3 ms·

> ...anybody who is in a hiring position for specifically engineering knows exactly how utilitarian the hiring process is. Well, it attempts to be utilitarian.

> ...anybody who is in a hiring position for specifically engineering knows exactly how utilitarian the hiring process is. Well, it attempts to be utilitarian. However, there are routinely articles on HN, from the perspective of the prospect and the company, about how much of a crapshoot interviewing for coding is.

rsj_hn 9y ago
Yes, it's a crapshoot. There can be bad hiring decisions. There can be bad business decisions. And people can be jerks or rude. People are passed over promotions. That happens all the time to everyone. But it's problematic when someone sees each event as happening to them as some conspiracy against their race/gender/age/whatever, as opposed to someone just disagreeing with them or dismissing their argument because they think it's a weak argument. And this is made worse because some people view everything through the lense of identity. It is literally the only developed conceptual framework they have. And when you try to explain, no, it was because I thought Y was better, then there must be unconscious bias, that you weren't even aware of. So fundamentally this debate is about the existence of an objective reality that does not have, at its base, gender or race or other attributes. Most of us who are engineers see a different base and it's pretty condescending to have someone explain to you that you prefer the person who aced the interview to the person who struggled because of some unconscious bias, or that you asked the wrong questions, etc. When I interview, I have a bank of questions and answers already written out. Each question has a well-defined definition of passing. I decide which question to ask based on the topics that the person lists on their resume. That is, question selection happens from reading the resume, and it happens before the interview. If they say they know javascript, I give them something from the JS pile. If they say they know crypto, I give them something from that pile. I don't want to ambush them with stuff they don't know. The assumption is that if they know the stuff they say they know well, then they can learn what they don't know as needed. And believe me, there are huge differences in how people do on these questions. For example, in the javascript test, I have some code with vulnerabilities that I ask them to find. I give them the code, explain what it does, explain what I want them to do, and either stay in the room or walk away, depending on what they want. After 20 min, I come back and ask them what they found. A large number of people find nothing. Some find one, but not the biggest one. Some find everything. Some are not able to understand how the code even works. This is for a security role. If they say they know crypto, then I have a bank of crypto questions. For example, I write down a common authentication and key establishment protocol. Then I ask them what could an attacker do if the last confirmation message was not sent. There are huge differences in outcomes. Some people are confused and have no idea. Others start confidently making stuff up. Others actually identify attacks that would be possible, demonstrating an understanding of cryptographic protocols. None of these differences are based on race, gender, age, etc. There isn't too much of a middle ground -- most people either succeed or fail. Now, they may have failed because they were having a bad day, or maybe they don't do well at technical interviews, or they might have lied on their resume, but there is an objective reality. This is what people mean when they say the interviews are utilitarian. But I honestly don't think that a large portion of the population is capable of understanding that. They really see everything in terms of identity, so the questions must have been sexist, or I was being unfair to the interviewee. Now the battle becomes, must everyone else also reject objective reality? Must everyone else throw away the notion of ranking candidates based on their ability to reason abstractly and demonstrate problem solving skills in a live interview, or must we just adopt a notion of "qualification" ? E.g. do we give everyone a job based on their GPA/list of degrees, or is there room for performance based evaluation of individuals? This debate is very important to tech, because there are a lot of incompetent engineers with degrees that really impact a team. There is a big difference in productivity between good engineers and bad engineers. Joel Spolsky is spot on here: https://www.joelonsoftware.com/2006/10/25/the-guerrilla-guide-to-interviewing-version-30/ https://www.joelonsoftware.com/2006/10/25/the-guerrilla-guid... In engineering, actual talent matters, and it matters a lot. This isn't like hiring a bus driver. You don't get promoted for having been there for X years, and you don't get hired based on your credentials. Your idea doesn't get adopted because "it's your turn" to have your way. Individual talent matters, and the fact that we may not always detect it accurately is not evidence of the hiring process not being utilitarian or any kind of identity-based discrimination.
- morgtheborg 9y ago
  I often think the bias occurs before the interview in terms of the pipeline. For me, that's my largest self-correcting area. There are groups of people or people who present themselves in a certain way that I simply don't want to interview because it feels like 'wasting my time'. Clearly that's bullshit and I make an active effort to overcome those biases and, if anything, now am overcorrecting (a better mistake to make imo). Once they're in the room, you're outline seems like a solid way to measure they're abilities. Of course I'm not saying there isn't opportunity for bias here as well (mainly in the selection of the question), but you've offset it magnificently by having written down the correct answer and (hopefully) prioritizing getting said answer rather than the manner in which it was conveyed ("I think..", questions rather than statements ("It's x, right?"), "Maybe it's..", etc. since those types of self-effacing comments are often seen and presented as polite forms of communicating for women).
  - rsj_hn 9y ago
    In my case, I'm not a recruiter, so I interview everyone they ask me to interview -- I've never played a role in the original vetting. I think most large companies have a formalized system of initial screening -- which has been criticized for relying on keywords and such. It's done by HR. After that, there's a phone screen which is much like an in person interview -- the questions should be pre-written based on the resume and generally they are. In my company I've worked to write up large banks of questions w/answers and distributed them to people doing phone screens for this purpose, and I've used that in my own phone screens. What people should do is formalize the interview process which helps in so many ways -- a repeatable process with repeatable questions lets you compare people across times and gives a lot of insight. It also helps reduce bias. I do think that small companies tend to do a worse job that big companies with established HR practices. That's not only true for HR, it's true across the board. Young people who are inexperienced are going to make mistakes. Certainly there is going to be "culture" type bias in start ups and other small companies at least during the screening process. I wouldn't say that's too relevant for "Silicon Valley", as you have these issues in small businesses in every industry.

