates. Candidates who were dinged for not being fast enough to solve problems, not having internships at ‘strong’ companies and who took too long to finish their degree. Only after hours of lobbying would they be hired.
> when tech companies talk about “lowering the bar” by hiring minorities they are actually just saying they don’t want to hire minorities since no one in tech actually has a bar that works very well in determining good versus bad hires regardless of ethnicity or gender.
This was the part I actually found most troubling. Google has an extremely high rejection rate - a large majority of their candidates end up getting rejected for "not being fast enough, not having strong internships" or for any number of other reasons. When the average Joe gets rejected for the above reasons, it's ok, but when a diversity candidate gets rejected for the same reasons, it's suddenly a problem?
And what exactly is the author advocating when he talks about how interviews are useless, and therefore, there's no point in having a bar at all? It just sounds like the author is saying that there's no way of figuring out who's going to be a great candidate, and therefore, you might as well just set some minimum qualifications, and then hire anyone who meets those minimum qualifications and happens to be a diversity-candidate. It's arguments like this that make people think that diversity == lowering-the-bar. Yes, hiring is extremely hard, but no, that doesn't mean it can be skipped or watered down. Every successful company needs to figure out some heuristic that will allow them to hire only the best candidates, and any diversity push needs to be made while still respecting this stringent heuristic. If you think that Google's heuristic in unfairly biased against diversity candidates, in a way that's unrelated to job performance, you really need to provide evidence for this. Simply saying that Google rejected some diversity candidates because they didn't pass their evaluation-process, is a really weak argument.
Overall, I agree with the author's thesis that the interviewing process is biased in many ways (race, gender, age, height, charisma, eloquence, etc etc), and that we need to do more to fix these problems. But the arguments he presents are so weak, and the insults he piles on MS/GOOG/FB are so unfair, that I find myself disagreeing with most of what he says.
I'm honestly confused by your response. Companies like Facebook, Twitter & Google hire 1% blacks in tech and 2% overall. Blacks as a percentage of college graduates in engineering or non-engineering fields are way more than 2% of graduating classes.
However most Silicon Valley tech companies under-hire from this demographic for their own reasons.
Now we need to make an implicit assumption that the enrollment and graduation rate of blacks is the same as the rate for the general population(I couldn't really find information on that, apologies). Using demographics data, blacks are ~15% of the overall population in the US. That means that about 0.75% of the total population are "blacks" && "engineer major". That would mean all the listed companies are overhiring black engineers if simply based on a college-graduate quota.
Besides that, I liked the gist of your post, but the end seemed incredibly disingenuous. Correlating Apple's number of black engineers and its value does not imply causation, and that statement seemed like an emotional appeal.
That's really weird math since you start with engineering college graduates then jump to talking about the entire population. So you're comparing apples to oranges.
Blacks are about 7% - 8% of college graduates with CS degrees each year. So one would assume that if tech companies were hiring US college graduates at an equivalent rate based on graduation rates then most tech companies would look like Apple that has 7%-8% black employees in their workforce.
> Blacks are about 7% - 8% of college graduates with CS degrees each year
Your data lumps together "Computer sciences," which probably includes Information degrees. If you look at CS degrees only it's about 2.5% in 2015 (https://news.ycombinator.com/item?id=12101868).
Your logic is faulty, because the pool of engineering employees is (roughly) limited to engineering majors. Nobody is complaining that teachers and nurses are underrepresented in the tech industry.
Regardless, the true number is around 7-8%, as is helpfully noted elsewhere.
How do you check that? I've found reporting on race in tech to just be absolutely horrible with numbers. Whole stories where the only number is "Google repots X percent of Y race." It tells me absolutely nothing without comparing that to demographics of where google has offices, and what the demographics of cs students are, among other things, and I've never seen that info.
The article addresses that directly. The author talks about Google more or less abandoning its Atlanta office, where a diverse workforce is easier to get than elsewhere.
