3 ms·
Careful what you test for though. If your interview requires people to throw a complete app together in 40 minutes, you'll be hiring people who can throw things
by ryandrake 9y ago
Careful what you test for though. If your interview requires people to throw a complete app together in 40 minutes, you'll be hiring people who can throw things together in 40 minutes. Maybe that's what you currently need, maybe not. There are programmers out there who can spew an entire app together in 40 minutes, but the result is buggy, memory-leaking, O(N-squared) pile of crap unsuitable for anything other than discussing in an interview. There are other programmers who would spend those 40 minutes on a single subroutine, but the result is code that is robust, bug-free, unit-tested and will likely never have to be revisited. Test for the type of behavior you need.
- mi100hael 9y agoPresumably the exercise is predicated on the notion that you give someone a task that they should be able to complete properly within the allotted time. Personally I think it's pretty reasonable to think if you're hiring for a Sr. frontend developer who claims React experience, that they could spin up a basic SPA with a few React components for ToDo items in under an hour. I know I could do it and I don't even particularly like writing frontends.
- deedubaya 9y agoIt defeats the purpose of saying "build a thing" if you disregard how it's built. That's what you're actually testing. If you don't care how the thing is built, you probably don't care how your production products are built either.
- arenaninja 9y agoI worked at a place where the test was always to toss a single web page together in 2 hours (candidates were free to spend more time but by 2 hours they had written enough code for us to weigh ability). The candidate we ended up hiring did _not_ finish, because he spent his time documenting both his frontend and backend work. The thoughtfulness that went into each of his functions was apparent, and I knew right off the bat this was someone I could learn from. So I think there's a middle ground that both sides could be happy with, but maybe it needs to be more apparent that interviews are more about the journey than the results (in fact, all 3 candidates we ever hired via this method did not have a complete solution, but had followed enough instructions to be close enough)
- nickpsecurity 9y ago"but the result is code that is robust, bug-free, unit-tested and will likely never have to be revisited. " That was me. I did less work upfront but they did less rework later. The rework costs several times more with potential to disrupt customers. My compromise to speed it up was prototyping in something dynamic to get an idea of the solution with basic tests. Then, convert it to static, efficient, and clean code w/ more thorough tests or analysis. I'll fail a whiteboard since it will be some bullshit. If I get asked, I'll probably do it in Prolog or Brainfuck just to mess with their heads. The, type same thing into computer interpreter to show it works. If they gripe about readability, refactor the Brainfuck piece by piece into something that looks like, but isn't, C# or Java. "Just one more thing..." (writes use atomics in top) "...and now it's immune to concurrency errors since it relies on atomic operations." I'll fail most of the tests but might also get an offer for C# or Java job by confused, impressed, and/or skeptical interviewers. It would at least be funny outcome to show up telling the company, "No I don't know C# or Java. Here's the Brainfuck I wrote for the interviewers. You didn't ask them to hire for that? Well, here's my resume. You want me to build something real for you?" ;)
- nailer 9y ago> O(N-squared) pile of crap It's a TODO list. If performance is an issue you've done something terribly, terribly wrong. Like write your own algorithm. > There are other programmers who would spend those 40 minutes on a single subroutine Something which has not shipped has no value. This is an excellent indication for a 'no hire'.