Ask HN: How do you get around question deflection when interviewing?

One of the most useful tactics in interviewing, or in any conversation, is to use silence as a tool. In your example, if they say "Deep learning", instead of responding with "that's interesting,..." stay silent for a few momenths. They will subconsciously jump to fill the gap in conversation, often revealing even more about themselves with the "unprepared" response you promted with your silence.

Another tact to take is being more "strong-armed". If they "dont want to bore you with the details", simply say "I would like to hear the details, break it down for me".

Simply don't allow them to deflect. If they deflect and haven't answered your question to your satisfaction, be straight forward and ask them to elaborate.

If you circle back and say something like, please bore me with the details it sounds interesting and they refuse that would be a deal breaker for me. It is in my mind a huge red flag.

> "I don't want to bore you with the details".

Response: "Try me."

Dig for the details and if the candidate can't provide them, then either someone else did the work, they used a library and don't actually understand it, or they made it up altogether. This is assuming, of course, that they did the work relatively recently (<5 years).

If they deflect by mentioning either IP agreements or NDA's, then remain on topic but discuss something else under the same umbrella. For example, if the candidate deflects a question concerning k-means clustering, then ask about k-nearest-neighbors. If they continue to deflect, that's a red flag.

I'm a pessimist when it comes to hiring, but there are a lot of imposters claiming undue credit in this industry.

I recommend finding the best bullshitter at your company and having that person interview the candidates with you. A good bullshitter can always spot another bullshitter.

Your role as an interviewer is to extract the information out of a candidate, otherwise we'd be using programs to do that for us because it's not just a 'yes' or 'no' answer. What I would do if I were you is to ask a follow up question - "I'm actually interested in learning more about that deep learning piece you mentioned, could you elaborate more on that, for example, is it using xyz technique or abc?". If the candidate can't answer your follow up question then follow up with "do you have experience in deep learning?". Only then, you'll get an accurate answer to your question. There's no guessing nor red flagging. If it's a yes then dig more until you get to understand the level of exposure to deep learning for this candidate. As you can tell, it involves 2 things from you: 1. be on top of deep learning if it's a requirement for this role, 2. being curious and ready to triple check your judgement. It's not a "I know, you don't know" type of situations, it's about trying to understand the level of a candidate within certain topics, while pushing the conversion in the right direction. That is your role as an interviewer.

Last but not least, it's totally fine if the answer you get is "I don't know anything about this topic". Try to find out if the candidate would be willing to learn deep learning and the reason why he/she would be interested in learning that. You may not have a deep learning expert in front of you, but maybe this person has the potential to become the next DL guru based on other things that could highlight excellent problem solving. You have to bridge the gaps as an interviewer by staying positive until the end - I'm sure this candidate has something to offer, let's try to find out what is that super power!

In the past, when i was asked about details of projects (non confidential) that i actually was working on - it always was fun for me to go into details and have conversation of equals instead of being interrogated.