Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

right. So when you’re done throwing the algorithm on the board, the candidate either implements it from memory because he’s a new grad from college, or has to deal with the overhead of communicating fending off your questions and “helpful” suggestions. It’s annoying and random and and gets you less interesting of candidates on net.

I swear, I was just dealing with this a week ago. “Excuse me, Mr. Interviewer. I can either answer your questions about what’s going on here or attempt to fix the last let of this code. Which is more valuable to you?” Proceeded to use silence to fix code. Puzzle solving uses the uninterrupted brain.

And just the other day, I had the SQL schema question to design a schema … for a calendar! Worst interview question I ever experienced, honestly. Nearly no one designs calendars and they’ve got lots of ways to trip you up because recurring events are exciting. (Also, when I did design something decent with two tables and was asked to query both with a single SQL statement my interviewer didn’t understand UNION but that’s another matter.)

I like take home projects augmented with short exercises to verify that I’m not a plagiarist who’s incompetent at code. If you want to assess my puzzle solving capabilities let me play a Zachtronics game like SpaceChem or the like.



I gotta ask... are companies you're interviewing with expecting you to have memorized particular algorithms? Or to be perfect with syntax?

In my most recent round of interviews, I found almost no one expecting me to recall a particular algorithm and implement it for a solution to a problem, but I did find plenty of questions that involved dealing with data structures, and writing my own code and algorithms to solve those problems.

Maybe I've been fortunate with not having been asked questions like "implement the quick sort algorithm"?


In this case it's a "distance between two employees in an org chart", which is basically a trivially modified closest-common-ancestor algorithm. The complication was that there was no ability to index into the tree at an employee and climb upwards, and they didn't want to incur the overhead of building that at the start.

That's fine in and of itself, but when I'm having a bit of a time pruning the common-ancestor paths, I want to use all 7 brain-registers for the problem instead of devoting 2-3 for communication, or, worse, having to flush everything from cache to analyze and respond to your "hint".




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: