5 ms·

So it's almost HN cliche to complain about these sorts of questions, and that's a fairly earned failure, but I just want to reiterate that these sorts of questi

So it's almost HN cliche to complain about these sorts of questions, and that's a fairly earned failure, but I just want to reiterate that these sorts of questions are one or two orders of magnitude more meaningful than the nebulous questions you sometimes see, "estimate how many auto mechanics serve Portland" or "I shrink you to the size of a pea and throw you in a blender, how will you survive?" or the like. At the core, what you should be doing if you want to hire properly, is to take a break approximately 10-15 minutes into the interview. Go get a bottle of water or something. During the next 1-5 minutes, jot down on a piece of paper your initial impressions of the candidate, your suspicions and biases. Some studies show that based on folks viewing the first five minutes of an interview, they can usually predict the outcome. One can interpret that in several ways, but the more cynical is to just say that most interviews are exercises in confirmation bias: you get some suspicion of the candidate, you spend the rest of the interview looking for evidence of your suspicions so you can feel good about your eventual predetermined decision. If you want to do better, it's very simple: do what a scientist does and flip this on its head. Write down your suspicions and then try to disprove them. You think that this person is too egocentric for your company? Then ask them to talk about how they have worked with others, ask what they'd do if a coworker was struggling with a personal issue, try to gather evidence that they can actually be compassionate and empathic. You think that this person cannot program their way out of a bag? Throw them a softball programming question like FizzBuzz. You think that they have encyclopedic knowledge of algorithms? OK, then ask them to implement a Dijkstra's algorithm and if they don't know what that is, a breadth-first traversal. What people are really sick of is that people use these tools as crystal balls into their own psyche: You as an interviewer tend towards "he got QuickSort wrong!" if you subjectively didn't want him in your company, "she got QuickSort almost right, except for some finagling around the pivot!" if you subjectively did want her. What's missing is deliberateness: choose what you want to know deliberately, choose the question deliberately to satisfy some goal: either saying "yeah my hunch was wrong about that, they did pretty well" or else "no, I really tried to give them every opportunity to show me that they can code and even trying to give them the benefit of the doubt I am still unconvinced."

Mc91 8y ago
> Some studies show that based on folks viewing the first five minutes of an interview, they can usually predict the outcome. One can interpret that in several ways I have interviewed dozens of people, possibly hundreds of people. I do get an impression within the first three questions. Some people I ask three questions covering different areas, and they knock all three questions out of the park. Some I ask three questions of and they kind of fumble around and give a very limited answer for two of the questions, and can't answer the third at all. In my experience, I can't recall ever seeing a divergence after this. If they fumble through the first three questions, they will fumble through the rest of the interview. If they can answer the first three questions coherently and in-depth, they will be able to answer virtually all of the subsequent questions coherently and in-depth. I'm not really confirming biases because I generally ask everyone the same questions. The only time I mix it up is if they're answering all the easy questions correctly, I start asking slightly more difficult questions which I don't ask much. I don't ask them much because if someone is stumbling over the easier questions, there's no reason to ask them tougher questions. Also, my first technical question is almost always an incredibly easy, standard question. I just ask it so they can answer one quickly and correctly and relax a little. But maybe 20% of people botch even this.
- crdrost 8y ago
  Now that's very interesting. I think you're right that within those first questions you do get some sort of perception about technical capacity in some raw form. Maybe it's even worth not trying to disprove that hypothesis by asking a separate question. But what I'm confused about, that I'd like to hear your opinion on, is why if you have this confidence that your first three questions contain 90% of the signal, that you spend the rest of the interview doing anything vaguely related to that? Like it sounds to me that you just told me, "I get a 90% accurate measurement in the first, say, 30% of the interview, and then I waste my time for the remaining 70% trying to up it from 90 to 98%," and that just doesn't seem like a great use of your time after that point. Can you help me understand -- why don't you pivot to something like trying to evaluate their conscientiousness or openness to learning or emotional intelligence or ownership of their previous projects and mistakes and such? Or do you mean that you ask these programming-type questions and you indirectly learn about those other dimensions of success? Also, I'd be curious to know what you do about teachability for some of the skills -- I know that sometimes it's just "we're hiring for a senior role and we cannot accept a junior/intern at this time," but very often it seems like it's just "we're growing in general" and then you are open to saying "hey we don't have an opening for you as a mid-level developer but we would be willing to hire you on as a junior-level dev and if you are as good as you say you are we'd also consider promoting you after your 90-day-review when we've seen you really shine." Are you trying to assess the level of the role that you would offer the candidate with your questions, in other words? Or are you just judging fitness for a fixed role?

Mc91 8y ago

> Some studies show that based on folks viewing the first five minutes of an interview, they can usually predict the outcome. One can interpret that in several ways I have interviewed dozens of people, possibly hundreds of people. I do get an impression within the first three questions. Some people I ask three questions covering different areas, and they knock all three questions out of the park. Some I ask three questions of and they kind of fumble around and give a very limited answer for two of the questions, and can't answer the third at all. In my experience, I can't recall ever seeing a divergence after this. If they fumble through the first three questions, they will fumble through the rest of the interview. If they can answer the first three questions coherently and in-depth, they will be able to answer virtually all of the subsequent questions coherently and in-depth. I'm not really confirming biases because I generally ask everyone the same questions. The only time I mix it up is if they're answering all the easy questions correctly, I start asking slightly more difficult questions which I don't ask much. I don't ask them much because if someone is stumbling over the easier questions, there's no reason to ask them tougher questions. Also, my first technical question is almost always an incredibly easy, standard question. I just ask it so they can answer one quickly and correctly and relax a little. But maybe 20% of people botch even this.

crdrost 8y ago
Now that's very interesting. I think you're right that within those first questions you do get some sort of perception about technical capacity in some raw form. Maybe it's even worth not trying to disprove that hypothesis by asking a separate question. But what I'm confused about, that I'd like to hear your opinion on, is why if you have this confidence that your first three questions contain 90% of the signal, that you spend the rest of the interview doing anything vaguely related to that? Like it sounds to me that you just told me, "I get a 90% accurate measurement in the first, say, 30% of the interview, and then I waste my time for the remaining 70% trying to up it from 90 to 98%," and that just doesn't seem like a great use of your time after that point. Can you help me understand -- why don't you pivot to something like trying to evaluate their conscientiousness or openness to learning or emotional intelligence or ownership of their previous projects and mistakes and such? Or do you mean that you ask these programming-type questions and you indirectly learn about those other dimensions of success? Also, I'd be curious to know what you do about teachability for some of the skills -- I know that sometimes it's just "we're hiring for a senior role and we cannot accept a junior/intern at this time," but very often it seems like it's just "we're growing in general" and then you are open to saying "hey we don't have an opening for you as a mid-level developer but we would be willing to hire you on as a junior-level dev and if you are as good as you say you are we'd also consider promoting you after your 90-day-review when we've seen you really shine." Are you trying to assess the level of the role that you would offer the candidate with your questions, in other words? Or are you just judging fitness for a fixed role?