Is Blind Hiring the Best Hiring? | Hacker News Reader

Is Blind Hiring the Best Hiring?

nytimes.com

64 pointstokenadult49 comments

Blind hiring is a good idea. But definitely cannot explain the bulk of the issue. It is systemic and the article completely writes of the idea that there is a pipeline problem - which ignores a lot of the pre-college factors that go into whether or not a student of a minority demographic can succeed in a top tier engineering program in particular (I mention engineering because the article refers to Silicon Valley jobs a lot).

Below are two pieces on the demographics of engineers at Harvard and Stanford: they show pretty clearly that the percent of Black and Hispanic students in these programs lags significantly to the US population percentage.

https://medium.com/@jcueto/race-and-gender-among-computer-sc...

https://medium.com/@winniewu/race-and-gender-among-computer-...

"Stanford Computer Science majors are Asian (46.4%), followed by White (38%), Hispanic (9.5%) and Black (6.1%)."

High school: Nationwide, black students graduated at a rate of 69 percent; Hispanics graduated at 73 percent; whites graduated at a rate of 86 percent.

13.2% of the population is black, 17.1% Latino, 63% white.

If White = (38%) then you would expect black to be: 38% / (.63 * .86) * (13.2 * .69) = 6.39%. So if anything Standford is about right for black people, and massively over represents Asians.

It seems like HS graduation rates are the real issue everything else just flows from that. But, income and graduation rates are linked so it's a self perpetuating cycle.

    > So if anything Standford [sic] is about right for black
    > people, and massively over represents Asians.

That assumes only local applications though. And even then ignores demographic prejudice in where to apply (which is probably more slight).

I would guess Asian students are a majority among international applicants to Stanford. If we then looked at high school 'graduation' worldwide as the baseline, that's probably an underrepresentation.

I am ignoring where people apply, just adjusting for % of population and % of high school graduates. There are a lot of ways to play with the numbers but that seemed like a good first approximation.

Also, international students are 8.70% of undergrads at Stanford, so they can't really change the stats that much.

Asians is a huge group. Would be nice to see a break down between Chinese, Indians and the rest of asia

Stanford and Harvard are top-level universities aiming to enroll (mostly) those who are in top 5-15% of national IQ levels. How would the characteristic of their diversity efforts change if instead of comparing to the US population percentage you start comparing to the population percentage at the right end of the IQ distribution?

Are you really suggesting that underrepresented groups are just less intelligent?

He's suggesting that we should dig up academic papers studying the topic and run the numbers based on the intelligence distributions found therein.

Here are a few review articles to get you started - more detailed studies can be found referenced therein.

http://psychology.uwo.ca/faculty/rushtonpdfs/PPPL1.pdf

https://home.ubalt.edu/tmitch/645/articles/roth%20et%20al%20...

http://psychology.uwo.ca/faculty/rushtonpdfs/PPPL1.pdf

http://matt.colorado.edu/teaching/highcog/fall8/nbbbbchlpsu9...

https://www.udel.edu/educ/gottfredson/reprints/1997mainstrea...

J. Philippe Rushton, who you cite directly twice here, ran The Pioneer Group, an organization that funds both research and public policy premised on the idea that "white persons" are superior to other "races"; it's an organization that was literally founded (prior to World War 2) to promote the eugenics ideas of the Third Reich.

Technically I cite him once, I copy&pasted the same thing twice. Anyway, suppose everything you say is correct; are he and the studies he cites factually wrong?

This reflexive denial of any possibility of group differences by the left is quite dangerous. In the short run denying reality and socially punishing those who acknowledge it works. But in the longer run you alienate people who don't like being lied to. Witness Trumpkins and the rise of the anti-Muslim right in Europe.

Why not just acknowledge reality and make a moral case that isn't based on incorrect facts? E.g.: "we should help refugees who need our help, and ISIS will try and bomb us because of that. We'll bomb them back if they try, but we won't give them what they want and stop helping people who need it."

