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

Don’t do interviews, do discussions

thinkingthrough.substack.com

Post reply on HN
Print view View original

221–230 of 432 posts

Re: Don’t do interviews, do discussions

#221

Post by Jensson »

vineyardmike wrote: post #214
Consultant32452 wrote: post #210

Earlier quoted context omitted.

This is almost exactly how I hold interviews, except they are usually I've found this technique to be extremely successful. It's possible I may have had some false negatives, but I've never had a false positive. Everyone I've recommended for hiring has been successful.

> I've found this technique to be extremely successful. It's possible I may have had some false negatives, but I've never had a false positive. This is the impossible problem with hiring. Every interviewer wants to minimize false positives (they're expensive!), but every interviewee thinks they're a false negative.

Which is funny since I bet most who thinks they are a "false negative" also would claim to have "impostor syndrome". Anyone with impostor syndrome would view themselves as a true negative.

Re: Don’t do interviews, do discussions

#222

Post by iamsb »

Having unscripted conversations is one of the best way to be swayed by unconscious bias in interviews.

Even though this advice sounds awesome, I will be cautious of putting it into practice without thinking through the bias problem.

I do remember reading multiple research papers on this, but unable to find them at the moment. From anecdote - In the last company I worked in London, only one team (DevOps) did not follow scripted interviews. It was the least diverse team, not just in terms of representation, but in terms of diversity of thought. Most of it was comprised of "tech-bros".

Scripted interviews do not mean you ask through a basket of questions. It just means that you stay within the guardrails of a set of topics and you go through all the topics. With in a topic, you have fair amount of flexibility. For example if you are hiring for a mid level Java programmer your topics may include - Java 8, Testing pyramid, Functional programming, type safety, developer safety(CI/CD/Rollbacks/Code reviews/Pair programming etc), some domain specific knowledge and so on.

Re: Don’t do interviews, do discussions

#223

Post by madmax108 »

Some of these are great tips and I agree wholeheartedly.

However, in personal experience, I've found that this works much better with more experienced developers rather than with junior engineers. Why? Because for some reason, a lot of junior engineers have been pavlov-ed into thinking every interview is for something at Google/Facebook scale (no matter where they are interviewing) and they start describing extremely convoluted designs using tech that they are also not completely familiar with just because they want to come across as knowledgeable.

Something that I struggle with in these cases is reining in the dev back to the "what" rather than the "how", because I've seen even good engineers go into this "let's add a system-bus for everything" way of thinking. Constraining problems explicitly tends to devolve into the "Interviewer-Interviewee information asymmetry" which is the same with most DSA problems (At least with DSA, most constraints are known by both parties).

On the other hand, almost every time I've picked an actual problem we have with a system, be it a bug or a new product or something else, as an interview question with an experienced interviewee, I feel like I've come out understanding the problem space AND solution space better just through the process of discussion and in multiple cases, actually ended up using a lot of ideas from these discussions, so interviewing feels much more "natural" and a "dominant strategy" in game theoretic sense.

Re: Don’t do interviews, do discussions

#224

Post by mark-r »

rsj_hn wrote: post #55
austincheney wrote: post #45

The biggest challenge with software interviews is that you don’t know when to lie. The process is maximally biased and so you have maximum incentive to lie. The only reason to not lie is reputation damage in the highly unlikely case you are caught. In the end you are either hired for more money or you are just wasting your time as a candidate. Most of us really want to be as honest as possible, not just because we ar…

> So, just lie. I think this is bad advice. I have never lied in an interview. I've also never had a job not offered to me if I made it to the in person interview part. This isn't to say that I have magical job-getting powers, but only that not lying has not hurt my chances. In one job I applied for, I didn't have a lot of domain knowledge, but I had knowledge in an adjacent domain and wanted to to jump over to this…

I once failed an interview, and I think it was because the interviewer laid a trap for me. He asked if I would use such-and-so algorithm to solve a particular problem, and I said sure. I had never heard of such-and-so algorithm, but I figured that I could look it up once I had the job. When I didn't get the job, I realized that such-and-so was probably made up by the interviewer to see if I was trying to BS my way into the job. Being honest would have been a better approach.

Re: Don’t do interviews, do discussions

#225

Post by Griffinsauce »

CalRobert wrote: post #26

I've finally reached a point in my career where I have a great paying job and like it well enough, and really don't give a shit what interviewers think. Paradoxically, I think I interview a lot better. I can steer conversation towards stuff I care about, and if they insist on being annoying, just thank them for their time and leave. Though this might just be a result of being pickier about who I interview with. If no…

The best experience I've had while interviewing is when I did not care at all about getting the job. It allows you to be genuine. Not playing the game of trying to show what they want to see but just truly being yourself.

It didn't have anything to do with my career though. It was early, I had an okay position and just encountered something really interesting and exciting.

So perhaps the learning is more about how and when you choose to go explore something that excited you rather than other reasons to find a new job.

Re: Don’t do interviews, do discussions

#227

Post by onion2k »

Consultant32452 wrote: post #210
swman wrote: post #197

I wish I could a 2-3 hour interview where I (or the candidate) showcase one of my projects and explain the architectural details and decisions in addition to showing any cool/hairy/insane code that got the job done. We can discuss those things and see how to improve them, or laugh at the crazy solutions. Honestly how many times do I need to rehearse these dumbass algos (blah blah blah, so I'll optimize for space with…

This is almost exactly how I hold interviews, except they are usually I've found this technique to be extremely successful. It's possible I may have had some false negatives, but I've never had a false positive. Everyone I've recommended for hiring has been successful.

Everyone I've recommended for hiring has been successful.

It's possible you're good at spotting good people, and rejecting bad people, but it's also possible that hiring is just easier than you think it is, and most people are capable of doing the jobs you hire for. You have no way to tell if you'd have the same result just hiring people by picking random resumes. Maybe negatives are just rare.

Re: Don’t do interviews, do discussions

#228

Post by jandrewrogers »

swman wrote: post #197

I wish I could a 2-3 hour interview where I (or the candidate) showcase one of my projects and explain the architectural details and decisions in addition to showing any cool/hairy/insane code that got the job done. We can discuss those things and see how to improve them, or laugh at the crazy solutions. Honestly how many times do I need to rehearse these dumbass algos (blah blah blah, so I'll optimize for space with…

The challenge with this approach as that many very competent people are not allowed to discuss their prior work in that level of detail. More practical variants of this approach use a straw man software design problem to talk to that will exercise diverse areas of experience.

Re: Don’t do interviews, do discussions

#229

Post by ryandrake »

iamsb wrote: post #222

Having unscripted conversations is one of the best way to be swayed by unconscious bias in interviews. Even though this advice sounds awesome, I will be cautious of putting it into practice without thinking through the bias problem. I do remember reading multiple research papers on this, but unable to find them at the moment. From anecdote - In the last company I worked in London, only one team (DevOps) did not follo…

> In the last company I worked in London, only one team (DevOps) did not follow scripted interviews. It was the least diverse team, not just in terms of representation, but in terms of diversity of thought.

Did this result in poorer job performance for the DevOps team, or any other negative business results that were specific to that team? If not, who’s to say which interviewing method was better or worse?

As an interviewer I’ve always felt very constrained by scripted interviews and “approved question lists”. I always struggle to really evaluate a candidate when I’m asking pre-selected questions without knowing why I’m asking those questions.

Re: Don’t do interviews, do discussions

#230

Post by mark-r »

I recently had the best interview experience of my life from an interviewer who had this philosophy. He's a professional interviewer who's done thousands of interviews over the past few years. He feels that the best way to gauge a candidate is not by their knowledge of minutia, but by whether you can trust the things they claim about themselves.

He's so passionate about the subject that he created a Youtube channel. It's aimed at both interviewers who want to do a better job and interviewees who want to influence their chances of success. https://www.youtube.com/playlist?list=PLhCnsRMXhadbiHsTcxMCg...

Post reply on HN