rsj_hn 9y ago

Yes, it's a crapshoot. There can be bad hiring decisions. There can be bad business decisions. And people can be jerks or rude. People are passed over promotions. That happens all the time to everyone. But it's problematic when someone sees each event as happening to them as some conspiracy against their race/gender/age/whatever, as opposed to someone just disagreeing with them or dismissing their argument because they think it's a weak argument. And this is made worse because some people view everything through the lense of identity. It is literally the only developed conceptual framework they have. And when you try to explain, no, it was because I thought Y was better, then there must be unconscious bias, that you weren't even aware of. So fundamentally this debate is about the existence of an objective reality that does not have, at its base, gender or race or other attributes. Most of us who are engineers see a different base and it's pretty condescending to have someone explain to you that you prefer the person who aced the interview to the person who struggled because of some unconscious bias, or that you asked the wrong questions, etc. When I interview, I have a bank of questions and answers already written out. Each question has a well-defined definition of passing. I decide which question to ask based on the topics that the person lists on their resume. That is, question selection happens from reading the resume, and it happens before the interview. If they say they know javascript, I give them something from the JS pile. If they say they know crypto, I give them something from that pile. I don't want to ambush them with stuff they don't know. The assumption is that if they know the stuff they say they know well, then they can learn what they don't know as needed. And believe me, there are huge differences in how people do on these questions. For example, in the javascript test, I have some code with vulnerabilities that I ask them to find. I give them the code, explain what it does, explain what I want them to do, and either stay in the room or walk away, depending on what they want. After 20 min, I come back and ask them what they found. A large number of people find nothing. Some find one, but not the biggest one. Some find everything. Some are not able to understand how the code even works. This is for a security role. If they say they know crypto, then I have a bank of crypto questions. For example, I write down a common authentication and key establishment protocol. Then I ask them what could an attacker do if the last confirmation message was not sent. There are huge differences in outcomes. Some people are confused and have no idea. Others start confidently making stuff up. Others actually identify attacks that would be possible, demonstrating an understanding of cryptographic protocols. None of these differences are based on race, gender, age, etc. There isn't too much of a middle ground -- most people either succeed or fail. Now, they may have failed because they were having a bad day, or maybe they don't do well at technical interviews, or they might have lied on their resume, but there is an objective reality. This is what people mean when they say the interviews are utilitarian. But I honestly don't think that a large portion of the population is capable of understanding that. They really see everything in terms of identity, so the questions must have been sexist, or I was being unfair to the interviewee. Now the battle becomes, must everyone else also reject objective reality? Must everyone else throw away the notion of ranking candidates based on their ability to reason abstractly and demonstrate problem solving skills in a live interview, or must we just adopt a notion of "qualification" ? E.g. do we give everyone a job based on their GPA/list of degrees, or is there room for performance based evaluation of individuals? This debate is very important to tech, because there are a lot of incompetent engineers with degrees that really impact a team. There is a big difference in productivity between good engineers and bad engineers. Joel Spolsky is spot on here: https://www.joelonsoftware.com/2006/10/25/the-guerrilla-guide-to-interviewing-version-30/ https://www.joelonsoftware.com/2006/10/25/the-guerrilla-guid... In engineering, actual talent matters, and it matters a lot. This isn't like hiring a bus driver. You don't get promoted for having been there for X years, and you don't get hired based on your credentials. Your idea doesn't get adopted because "it's your turn" to have your way. Individual talent matters, and the fact that we may not always detect it accurately is not evidence of the hiring process not being utilitarian or any kind of identity-based discrimination.

morgtheborg 9y ago
I often think the bias occurs before the interview in terms of the pipeline. For me, that's my largest self-correcting area. There are groups of people or people who present themselves in a certain way that I simply don't want to interview because it feels like 'wasting my time'. Clearly that's bullshit and I make an active effort to overcome those biases and, if anything, now am overcorrecting (a better mistake to make imo). Once they're in the room, you're outline seems like a solid way to measure they're abilities. Of course I'm not saying there isn't opportunity for bias here as well (mainly in the selection of the question), but you've offset it magnificently by having written down the correct answer and (hopefully) prioritizing getting said answer rather than the manner in which it was conveyed ("I think..", questions rather than statements ("It's x, right?"), "Maybe it's..", etc. since those types of self-effacing comments are often seen and presented as polite forms of communicating for women).
- rsj_hn 9y ago
  In my case, I'm not a recruiter, so I interview everyone they ask me to interview -- I've never played a role in the original vetting. I think most large companies have a formalized system of initial screening -- which has been criticized for relying on keywords and such. It's done by HR. After that, there's a phone screen which is much like an in person interview -- the questions should be pre-written based on the resume and generally they are. In my company I've worked to write up large banks of questions w/answers and distributed them to people doing phone screens for this purpose, and I've used that in my own phone screens. What people should do is formalize the interview process which helps in so many ways -- a repeatable process with repeatable questions lets you compare people across times and gives a lot of insight. It also helps reduce bias. I do think that small companies tend to do a worse job that big companies with established HR practices. That's not only true for HR, it's true across the board. Young people who are inexperienced are going to make mistakes. Certainly there is going to be "culture" type bias in start ups and other small companies at least during the screening process. I wouldn't say that's too relevant for "Silicon Valley", as you have these issues in small businesses in every industry.