Also, the research doesn't even support any white superiority claims. On most dimensions - intelligence, criminal propensity, conscientiousness - it suggests that Ashkenazi Jews, Chinese and Indians are superior to white people. Whoops, looks like your attempt to conflate human intelligence research with Nazis kind of backfired in a comical way.

Hold on. I'm not answering anything else in this comment until you acknowledge that opposition to American neo-Nazism isn't "leftism". I have plenty of conservative friends, and am interested enough in conservative ideas that half of HN seems to think I work for the GOP. None of them would share a room with Rushton. The first person you cited in this discussion helmed an organization whose charter was to create an American Lebensborn system. Not, like, "something that would effectively function like the Lebensborn", but literally take the template of the Lebensborn and apply it America.

I acknowledge this.

Now let me note that I linked to some studies, and rather than engage with their content, you just compared a guy to the Nazis for citing work that contradicts Nazi ideology without even discussing whether it is correct or incorrect.

Let me also point out that you know my general background; spend large amounts of time in Asia, advocate open borders vs nativism, have slept with 2 white ladies in the past 5 years (one Jewish), and argued against white superiority in the very thread you are replying to. Do you really think I'm pushing some sort of pro-Nazi position?

I don't have time to fully answer this right this second but I want to duck in real quick and say that I don't think you're pro-Nazi --- but please find a better way to say that than "I've only slept with...".

Seriously? Ok, first of all, why would universities use IQ as an enrollment measure? IQ is a general (and contested) measure of potential, not of effort, accomplishment, interpersonal skills, or any of the many other qualities that make a person successful in college and later in life. That's why the admissions process uses a combination of test scores, GPA, essays, and interviews. If you just enroll the top 5% IQ students you could easily end up with a class full of intelligent slackers who fail out in a year because everything in high school was easy.

Second, IQ and race is a highly studied area and the differences are likely to do with many of the exact same factors that lead to socioeconomic inequality between the races. Remember: IQ is correlated with general intelligence, but there are many reasons people do better or worse on IQ tests beyond that. You should really read the Wikipedia page on the subject and take a look at some of the cited research: https://en.wikipedia.org/wiki/Race_and_intelligence

Lastly, to answer your question, if universities just wanted the highest IQ students they would overlook American students entirely and fill their classes with East Asian, and maybe European students instead.

This is true. Various studies have been done which measure how much of college GPA and other such things can be predicted based on academic inputs.

http://ftp.iza.org/dp8733.pdf

http://www.mindingthecampus.org/2010/09/the_underperformance...

It turns out that "underprivileged" people actually do worse than what IQ and other academic measurements alone would predict.

If universities actually wanted to accurately choose the students most likely to succeed, they would penalize blacks, hispanics, males and people with low family income.

Second, IQ and race is a highly studied area and the differences are likely to do with many of the exact same factors that lead to socioeconomic inequality between the races.

If this is true, you should be able to run a multiple regression analysis which takes into account those "exact same factors" as well as race, and which accurately predicts outcomes. Further, the coefficient on race should be close to zero.

(This is the exact opposite of what is found in the studies I link above.)

Can you cite a study demonstrating this?

Did you read the link I provided? It covers exactly what you're asking for with multiple references.

The wikipedia page agrees with what I said: Different aspects of the Socioeconomic environment in which children are raised have been shown to correlate with part of the IQ gap, but they do not account for the entire gap.[79] Generally the difference between mean test scores of blacks and whites is not eliminated when individuals and groups are matched on SES...

I haven't read every study cited there, though I've read a few. Which ones (you claim multiple) do you believe provide what I'm asking for?

What are you talking about? That was about half of the article.

> A new study from Harvard Business School backs up this line of reasoning. It found that when service-sector employers used a job test, they hired workers who tended to stay at the job longer — indicating that they were a better match. When employers overruled the test results to hire someone for more subjective reasons, the employees were significantly more likely to quit or be fired.

That's an interesting finding considering that standardized tests are often criticized for not promoting diversity. Ive read that standardized tests were originally intended to combat bias, and i think they are fairly successful in that regard. It's a shame that they have fallen out of political favor in favor of looking for "the right type" of subjective qualities of candidates.

Standardized tests are rarely job tests. Non-job-specific standardized tests may eliminate bias with regard to what they are designed to measure, but the choice of such a test itself introduces bias to the extent that the test favors one group over another in ways which do not reflect the job for which it is given, even if it is unbiased in what it is intended to measure. (This is the basic reason behind the case that is often mispresented as outlawing standardized IQ tests in hiring, which did not actually do that but instead required that, to the extent that such tests disproportionately disqualified members of a protected class, that there be evidence that the test was a real measure of performance on the job for which it was used as a hiring filter.)

Yes, no test is perfect but I think its better than the alternative. The best part is that its fair in the sense that the person knows what to expect. A lot of people feel disillusioned about processes that have a subjective element, so much so that they don't participate. If I were told the precise steps that all but guarantee an outcome, that person would likely find that preferable to the alternative. It would build trust into the system, even though the objective methods may be too rigid at times.

Businesses have all the incentive to create tests that measure real performance. These kinds of problems are best solved in the highly competitive market place for talent. I don't think some politician would do a better job at determining what makes a good employee and how to hire that employee.

I've read somewhere that the push for "well-rounded" candidates in colleges was pushed partly to stop the flow of very strong asian candidates displacing others. That's why schools that rely primarily on tests, like the elite Stuyvesant high school, end up with disproportionately asian students (Stuyvesant is 72% Asian). They get criticized for some reason due to the outcome of their selection method. Perhaps going to a school with predominately Asians has downsides for the student, but that's really a judgement call that should be left up to the parents. And the rationales normally used to prove racial bias for a test doesn't really hold water since the groups doing well are usually first or second generation immigrants. I think rigging the system in favor of one group is pretty bad, but rigging it just so that its harder for another group may be even worse.

This is getting ridiculous. Standardized test is ONLY test that is fair. The very first moment people will get different tests based on their race will be moment when racism started and will shortly follow by employer discrimination.

I had a customer-facing programming job (something like a technical consultant). I had to speak English understandably -- even over the phone -- and be a polished, likeable, and competent representative for the company.

It's hard to imagine how this could be done without "flawed human judgement".

Some jobs might be form-filling or widget-building in a dark basement, and you might be able to get away with this. But a lot of jobs are not that.

The rules for sales are... different, but sales (and to a lesser extent, consulting, at least the "make the customer feel special" part of consulting) exists in what most of us consider to be a kind of ethical gray area. You're trying to tell the customer "I'm one of you"

Most technical jobs are not like that. Most have pretty minimal social requirements. If you can't work with a programmer who doesn't look like you, I'd argue that the problem is with you, not with the person who looks different.

Every time I argue that competency should be more important than 'cultural fit' people bring up the example of the asshole. But the asshole isn't usually the one rejected for 'cultural fit' - assholes usually do fairly well during the interview.

And there are technical ways to find the asshole. Send them some bad code for review and see how they review it; ask them to work with the author to get it merged. (Enough code review happens in open-source projects over email/GitHub, over text, with pseudonyms, that having it be anonymous is entirely realistic.)

Alternatively, keep an eye out for asshole behavior -- to any employee, not just those on the same team -- during the first few weeks, and be willing to fire.

I'm not talking about you, per se, but about hiring for such a job. One could justify a face-to-face interview for a job like that, but if blind hiring became commonplace, you might have to settle for second best in terms of technical ability. This is already typical of technical sales jobs.

And second best in terms of sales ;)

As the article says, the first rounds of interviewing (seeing if you have the technical skills to consult) can be done without seeing you. That sort of thing routinely involves human judgment, but does not need it. Your personability, which requires human judgment, can be measured afterwards.

It seems to me that the same argument could be made against blind orchestra auditions. You could potentially play your instrument amazingly well, but also be unpresentable on stage. Maybe you refuse to wear clothes that fit. Maybe you pick your nose during rests. Maybe you shake your leg subconsciously. If those are real problems, they can be found after hearing a good audition, and they won't allow unconscious bias to interpret the human judgment of whether the audition is in fact good.

I once had an amazing experience with an organization who explained on its website that you should remove your ''personal information'' from your resume as they do not select based on name, gender, age or race. I did not get hired, but to this day I still recommend that organization to my friends!

Would you mind recommending it to us?

As the article notes, software engineering is an interesting discipline in that an engineer's main responsibility is to produce a discrete body of work which can be peer reviewed. I don't want to dismiss culture entirely, but it's a very vague idea that gets used to justify a lot of negative hiring practices. I think it's less important to assess "culture fit" than it is to do a very thorough review of a candidate's prior work. The candidate can be included in this review and even asked to explain/defend particular choices they made while creating that work -- since sometimes external constraints force non-ideal technical decisions and if the candidate understands that it's a sign they're a strong engineer.

Why not hire the best engineers you can find and then let them define the culture of the company? People who have a hand in defining the culture of their team/company tend to be more motivated and invested in it.

But their prior work is their cultural fit. Has this person worked for F100s, startups, in academia, as a contractor, etc etc etc. Your test would rule out anyone looking to change direction regardless of the reasons.

Because culture does matter, even in tech.

Consider a company where social inclusion is important. That'd be a terrible culture fit for me. When debating a technical choice, I'll say "do we all agree to run X experiment, outcome Y => choice A, outcome Z => choice B? Then we can all STFU and stop arguing." In the happy/social/fun environment, this is terrible. The promoters of B might feel unhappy, harming team cohesion, and that could be more important than 10% lower latency.

On the flip side, consider the confrontational "put up or shut up" environment I tend to prefer (and create). In this environment, "you suck at SQL" is just a factual statement that you can choose to remedy or not. Some folks really don't like being told "experiment failed, your idea is wrong, be smarter next time". Even if they are technically competent, they need a form of social interaction at work that they don't get.

There are a variety of cultural dimensions like this that matter. Another is implicit vs explicit - whether you can navigate implicit job responsibilities or they need to be spelled out. Much as we'd like to act like we are just machines taking money as input and code as outputs, this stuff does matter.

How about "the experiment failed, your idea was a good hypothesis, better luck next time"? Instead of "you suck at SQL", how about "you should probably read up on the difference between GIN and GIST indexes in Postgres and maybe try out window functions - they make this problem a lot easier to solve". It seems more productive to give people ways they can improve their skills rather than tearing them to shreds just for entertainment.

You can have an environment that is driven by good engineering (testing assumptions, measuring etc.) without people being outright insulting to one another. All it requires is, oh, minimal human empathy and a willingness to spend 10-15 minutes every so often explaining unfamiliar concepts to fellow team members who haven't fully grasped it.

It seems more productive to give people ways they can improve their skills rather than tearing them to shreds just for entertainment.

That's a valid elaboration focused on a single specific instance.

It doesn't mean that "you suck at SQL" isn't also a valid critique. For example, while GIN/GIST indexes will help them solve this problem, what that person may need is to actually just learn a lot more about SQL overall (rather than just one specific trick). "You suck at SQL" isn't a thing that's valid (read: correct) to say to someone just because they don't know about a window function. It's a thing that is valid to say when a person repeatedly demonstrates fundamental misunderstandings of SQL on many occasions.

It's also useful information. I've been told that I suck at UI, repeatedly. I accept and internalize this fact. As a result, while I do sometimes think about customer experience, I very rarely push hard for a specific UI or worry too much about details. I'll suggest that maybe a UI is confusing and they should hallway test it and then afterwards do nothing but insist that we abide by the test results (whatever they were).

Based on the mood affiliation of your comment, it sounds like you dislike this sort of interaction. You prefer something with more "empathy" and less "tearing them to shreds". That's exactly what I'm talking about when I say culture matters. You'd probably hate being on my team and I'd probably hate being on yours, even if we are both highly competent devs.

>That's exactly what I'm talking about when I say culture matters. You'd probably hate being on my team and I'd probably hate being on yours, even if we are both highly competent devs.

This shouldn't be a matter of team culture. Socially mature people should be able to adapt their communication to different people's personalities without any difficulty. There's no reason you can't say "Hey Fred, your foo implementation sucks." in one breath, and "Hey Bob, this bar implementation might be a bit better if you looked into baz." in the next. And I'm not sure why you find the latter so unpleasant.

There is also no reason why Bob can't adapt his communication to whoever is saying "your foo implementation sucks".

Culture is a set of prevailing norms about which group of people are the ones who should adapt. In some cultures it's Fred who adapts, in others it's Bob.

Let me emphasize that I'm not advocating that one culture is superior. Bridgewater is a highly successful "you suck here's why" culture, Shutterstock is a highly successful "lets all think about feelings" culture (to name two NY companies known for culture), and I don't think either would be improved by making them more like the other.

I'd just be far more willing to work at Bridgewater (at least if they opened a NYC office - who wants to go to CT) than at Shutterstock.

And I'm not sure why you find the latter so unpleasant.

I find it to be an exercise which saps my mental energy and distracts me from increasing shareholder value. The less I think about feelings, the more I'm thinking about getting more sales for my customers.

I'm not sure why some people find open office plans so unpleasant. But I accept that some people really don't like them and I have no problem if companies choose private offices to appeal to those people.

> There is also no reason why Bob can't adapt his communication to whoever is saying "your foo implementation sucks".

Except that it's a lot easier to not say "your work sucks" than it is to not feel upset when someone does. And most people will indeed adapt over time, so it's a pretty silly reason not to hire someone.

>I find it to be an exercise which saps my mental energy and distracts me from increasing shareholder value.

The thing is, I don't feel like I need to actively think about people's feelings to not say "this sucks". I just do it naturally in response to different social contexts.

It's easier for you. For me the reverse is true.

One reason to avoid hiring people who don't fit the culture is to avoid entryism. If a group allows just anyone to enter, the newcomers may gain power and destroy the original culture. This is particularly true if a culture is significantly outside the norm.

For example, witness Github's transformation from flat hierarchy and meritocracy to traditional corporate hierarchical politics: http://danieltenner.com/2016/02/06/github-the-quiet-death-of...

The thing is, I don't feel like I need to actively think about people's feelings to not say "this sucks". I just do it naturally in response to different social contexts.

I don't need to actively suppress negative feelings when people give me factual feedback. It just happens naturally. People are different.

If I'm interpreting your posts right, it sounds like you feel your personality profile should be privileged over others and that all organizations should adapt to suit you (leaving me unable to find a place that suits me). Perhaps I'm wildly misreading you and interpreting differences in mood affiliation for differences in beliefs? Could you clarify what you are advocating for?

>It's easier for you. For me the reverse is true.

What I mean is it's easier for you not to say "sucks" than it is for them not to feel upset. Speech is active, but emotions are passive.

>I don't need to actively suppress negative feelings when people give me factual feedback.

I have no problem with merely "factual" feedback. What you're talking about goes beyond the facts to put a negative and personal spin on it. Moreover, it suppresses the actual facts ("this SQL is bad because...") in favour of that negativity.

>If I'm interpreting your posts right, it sounds like you feel your personality profile should be privileged over others and that all organizations should adapt to suit you

I think the personality trait of "upset by strongly worded negative language" is incredibly common, so being able to communicate with people who share it is part of basic social literacy. To me, your saying it's too much effort sounds like those people who say using correct spelling and grammar is too much effort. If you can't do it without thinking, you should learn, because it's a widely applicable skill.

