Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

If you care about diversity, don't just hire from the same five schools

blog.interviewing.io

Post reply on HN
Print view View original

371–380 of 496 posts

Re: If you care about diversity, don't just hire from the same five schools

#371

Post by fareesh »

lutorm wrote: post #240
fareesh wrote: post #203

As an outsider in another country, my impression seems to be that the quest for diversity seems to have turned into something of a religion in several cultures. In principle, as a solution to the problem I am principally all in favor of anonymous, faceless, no alma mater preference etc. interviews and application processes, and/or whatever other measures are suggested to remove the possibility of bias or unconscious…

The underlying assumption is that, if you average over a large enough population, there should be no underlying consistent difference between subsets selected on gender, race, or class, so a difference of outcome is presumed to be a result of difference of opportunity. The underlying assumption is definitely an assumption, but it seems like a good zeroth-order assumption to me.

Right - like I said, fighting equality of opportunity is worthy and must be pursued relentlessly.

What seems to be happening (the impression I get) is something analogous to this:

I have a game released on two mobile platforms - android and iOS. The iOS players are disproportionately scoring better than the android players.

Some are suggesting that the server is bugged/programmed to give iOS players better scores. I reject this.

Some are saying that the android user experience has conditioned those players to be satisfied with a low score. I reject this.

Some are saying iOS players are superior because they bought an apple product. I reject this.

My opinion that it is likely that the android release has a lot of performance issues which impede the player's ability to get a high score on that platform. I test this out by doing things like getting an android player to play on an iOS device, and checking if their performance improves. I do the reverse and check accordingly too. I do various tests on fps, input lag, etc.

If my experiment fails to show anything conclusive, I should be open to re-evaluating some of the things I rejected, but I refuse to do it out of zealous belief in my hypothesis. The answer must be optimization, I insist.

If my experiment succeeds, I should tweak the android code till the disparity goes away. Instead, I just hardcode +500 to the android scores and pat myself on the back.

There seem to be cases where the diversity disparity is "solved by hardcoding", and other cases where the data doesn't support the hypothesis, but there is an ardent refusal to accept it, as if it would be blasphemous to suggest otherwise. These are pretty much the only things I take issue with.

Re: If you care about diversity, don't just hire from the same five schools

#372

Post by eropple »

Spooky23 wrote: post #307
KGIII wrote: post #195

Earlier quoted context omitted.

I wonder if there is a competitive advantage to be had there? UMaine is a pretty good system. When I retired, I taught math at UMF for a little while, so I got to see the system from the inside. UMF has one of the best teaching programs in the country. They even have a CS path that is judged well. Maybe recruiting at the small State universities is a potential edge?

Please don’t. Our hiring pipeline is almost completely from these schools. You get great talent that big tech companies won’t interview!

And you can underpay them, which is a nontrivial part of my concern.

No insult intended, I would rather they benefit and not you.

Re: If you care about diversity, don't just hire from the same five schools

#373

Post by codeonfire »

drewg123 wrote: post #37

When I was at Google, I was asked if I wanted to make a recruiting trip to my alma mater. I was excited until I found out they were talking about my grad school where I got my MS, and not UB, a large state school where I got my BS. I told them I'd be happy to take a trip to recruit at UB. I got mostly crickets back from that reply. However, I ended up getting signed up for a series where they had a panel of a few HR…

They understand perfectly. They are hoping for rich, well connected kids from rich families that can bring in money and power into the organization. They may have tried some "ordinary" people before and were befuddled when it did not bring in $100 million dollars and several political contacts.

Re: If you care about diversity, don't just hire from the same five schools

#374

Post by lutorm »

dsacco wrote: post #214
zulrah wrote: post #185

If you care about diversity stop interviewing on these stupid algorithmic questions! Getting a job literally depends on buying the cracking the coding interview book and solving these stupid problems. I know some of my course mates who worked on very interesting projects but couldn't get a job because they couldn't white board a tree balancing problem :D

I agree that whiteboard coding interviews that assess mastery of algorithms and data structures with low applicability to the role are not optimal for hiring. However, I don't really agree that this is a "diversity" problem, except perhaps weakly so in that it optimizes against people who don't want to study algorithms for interviews. But interviews by definition optimize against some subset of the general population…

The article touched on this a bit. If you have a nonstandard background, you are less likely to have prepared for these kinds of situations because it's less likely you have parents, mentors, or peers to learn these things from.

Re: If you care about diversity, don't just hire from the same five schools

#375

Post by lutorm »

fareesh wrote: post #371
lutorm wrote: post #240

Earlier quoted context omitted.

The underlying assumption is that, if you average over a large enough population, there should be no underlying consistent difference between subsets selected on gender, race, or class, so a difference of outcome is presumed to be a result of difference of opportunity. The underlying assumption is definitely an assumption, but it seems like a good zeroth-order assumption to me.

