Hacker News Reader

> Some people are really good at interviewing and being charismatic enough to convince people to hire them.

Totally agreed. Traditional interview processes select for people who are good talkers. That doesn't correlate well with technical skill: you over-hire glib people and under-hire people who aren't. E.g., the shy, awkward, and anxiety-prone.

When I run an interview process, I focus on making it as much like real work as possible. Some pair programming, some technical discussion, some joint product collaboration and systems design. It's my firm belief that if we want to know if people can do a thing, we should try doing the thing with them. It's not perfect, but it's way better than asking people about their 10-year career plan or having them solve Mensa puzzles.

>Traditional interview processes select for people who are good talkers. That doesn't correlate well with technical skill: you over-hire glib people and under-hire people who aren't. E.g., the shy, awkward, and anxiety-prone.

This is true: but "technical skill" is not the ONLY hiring criteria that should be considered. (I say this as a fairly shy, awkward and anxiety-prone person, who has had his share of interviews-gone-wrong).

I'd posit that you DO want to hire people who are good talkers, and are glib. They help build teams. They help with information flow. The greatest ideas in the world are worthless if they can't be executed by a team that doesn't understand them, and the person who came up with it can't communicate that idea effectively. One might say that that is what code is for. But that's not true. If code were to communicate between the developer and the machine, we'd all be writing in machine language. Code is a communication tool for developers to collaboratively tell the machine what to do. And that has to be supplemented by human, interpersonal skills. (especially when you have collaborators who are not coders, like business and finance people who organize the teams and manage projects and programs).

I like that you focus on pair programming. One horribly toxic factor I see at way too many places is where management pits programmers against each other like it's some kind of competitive sport, and collaboration is frowned upon because "if you can't figure it out by yourself, you're a burden on the rest of us".

I am certainly not arguing for only hiring based on technical skill. Perhaps you didn't get the chance to read all of my comment, but I make it clear collaboration is important. I'm also not totally sure you know what glib means, so I'll just quote the definition here: "fluent and voluble but insincere and shallow".

Basically, I think workaday collaboration and communication is a pretty different skill than on-the-spot glibness. If I'm hiring a PR person or a press secretary, glibness is a valuable skill, because an important part of their job is to talk over people, to favor sounding good over consideration or substance. But for a software team, I value more substantial engagement and communication. So no, I'm happy to hire people who are not glib. Have done before, will do again. And for those who do happen to have the gift of gab, I want to be sure they have it in check and only deploy it when necessary. A colleague who can say, "I don't know" is way better than one who glibly avoids that out of pride.

> A colleague who can say, "I don't know" is way better than one who glibly avoids that out of pride

An interesting interview question could be to study the job applicant's background and then ask questions that, given the background, s/he doesn't know -- then, will s/he tell you this, or make up nonsense

What you may be missing is how hard it is to actually create a good take home. Every take home test I’ve seen was rife with potential for the problem to explode in complexity. Even things that seem simple like names and dates have so many potential pitfalls that it can be impossible to tell as an applicant whether they intentionally laid a trap or not. Then there’s the incidentals, should I send them a docker image to increase the odds it’ll work on their machine? Oh, they’re going to want me to extend this in real time, should I include other nice to have services that this problem could dovetail into needing, like redis?

As far as I can tell, any company that isn’t willing to pay contracting rates to solve real problems on their stack is likely being disengenuous with their take homes and largely biasing against experienced devs who aren’t as likely to waste their time. And worse with a take home, is that they have no skin in the game. With an in person interview they lose at least as long as the inverview. With a take home they lose nothing except short email exchanges.

A code comment saying “Here’s a potential pitfall we could discuss addressing” is often more valuable than a solution. Every software project has more problems than it has people-hours available. Every team has the engineer who spends a week fixing an edge case in their ticket that no customer cares about. Demonstrating awareness of this makes you an attractive candidate.

Take homes are easy to iterate on. If multiple candidates can’t understand the problem or waste time with over-complicated solutions, you update the instructions.

If you’re constantly worried about “traps” then you might not want to work for those companies anyway. Take homes should be straightforward and similar to solving real problems.

> As far as I can tell, any company that isn’t willing to pay contracting rates to solve real problems on their stack is likely being disengenuous with their take homes

Take homes are contrived problems, not real work. Every candidate receives the same problem so they can be compared.

No company should be giving employees actual unpaid work to do as part of the interview. That’s more of an internet trope than a real problem because no rational company is going to be sending their codebase to applicants to add features to.

> What you may be missing is how hard it is to actually create a good take home

The biggest issue I see are people underestimating the setup time for "simple" projects. You're a big software company that doesn't start new projects regularly, I'm a developer at a big company that doesn't setup new projects regularly, why do you expect me to remember how to setup "professional quality" projects in a couple of hours?

A simple console app might test if I can code, setting up a new MVC project with authentication, unit tests, IoC, etc is testing how well I can google shit I haven't had to setup for years.

It depends on when in the interview process you give the task, and if they're interviewing with any other places. If two places do this, a person now needs 8 hours or 12 in the week you give them. Ask for the task to be done before properly interviewing them and that presents more problems I hope that last one dies.

We spent some time on making up a quite simple project with a readme about the tasks to be implemented. It's something that can be done in half an hour when you are fast and we always thought we need to make it a bit harder. Turns out it's good enough to weed out quite many people who can't keep a deadline or don't get their shit together in other ways. From reviewing the code you can judge how people think, if they keep their files and code in order, use git etc. We also prompt people do document their process in finding a solution.

Interesting, could u share a bit more pls?

We made up a problem with technologies that we use at work targeted at frontend engineers. So we set up a project with a Symfony backend and made the candidates create Twig templates and SASS styling based on Bootstrap. Then we put that up on Github with a README. So the candidates could focus on the frontend stuff without caring for the backend. Still they had to figure out how to set up SASS compilation and modularization of Twig templates.

As I said it's not very hard but you can get bonus points if you come up with considerations for responsive design or stuff like that. Overall the aim is to create readable idiomatic code and not come up with something clever that no one but you can understand.

I guess that's a very specific task that's of no great use to others but the point is to have a specific problem with some blanks that can be filled out with a reasonable amount of work by the candidates. Plus the general workflow using git and such.

Although, if you wanted them to do a longer project, Basecamp’s method where they pay the applicant to do the project seems like a good way to do it. And you’ll get the most complete picture of their expertise (of course, you wouldn’t want to do this until the last step of the process).

This can get sticky with employee contracts, that may (and likely do) forbid an employee from doing contract work for another company, at least without permission.

I would find 4 hours a lot. It is basically wholw afternoon.

Ever been to an onsite interview? It'll take you the same amount of time. Or longer. While being far higher pressure. And probably involve writing code on a whiteboard instead of an IDE.

I was not on over half day long interview that would primary involve whiteboard coding.

Does a take home eliminate the onsite interview? Or is it just additional time?

Even if it is additional, it's not consecutive. The OP merely said 4 hours seemed like a long time. My point was not that this 4 hour block would or wouldn't obviate the need for the other; just that the other means 4 hour blocks are standard.