Even if Blacks get CS/Engineering degrees, they can still be less competent than Asians/Whites. Tech companies attract the most talented people (e.g. Google gets millions of job applications per year) and they can pick those who are the most competent/best in class. Those are usually Asians and Whites. For example, if 10% of all CS degree holders are Black, you can't expect that there will be 10% of Blacks at tech companies.
Hilarious or not, that's the (replicated) result from the literature:
> Implications from this study are much the same as those stated by Frey and Detterman, namely, that SAT appears to be a measure of general intelligence and is a useful tool in predicting cognitive functioning
Yes, it's an indicator of competency in mathematics, which in turn makes a good computer scientist. What else would be a better indicator of math competency than math tests?
The hilarity is in how this perpetuates the reasons for excluding a class of individuals from participating in the software creation process.
Good in math probably won't give you the empathy and insights into how technology could help rural Kentuckians.
Additionally, software today is very much a 'poke at the frameworks, and mash it together' sort of affair. When I joined MS 17 years ago, knowing how to use WinDbg was probably a big skill to have. Today, it's more about knowing how to use various JavaScript frameworks.
Math specifically comes into play perhaps when talking about data science specifically, but there's an app for that.
If Math is a stand in for an ability to think critically, and apply the scientific method, then biologists, physicists, and even social scientists, might be just as qualified as any engineer.
So, yah, one more exclusionary myth about the qualifications of engineers, and who's best at it (whites and Asians? really?).
The hilarity is in how this perpetuates the reasons for excluding a class of individuals from participating in the software creation process.
You are excluded from companies using computer science if you are bad at math. Blacks are worse at math than Whites and Asians, therefore they will be hired less - especially in companies where everyone wants to work.
Good in math probably won't give you the empathy and insights into how technology could help rural Kentuckians.
Will being bad at math give you that? What does building software have to do with rural Kentuckians and how are black software engineers going to help you with that?
Additionally, software today is very much a 'poke at the frameworks, and mash it together' sort of affair.
Are you trying to say that Blacks are as good as Whites/Asians at poking at frameworks. If they are bad at math, maybe they are bad at this as well. BTW, I am referring to computer science, not knowing JS frameworks. Computer science requires mathematics.
So, yah, one more exclusionary myth about the qualifications of engineers, and who's best at it (whites and Asians? really?)
What is mythical about this? I've given you the data that clearly shows that Blacks are worse at math than Whites/Asians. They also perform worse at SAT-Critical reading.
Well, I suppose this is fun at a certain level, because there are assumptions, and conclusions, built upon assumptions.
I'll try to clarify my position.
The back of my Microsoft badge says something about helping everyone on the planet.
Such a mission is as much about emotional intelligence and an ability to be empathetic as it is about actual application of technology.
Let's imagine the very high IQ'd high scoring person, who grew up in a certain environment has the task of creating software for someone in a rural community they know nothing about, and they will not visit. They are very clever, and they'll come up with something.
Now, let's imagine a child who grows up in that rural community. By standard testing methods, they don't have the same level of skills as the high SAT scoring individual, so they never get hired into tech, they're never consulted about what might help the needs of their community. They are in fact intelligent, just didn't go to a very good school, nor had the various opportunities that the high scoring SAT person did, so they're out of the loop.
So, in my mind, if I have a mission to help such people, it seems the best way I could do that would be to incoroporate them into my dev team in some way. That might include providing them with additional educational opportunities they might not have otherwise had. It might be a temporary contracting or consulting role rather than full employment.
It does not mean total exclusion from the creation process, which is what occurs today based on things like a test score, or a college affiliation.
This is the argument of why diversity matters in tech. All the SAT scores, college graduation rates, and the like, are just the pieces we're moving around to justify our reasoning for remaining fairly homogenous in our hiring practices.
At this point, you don't know me, my history, ethnicity, nor accomplishments in tech. The full reveal might be enlightening.
What else would be a better indicator of math competency than math tests?
Have you investigated why there's a difference? Your comments seem to be leaning toward dismissing this problem as not in need of an explanation, but the explanation matters a great deal. Do you believe those scores could be changed by changing educational approaches? Do you believe there's something genetically innate causing the lower scores? Do you believe there's some social factor or combination of factors which could be changed?