Right - like I said, fighting equality of opportunity is worthy and must be pursued relentlessly. What seems to be happening (the impression I get) is something analogous to this: I have a game released on two mobile platforms - android and iOS. The iOS players are disproportionately scoring better than the android players. Some are suggesting that the server is bugged/programmed to give iOS players better scores. I…

If my experiment fails to show anything conclusive, I should be open to re-evaluating some of the things I rejected, but I refuse to do it out of zealous belief in my hypothesis.

You are arguing a straw man. The "experiments" in this case are showing ample evidence of biases, so as long as those biases exist the underlying hypothesis can by definition not be falsified.

Re: If you care about diversity, don't just hire from the same five schools

#376

Post by john_moscow »

fareesh wrote: post #371
lutorm wrote: post #240

Earlier quoted context omitted.

The underlying assumption is that, if you average over a large enough population, there should be no underlying consistent difference between subsets selected on gender, race, or class, so a difference of outcome is presumed to be a result of difference of opportunity. The underlying assumption is definitely an assumption, but it seems like a good zeroth-order assumption to me.

Right - like I said, fighting equality of opportunity is worthy and must be pursued relentlessly. What seems to be happening (the impression I get) is something analogous to this: I have a game released on two mobile platforms - android and iOS. The iOS players are disproportionately scoring better than the android players. Some are suggesting that the server is bugged/programmed to give iOS players better scores. I…

And now assume that your app is competing with another app and has an issue of Android users frustrated about the low scores.

So while you are figuring out heads down in a profiler why your Android users could be more challenged, your competitor simply hardcodes the +500 to the score and makes a big PR announcement how they restored the justice to the previously discriminated users.

Guess who's getting more sales next quarter. Unfortunately that's how politics works.

Re: If you care about diversity, don't just hire from the same five schools

#377

Post by hasenj »

> While enjoying a nice free meal in Harvard Square, he has the opportunity to ask these successful engineers questions about their current work. If he likes the company, all he has to do is accept the company’s standing invitation to interview on campus the next morning.

What? That sounds almost comical. It seems obviously false, but just in case I will ask:

Is that really what happens at Harvard?

> Anthony goes to a state school near the town where he grew up. He is top in his class, as well as a self-taught programmer, having gone above and beyond his coursework to hack together some apps.

Are these "todo list" apps or legit impressive feats? If the former, it almost has no value. If the latter, they will have absolutely no problem getting hired.

Re: If you care about diversity, don't just hire from the same five schools

#378

Post by fareesh »

lutorm wrote: post #375
fareesh wrote: post #371

Earlier quoted context omitted.

Right - like I said, fighting equality of opportunity is worthy and must be pursued relentlessly. What seems to be happening (the impression I get) is something analogous to this: I have a game released on two mobile platforms - android and iOS. The iOS players are disproportionately scoring better than the android players. Some are suggesting that the server is bugged/programmed to give iOS players better scores. I…

If my experiment fails to show anything conclusive, I should be open to re-evaluating some of the things I rejected, but I refuse to do it out of zealous belief in my hypothesis. You are arguing a straw man. The "experiments" in this case are showing ample evidence of biases, so as long as those biases exist the underlying hypothesis can by definition not be falsified.

Perhaps I am not very well read on the subject. My argument was rooted in the attitude I came across when this story was trending (linked at the end of this comment). This research might have been disproved later on, I never followed up on the story. What I was referring to was the readiness with which some were willing to stop pursuing this sort of trial.

My intent wasn't to strawman - if this is in fact non-existent then I concede that point.

http://mobile.abc.net.au/news/2017-06-30/bilnd-recruitment-t...

Re: If you care about diversity, don't just hire from the same five schools

#379

Post by KGIII »

Spooky23 wrote: post #311
gonehome wrote: post #112

Earlier quoted context omitted.

I agree - and this is even more true at state schools than the slightly more selective schools (like RPI that I went to). State schools have people that got into MIT, but didn't go for financial reasons or family reasons. Nobody at RPI got into MIT because if they did they would have gone to MIT (maybe some rare exception exists, but this is generally true). School admission is obviously not a perfect indicator of ab…

Plenty of people decide against MIT or leave early. It’s a pretty hostile environment for many people.

I'm not sure I'd call MIT 'hostile.'

I admit I'm heavily biased, but I'd say it is demanding. If universally demanding excellence is 'hostile,' then I guess I can agree with you.

Re: If you care about diversity, don't just hire from the same five schools

#380

Post by dlwdlw »

Wouldn't this require at least implicit acknowledgement that top tier schools are not diverse? Subsequently all existing people already at companies need to acknowledge this as well. (Since they are from those schools) It would require breaking the illusion that's it's those "others" who are the "real" problem, both by schools and companies.
Post reply on HN