It's not easy for me to actively model the mental states of others at all times and tailor my words in order to prevent those states from entering negative territory. I can do it, but it's not as simple as you imply.

Moreover, it suppresses the actual facts ("this SQL is bad because...") in favour of that negativity.

Or it suppresses the actual facts because I didn't feel like writing a longwinded example. "Too few examples" and "too little detail" is a critique people rarely make of me.

I have learned - anyone outside the mainstream has to. But I think you underestimate the effort that it takes to adapt to a culture or way of thinking alien to your own.

But that's not the question here. The question is whether it's acceptable for a group of people with non-mainstream tastes to build and protect a culture where their preferred modes of interaction are the default, and to prevent others who might change that culture from entering. I get the impression you are saying it isn't. Is my impression correct?

I believe you are probably right that 80% of people think like you and 5% think like me. I don't believe that this implies that 100% of organizations should cater to you. On the contrary I think if you don't like my organization with a niche culture, it should be very easy to find a more feelings-driven organization and shift to it.

> I'm not sure why some people find open office plans so unpleasant.

Funnily enough, I recently outlined the reasons here (applicable specifically to those with hearing loss as well as everybody else), with reference to a whole range of scientific studies. https://tommorris.org/posts/9403

> It doesn't mean that "you suck at SQL" isn't also a valid critique.

Completely disagree. "You need to learn more about SQL" communicates the same information without the gratuitously negative tone.

> You'd probably hate being on my team and I'd probably hate being on yours

It's pretty clear that I would hate being on yours, but I'm not so sure you'd hate being on mine. It is possible to communicate directly and effectively about technical matters while still being courteous and professional. As team lead I strive to set this tone, in large part by being open to criticism about my own work.

>In the happy/social/fun environment, this is terrible. The promoters of B might feel unhappy, harming team cohesion, and that could be more important than 10% lower latency.

I don't think anyone competent wants to be on a team that makes bad technical decisions for the sake of people's egos. If someone has hard evidence that choice A is better, but is unable to communicate that in a way that doesn't harm team cohesion, then I'm inclined to say they must have pretty poor social skills.

All you have to say is something like "I understand your reasons for expecting choice B to be better, but I'm afraid this evidence shows otherwise." instead of "experiment failed, your idea is wrong, be smarter next time". Especially since, if you needed an experiment to work out the correct solution, then predicting the wrong outcome can't have been particular stupid.

>In this environment, "you suck at SQL" is just a factual statement that you can choose to remedy or not.

Maybe it's "factual" in some sense, but aside from being unnecessarily harsh, it's also pretty vague. A more useful statement is "Your SQL experience doesn't seem to be sufficient for this task". (EDIT: What I'd actually do first in this situation is ask them how confident they felt in SQL, giving them chance to save face by voicing the criticism themselves.) Of course ideally, you have an environment where this person felt comfortable saying "I don't have much SQL experience" at the time they were given the task.

EDIT: Also, to be clear, I don't think everyone has to talk like this. If two people in an office are happy to trash talk each other, that's fine, just as long as they can change gears when talking to less confrontational people.

Haven't we already decided that the answer is definitively no? That the "rockstar" developer who is also a complete prick is a bad hire?

It really, really depends. If you've got a large team and a mature codebase, a rockstar programmer who's a douche is usually worse than useless.

But if you've got a small team and are making major architecting decisions, the competent, meticulous, anal-retentive jerk can be perfect. A good manager knows how to get the best work out of everyone, and managing different/difficult personalities is an important skill.

That's fine as long as you're happy having only one person at that level of competence. You're not going to get a second one to stay long.

Relevant paper: Unintended Effects of Anonymous Resumes http://ftp.iza.org/dp8517.pdf

I am pro blind hiring. You should get hired based on your skills and abilities, not your personal background.

Reply on news.ycombinator.com