Hacker News Reader

Comment by _m8fo | Hacker News Reader

Couple points:

First, this is where the meat is: https://www.dol.gov/newsroom/releases/ofccp/ofccp20170118-0

Second:

"...Oracle nevertheless preferred Asian applications over other qualified applicants in the Professional Technical 1, Individual Contributor Job group and in the Product Development job group at statistically significant rates." [1]

[1] https://www.dol.gov/sites/default/files/newsroom/newsrelease...

---

Perhaps I'm naive, but I'll say it again and again until someone with influence hears me: large companies should do anonymous interviewing. I've interviewed with Oracle, Cisco, and many of the "old corporate-y" companies. There's ZERO reason the interview process can't be completely anonymized. Their interviews (from my limited experience) are completely impersonal and done on an ad-hoc basis anyway.

That being said, it seems this issue may be more of an H1B1 issue, which inherently cannot be made anonymous.

View on HN

When you say anonymous, how do you mean it? As in: the interviewer meets face-to-face without a resume in front of them, or the interviewer has a resume but the interview is done over the phone, or a phone interview with no resume?

Well, for one, you can hide contact info from a resume (which is still stored in a system somewhere, with a unique ID).

And you can also make the early stages of your interview email or text-chat based (or just a Google Doc, like some companies do), to eliminate any potential bias in things that can be inferred from voice like accent, region, and gender.

Obviously once they're on-site for an interview, you can't anonymize much anymore, but anonymizing every stage before that seems like a win-win to me (excluding potential costs of implementing the system).

Well, I can tell the way sentences are formed in many cases if I'm dealing with a native speaker or not.

Would you kindly do the willingly and explain how? :)

Explaining the human brain's ability to pattern match will take more than a few sentences.

It's pretty easy, but also pretty rough - if you're good, you can certainly distinguish between an American, a British or an Indian English-speaking person writing.

I could be missing the mark here, but I'm pretty sure you were replying to satire. Based on the grammar used and smiley face.

Maybe add </s> next time, although the smiley should have been a dead giveaway.

You mean do the needful?

But whether someone is a native speaker is a trait that someone can legally and rightfully discriminate over. It's perfectly reasonable to reject someone if you think their English skills aren't sufficiently good. It's not reasonable or legal to reject them just because they were born in a different country.

If you anonymize it and make it completely chat based how do you make sure that the person you are chatting with is really the applicant and not someone else?

You'd need a third party that you both trust. If you both don't trust anyone, there are ways around that too...

Blockchain for hiring/interviews?

Blockchain + machine learning

http://www.experian.com/decision-analytics/identity-and-frau...

It's difficult. I remember coming out of school, employers tried to anonymize for various things, especially with respect to things like race. Black students would, in an anonymized resume with contact info removed for example, state that they were "Treasurer for the Black Students Association" or "NAACP Volunteer", or things to that effect.

My point is that I think that an Asian person, applying to a company known to be favorable toward Asians in hiring, would signal their ethnicity somehow in their resume. At least, that would be the smart play for that person.

Build a loop into the workflow for "You're insufficiently anonymous, resubmit", with a special way to enter "significant award/position/membership with deanonymizing title, X level of prestige/selectivity/membership"? Might be valuable anyway so you don't accidentally get peoples' personal biases into play - different people will weight "NAACP award" very differently.

I find it pretty easy to distinguish between people educated in the US and those not educated in the US after a fairly small sample of writing.

Interesting. Any telltale signs? I'm assuming it's not the difference between native English speaker and ESL, but specific words or grammar patterns? Can you tell from my comment history whether I was educated in the US?

I spent <5 minutes on it. I'd guess that you're not US educated (or at least not a native US English speaker).

The "tell" that I'm primarily relying upon is that you often drop articles where a US-native speaker would use them. Just one example: https://news.ycombinator.com/item?id=13197570 (that was as far back as I went) I think most native US speakers would use "the gun industry" and "the film industry". Several later comments drop an "an" or "a" article as well.

You also have a habit of using a comma splice rather than a semi-colon to join sentences, but that one is harder to place geographically, as it's a somewhat common trait within the US as well. In my experience, it is often a trait of European native speakers who learned English as a second language, but is less reliable than dropped articles.

Correct on both counts, although it's not quite clear how much of the difference is from place of education rather than native language. Math is the same no matter where you study it, at least in theory. Probably need a bigger sample.

Dropped articles is a good indicator of (most) Slavic language speakers, but not of most European languages. E.g., French would not have dropped those definite articles either.

Definite article rules do vary a little bit, though, so extra definite articles would be a good indicator of European native speakers...

That was an incredibly obvious example. My favorite telltale sign for native English speakers is the rampant misspelling of trivialities like it(')s, you(')r(e), etc. I've seen so many well educated peers fail at basic grammar, or misspelling common words (e.g. electrisity, from a CS undergrad) that I'm comfortable making this generalization.

The question was not whether I'm a native English speaker (obviously not), but whether I was educated in the US. I believe the latter is a much harder question.

My favorite sign for native English speakers is "could of" instead of "could have"/"could've".

I have never witnessed that mistake with non native English speakers.

communication skills and personality can be really integral to the hiring process. many would rather have a good communicator than a slightly better engineer. your way removes this.

I think they're referring to something like the blind auditions orchestras used to deal with their own gender disparity issues.

Native English speakers would then need to fake an "asian" accent?

I applied at Treehouse learning and they have a take on 'anonymous' hiring. It's 'anonymous' in that they don't see the resume you submit, or so they claim, until after reviewing short essay answers to some questions.

No knowledge of tools used, name, DOB, companies you have worked at previously, is where Treehouse starts in assessing your application materials.

These were silly simple questions, like assign an empty array to a variable in Python, then append N items to it, and print it on the screen. No FizzBuzz at this stage anyway.

More like "describe how you would handle situation X" or "walk through how you'd solve this real problem we recently saw, given this information".

My intuition suggests they are assessing communication skill first. stuff like tone and clarity in writing. I can't say if that's accurate or they were looking for a "right answer" type response, however.

FWIW, it was actually the easiest application process I've been through. Something about the anonymous part made me feel more relaxed. Like I'll be judged on my own words, not a bullet list of skills. That said, didn't hear from them. :-/

We built voice modulation to mask gender in technical interviews.

https://blog.interviewing.io/we-built-voice-modulation-to-ma...

Your project is something I had in mind as a potential idea for a long time, and the fact that you went as far as doing that is insane, in the best possible way. Kudos, guess you've just gained a new newsletter subscriber.

I hope your anonymous interview system can authenticate applicants properly. Because we get cheating attempts in the online screening test we use for hiring. Recently, we got five submissions from recent grads of universities around the country. They were modified trivially (with different indentation and variable names) but clearly drawn from the same work. There is no way of knowing if someone was hired to do the test, and didn't share the results. We have also gotten people in on-site interviews who were not the same as the person in the phone interview.

"...Oracle nevertheless preferred Asian applications...at statistically significant rates."

I would be very wary about what counts as 'statistically significant'. If I flip a coin 10 times and it lands heads 10 times, is that 'statistically significant'? Not if I'm going through a jar with 1000 coins.

And splitting Oracle's workforce into 'job groups' makes it much easier to find spurious rates of discrimination. Eg. there is discrimination in the 'Professional Technical 1' group but not the 'Professional Technical 2' group...etc.

(This discussion reminds me of this Dilbert cartoon: http://dilbert.com/strip/2001-10-25 )

Yes, or the xkcd with the jelly beans.

Government civil service was like this. You'd take a test and the top 3 scores were hireable.

There are plenty of affirmative action organizations that would be very upset with anonymization of applicants.

Out of curiosity: do you know of any organizations that have voiced such opposition?

No, because I don't know of anyone who has made a serious suggestion that such anonymization be carried out.

There are plenty of advocacy organizations that badger, and attempt to shame, companies into buying their "consulting" services. Services that consist primarily of coming in, counting your employees by the hue of their skin, or whether their genitals are innies or outies. Then, after you've paid them stupid amounts of money, they provide a seal of approval. See Project Include or the Rainbow Coalition.

One reason why I believe it won't matter is such: Companies require a real name. Most require social profiles or a website. If you refuse to give your real name and supporting materials, you will be passed over.

Of course you don't submit an anonymous résumé; companies handle this. HR departments can have specific people outside the selection process hold the table matching candidates' realnames and candidate ID #.

I've never heard of a company that requires a social network profile or website.

Technically, they don't even need a name or any other personal info until they're ready to make an offer.

No reason, I agree! This sounds like a website some startups can design to automate the interview process and anonymize candidates info so the employer can't be biased.

I'm sure that even conversations can be encrypted so the job seeker and the interviewer can see, but nobody else can.

Just make a Chat web client based on the FTP servers or whatever works for everyone.

This would be good for those of us who are 40 and over and have a lot of experience and skills.

But how would you guarantee to meet the quotas with anonymous interviewing? There was for example a study showing women did worse in (phone) interviews when they were masked as men (digitally replacing the voice). So the assumption that everybody would do equally well with anonymous interviewing seems to be wrong.

Reply on news.ycombinator.com