You don't get to just say "oh, score gap, question answered".
It's quite surprising to me that Apple seems to be doing so much better at diversity hiring. I think Google, Facebook, and Twitter are quite serious with respect to diverse hiring / I find it hard to believe Apple's methods wouldn't be found out and copied by any of them if they worked so much better.
Has anyone looked into the possibility that rather than Apple being uniquely good at diversity hiring, Apple is just uniquely good at gerrymandering their diversity report?
In Apple's 2014 EEO-1 they distinguish between professionals and technicians. Professionals are 1.7% black and technicians are 11.3% black. I assume both of these are lumped into a single "tech workforce" number.
Wow, thank you sir. Note that 1.7% is roughly the same percentage that Google, Facebook, and Twitter have according to OP. I hope OP sees this and reevaluates his position now that Apple's diversity numbers no longer appear to be 4x higher than the rest.
Did we read the same post? The whole point is that they are not actually serious. There is a lot of lip service, but then diverse candidates in practice are stonewalled at every step: by location, by academic background, during evaluation etc.
If they were serious, they could look beyond their institutionally-placed (and fairly arbitrary, let's say it) obstacles, like Apple did for his friend; but they won't, which means they are not serious.
Yeah. Going further, another part of "being serious" is training and a demonstration of commitment to the employees after they're hired.
Something like "Yes, this individual did not have a great education in youth because their local school district sucked. Nevertheless, this individual has potential. So we're committed to keeping this person on board and providing the necessary training. We will go to great lengths to make this person a useful and vital part of this company."
Amen to that. It's a whole pipeline thing that begins before they even interview, and extends at least as far as their first few promotions.
And you do it not just because you want to tick off some numbers, but you do it because a diverse workforce is going to create a more valuable and relevant company.
Right, that's the thesis of the article. The basis for concluding that Google, Facebook, etc were not "actually serious" was their numbers being lower than Apple's numbers.
I was asking if perhaps OP's data analysis was flawed because (for example) Apple might be counting Genius Bar employees as "tech" for the purposes of the diversity report... ie maybe Apple has the same difficulty hiring diverse candidates as Google for similar positions, OP's friend's experience notwithstanding.
on top of that, apple is much more of a device company than the others. this means that they will have a lot more people on the technical staff that aren't developers and those other positions may have a larger candidate pool of blacks.
I'm sorry but I get bothered by the current national focus on rewarding or advantaging diversity based on arbitrary traits such as skin color and gender.
I'm 5'8 and male. Height tends to correlate to success and salary. Should I receive special treatment because of my disadvantage? Perhaps a certain ratio of company board members should be required to be under a certain height threshold.
I've also been bald since I was 18, a trait I have no control over, but which likely led to some lost opportunities (with the opposite sex at least). Do I deserve something special for that?
Companies should hire on the basis of merit, not skin color or sex.
I did undergrad EE/CS at Illinois/Urbana and I don't recall any blacks in any of my classes at all.
I've worked at high tech firms and never see black programmers/engineers.
I've worked for a Wall Street bank and never saw black programmers/engineers.
I've been to many tech Meetups and conferences and other conferences where software plays a major role but I've almost never seen black engineers/computer scientists.
Some of the meetups / conferences I see few women, but I do see women in the aggregate, but almost never blacks and I'm in NYC where there are many blacks.
I've been to Google Tech talks in the Google building in NYC yet never see blacks.
Since the meetups don't require you go to "the right school" etc. and are free to attend it shows that there just are that many blacks that seem to be interested.
Computer software and hardware are those areas where competence is extremely important. Hire on merit, not on other indicators.
As someone who grew up poor (albeit white), I'm going to try to explain a core problem to you - role models.
When I looked at the adults around me growing up, I didn't see educated people. The responsible adult professionals were motorcycle mechanics, truck drivers, stuff like that. The less responsible were day laborers or petty criminals. The only people I encountered who were educated were teachers and doctors, and the wealthy men my father often worked for whom he condescendingly called "edjicated mow-rawns" in his thick Kentucky drawl. What I knew of success was seething class resentment.
