Hacker News Reader

Comment by Danihan | Hacker News Reader

I believe that treating everyone as individuals, rather than as stereotypical groups, is the only way forward. It's the only truly fair approach.

What ever happened to the notion of being color-blind when it comes to policy enforcement? AKA, actually treating people equally, based on merit?

If biases are really that big of an issue (are there studies that show this is true in tech?) then what is wrong with "blind-hiring," instead of the current "diversity-conscious" hiring? You don't have to get to know someone's personality at a deep level to make a hiring decision, you need to know their skill level and aptitude.

It worked to remove the gender gap in orchestras. Why wouldn't it be good to use in tech?

http://gap.hks.harvard.edu/orchestrating-impartiality-impact....

View on HN

You don't have to get to know someone's personality at a deep level to make a hiring decision, you need to know their skill level and aptitude.

I agree, but how do you estimate their aptitude in an unbiased way? That mostly rules out face-to-face conversations, which is what most companies use.

Aptitude tests? I feel like those have a bad reputation, at least in Bay Area tech companies. Are there good tests we should be using? How do you customize the test to fit your own company? To the extent that "cultural fit" is important for effective teams (and isn't simply a way of excluding women, black people, etc) how do you test for that?

I personally believe including "cultural fit" in hiring decisions is introducing massive amounts of bias, almost by definition.

Aptitude can be figured out by something as simple as SAT or GMAT scores. If universities use those test scores, why shouldn't employers?

Skill level is determined by doing tasks very similar to the ones that will be given at the job. You know, like reversing red / black binary trees in memory. ;p

Mainly because those tests aren't really a good measure of aptitude, and minoriies and women tend to (at least historically, idk about now) score lower. So you would end up with a company full of white, Asian, and Indian guys. (Not making a judgement here just pointing it out)

You could also use the test as a filter mechanism, but I'd just not take the test unless you paid for it as the recruiter. Even then they require months of serious study for most people. It just doesn't work well overall. Take home "work samples" tend to be the preferred method right now and they seem ok as long as they aren't abused.

How about an artificially anonymized process. HR does know, or assumes, your gender from receiving your CV. But from that point onward your identity is anonymous.

For instance, HR creates a throw away email which will be used during the hiring process to coordinate the rest of the tasks. The coding interviews could use one of the many platforms we have for shared/same space coding with an added chat box for talking your way through the problem so to say.

And so hiring decisions are done by interviewers without knowing the gender or how the candidate looks.

I think this would allow for a good level of blind testing, but would provide some downsides in the side of cultural fit screening. It's a lot easier to pretend you're not an asshole in asynchronous text-only communications.

I guess everything carries a trade off.

> And so hiring decisions are done by interviewers without knowing the gender or how the candidate looks.

It might make the situation worse: http://www.abc.net.au/news/2017-06-30/bilnd-recruitment-tria...

It's very sad that "free from sexist, racist, ageist biases" is considered "worse", surely?

Do you agree?

Yes, I agree (check my comment history over the last few days to see which side I fall on).

I meant worse for the problem the GP was trying to solve.

The method (free from sexist…) is better but the outcome (employee similarity to general population) is worse.

I guess it really comes down to the ethical framework you accept as valid, deontological or utilitarian...

> If biases are really that big of an issue (are there studies that show this is true in tech?)

I don't know if there are studies, but I absolutely know toxic and unfriendly environments exist. I don't know how you'd quantify the effect; if you made up some metric where you looked at how many women COULD be in tech, there's huge lost productivity, but that's not necessarily meaningful.

Even clearing the hiring hurdle is not nearly enough. Hiring someone who your culture treats like crap is not going to help you or them. If the person is actually very competent, but consistently treated as a newbie, their work will be sub-par and they will burn out and leave.

It turns out you need managers who can actually see people, how they interact, and manage them on a personal level. Set up mentoring for those who need it, put people who like to work alone tasks that can be handled alone, people who like to be on big teams on big teams, etc.

There's no way to exhaustively list the things you could do, and that's the point - it's a big, hard job that is a job, that I think SV too often wishes didn't exist.

Doing blind hiring for software is really, really, hard. It's unsurprising that most interviewing pipelines end up being conversations + some whiteboard coding, because coming up with something standardized, systematic, and as blind to biases as possible is something we really haven't figured out yet.

>What ever happened to the notion of being color-blind when it comes to policy enforcement?

A few people didn't have the self control to pull it off, hired a bunch of white dudes, fired a bunch of "diverse" people and basically showed favoritism to people like themselves (i.e. mostly white men) which increased the disparity. After awhile this pattern became obvious and HR departments instituted quantitative policies because doing it poorly in a legally defensible way is better than trying to do it well at a risk of doing it poorly enough to get sued.

Reply on news.ycombinator.com