Hacker News new | ask | show | jobs
by brailsafe 2175 days ago
Afaik, research suggests that interviewers know after 20 seconds whether they'll hire someone. I don't have references and don't know if this would translate well to tech interviewing, but I'd wager that the vast majority of a 7h interview day is totally pointless.
5 comments

> "...interviewers know after 20 seconds whether they'll hire someone..."

Without even seeing a single like of code from the candidate? That definitely doesn't translate to tech interviewing. It takes at least 30 minutes to get enough code out of the candidate to form even a first impression.

> It takes at least 30 minutes to get enough code out of the candidate to form even a first impression.

Yeah, gonna stop you there. It takes less than 30 seconds to get a first impression. That's the whole point of first impressions. As soon as you see a person, you have a first impression.

I've talked to many interviewers - they all agree that they basically have an impression formed within the first few minutes and then use the interview to solidify that impression. It is wildly uncommon for them to come in, have a very negative first impression, and then have it completely turned around by the end.

Grandop said

> research suggests that interviewers know after 20 seconds whether they'll hire someone.

That’s not first impression, that’s a meaningful ability to see if I hire someone. I can tell you that without seeing code/discussing some previous experience there is zero chance I can form hiring decision for most reasonable candidate (the only 30 second hiring impression I can form is perhaps a negative one).

I meant to be less specific than "hire someone", but I'd wager that once a subconscious impression is formed, you'd be more likely to bias your interpretation of whatever else they say.
Right, I am sure I have a first impression, which could be good or bad, and that impression does inform my later views. However, just objectively I know for a fact that I declined some candidates I initially liked and I hired a few people who initially was super shy/not communicative.
This has definitely not been my experience when I do tech interviewing. Lots of people can talk the talk...
I believe it was more of subconscious thing, but again, just throwing it in. Might have been SciShow that remarked on it.
I wonder how much of that is like-me bias with the risk that 3 short interviews might get us (even) more homogenous employee populations?
I rarely have encountered an employer that is truly - actively - trying to avoid like-me bias. If they were even remotely successful, your interviews wouldn't almost always be constructed of 3 races... and various teams at orgs wouldn't have such wildly lopsided demographics.
Wouldn’t entire elimination of irrelevant bias mean that teams would have wildly varying demographics?

If I have 16 teams of 4 people each, I should probably expect to find an all female team and an all male team, if we assume that gender has no bearing on individual performance or on team selection and that gender bias was entirely removed, leaving the gender of each team member to a coin flip?

Similarly, I’d expect to find a fair number of teams who all had O+ blood type today, as I think we do not discriminate on that basis.

Team size is going to be important to note. The teams I'm talking about are frequently near 10 or more members.

As your point becomes more important as the size gets smaller. If it were all teams of 1 then each time would have incredible bias...

> I wonder how much of that is like-me bias

I am just crawling out of 4 months of hell of interviews and homeworks. And I can tell you: probably 90% of it. :(

How can I appear more like my interviewer in those first 20 seconds?
20 seconds? Nonsense.

Thirty minutes is usually enough to form an opinion, but the purpose of the other thirty is to gather supporting evidence, so that I can justify my hire/no hire decision.

If you are letting your interview candidates bomb their interview in less than 15 minutes, your expectations for them are highly unrealistic.

Yeeeahhhhh...

Which shows that bias is 99% of that thought process.