I don't think I knew a single adult who was an engineer or technical professional. But I was clever, and read a lot of science fiction. I was originally interested in automotive/motorcycle engineering, having grown up around racing. But I discovered computers in high school, and that was that.
The boundaries we can imagine are limited by our childhood experiences. If you grow up wealthy and privileged, surrounded by doctors and lawyers and executives, you expect adulthood to look like that. If you grow up in a world where the only adults you know are hustlers and criminals, that's what you can imagine.
So why aren't there more black engineers? In part, because there aren't black engineers. And what black engineers there are, are underemployed, because they didn't necessarily go to the elite schools or have the other earmarks of privilege.
Fixing this is going to take more than the myth of rugged individualism.
Glad to hear you made it out. But seems like your own story gives lie to the theory that role models are necessary. You didn't have any; didn't need them. Just needed opportunity.
It's possible to escape without role models - but it's a lot harder. And quite frankly, I'm still privileged. I'm white. I'm male. If I'd been black, I don't think I would have escaped. The fact that I'm a white male means that, once I've melted into the new System, no one notices me as different. One of my best friends, who is far more professionally successful than me, is black. It still trips him up all the time. If he drives through the wrong town, cops don't see a doctor who shifted careers to finance. They see a big dark-skinned black dude driving a car way too nice for him.
White male engineers, working careers surrounded by white males, socializing mostly with successful white professionals, typically have no clue how everyone else lives. Their advice is mostly useless. Or, as a female friend recently put it, it's like someone whose entire experience with horses is owning a stuffed unicorn giving advice on how to ride to people struggling with actual horses.
How many events have you attended as the only white person?
How comfortable did you feel?
Do you think your merit was accurately reflected at these events?
The attendance of the events have no function to keep people out so merit is not required, although there probably is some correlation with merit.
The Meetup events that I attend (in NYC which is very multicultural) has people of all skin colors, not just white people. Lots of Indians, lots of Asians, just all sorts of colors, so it is not like one black among whites.
At any rate, if blacks want to advance technically and get jobs, going to Meetups and connecting with others is a good way to do it. All these "diversity" initiatives are the wrong way to do it.
So the lack of black participation in free tech events === lack of interest from black people? That's a stretch of a conclusion rife with confounding variables.
Could you please explain in plain english with examples exactly what you mean by that statement?
NYC where I live has at least 1 million blacks.
NYC has many hi tech firms
NYC has many banks and financial institutions and media groups that all employ programmers and engineers.
It has many tech Meetups.
It has many opportunities to attend tech events that are free, and where there are no restrictions on attendance.
Where I can agree is that there are basic opportunities (e.g. networking) in the tech community that Black Americans may not take interest in. The stretch comes when you equate that to having no interest at all in technology (correct me, if I've misconstrued your point there).
What makes it a poor approximation IMHO is that you place assumptions around how to "make in tech". "Not participating == not interested" assumes that there is some universally accessible and well-known path to working in tech.
Now, if you are saying that you've taken advantage of all of those events/opportunities (from your OP) because one day you magically became interested without implicit/explicit involvement from role models, loved ones, inspirational teachers, etc. then I will concede your point... but that would make you exceptional IMHO.
>> Where I can agree is that there are basic opportunities (e.g. networking) in the tech community that Black Americans may not take interest in.
I look for people who don't do the job for the money but because they enjoy it and they have a passion for the work and technologies. Today I don't work with the tech directly but apply it to a particular domain but I need to keep up-to-date with the technologies and work with them on my laptop.
Many of the meetups I've gone to (including Google Tech talks in NYC) are ultra cool. I don't go for the networking, but because I want to learn about the new technologies and ideas and ask questions. While attending, of course I end up networking and learning more about what other attendees are up to.
In my case, I paid for 90% of my tuition and housing/living costs for undergrad by programming computers beginning in high school. I could not get a penny of loans or grants in financial aid. But because I had so much work experience, my many campus job interviews were about the interesting projects that I'd programmed and I got many offers from great firms.
In summary, attending the Meetups is not about the networking per se, but that the topics are ultra-cool applying different technologies to many domains (finance, healthcare, media, urban planning, etc. etc.) and its fun to meet people with similar passions and interests.
Not only do I not see blacks, but I also do not see hispanics at these events.
Instead of whining about diversity, create a github account, do some interesting projects with spark and machine learning or one of the other spark libraries. Do it with Scala or use Python or R. Solve, some interesting problem with all of the data out there or do some volunteer work while in school to do such a project. Interview with that.
There used to be severe amounts of antisemitism in this country. Even as late as 1997 Donald Trump sued so that Jews and blacks could attend country clubs in Palm Springs.
Richard Feynman wanted to attend Columbia University but was turned away because of the Jewish Quota.
But even with the antisemitism there were always firms who ignored the prejudice and hired on the basis of merit where merit was important. There were never diversity programs.
I believe there are firms in tech and other firms that need good computer tech that hire on the basis of merit and could care less the sex or color of skin of the individual. Who would want to work in a firm that isn't merit based anyway?
But even with the antisemitism there were always firms who ignored the prejudice and hired on the basis of merit where merit was important. There were never diversity programs.
And there were and still are plenty of firms which didn't/don't hire based on merit. Your argument here is basically "oh well, let's just sweep that under the rug and be hopeful that in the future it'll get better". Diversity programs are about taking the norms of the future and starting to implement them now, because telling people to wait another generation, or another century, for the discrimination against them to fade away naturally is not a great thing to do.
Someone who is highly competent, is deserving of merit, never want to work for a firm that doesn't hire on the basis of merit. So those firms that don't hire on the basis of merit are turning these people away. There is a shortage of people with competence and merit in computing, so that person will find a job and with a firm who values that person for their abilities.
Just as the case with antisemitism, no need for diversity programs. Physics Nobelist Richard Feynman was turned away from Columbia University because of the Jewish Quota. He went to MIT instead. MIT wanted someone like him and didn't care whether he was Jewish or not. The fact that MIT accepted him and Columbia did not demonstrated that MIT was the better school.
In a world where every market participant was perfectly informed and perfectly rational, and always acted in perfectly enlightened long-term self-interest, then you might have an argument.
We do not live in such a world. Your Feynman anecdote is a rare exception rather than a norm. Tech companies and software engineers will give up their racial and gender biases only when absolutely forced to, and at the moment nothing in the market is forcing them to -- you might not become the most profitable company in the world, but you can have major bias and still do well enough that it's simply not perceived as a problem for the companies which do it.
>One of the open secrets of working in technology is that technical interviews are completely worthless as a predictor for whether someone is a good hire or not. [...]
>> We looked at tens of thousands of interviews, and everyone who had done the interviews and what what they scored the candidate, and how that person ultimately performed in their job. We found zero relationship.
> It’s amazing to think that Google found zero relationship between an interviewer saying “Hire” and whether the candidate was actually a good hire or not.
The fact [the scores of someone who gets hired] are not predictive of their eventual performance, does not mean that hiring process is completely random, or worthless.
In order to tell that, you would need to look at the cohort of people who were 'no hire' and compare them to the 'hire' cohort; it's not enough to just look within the 'hire' cohort.
In other words, the hiring process is (1) designed to be a Classifier, typically tuned with an emphasis on having a low false positive rate. Hoping that (2) the scores of the small subset of candidates who are classed 'hire', also act well as a regression to the candidate's eventual performance, is a big additional ask.
Just because there is evidence that (2) is untrue, or particularly noisy, does not necessarily mean that (1) is broken.
I don't see a great way of solving this, without a large company like Google also randomly hiring 'no hires' from different stages in their hiring process (and subsequently measuring their performance) in order to quantify the performance of subsequent steps in the hiring process - not aware of anyone who has done that; it would be really interesting to read how a company who does that gets on, but an expensive experiment.