http://stephaniehurlburt.com/blog/2016/10/5/a-few-stories-of...
http://stephaniehurlburt.com/blog/2016/10/5/a-few-stories-of...
Giving a take-home exercise and having part of the interview be a collaborative review of it? Good idea, works well.
Developing exercises where interviewer and candidate pair on a problem, using real coding tools and looking things up as necessary? Good idea, works well.
"Here's the whiteboard, here's a problem, code it up" -- doesn't usefully distinguish qualified from unqualified candidates, doesn't give you meaningful information about job-relevant skills, serves only to make people squirm and serve as a "yeah, but I had to do it to pass the interview, too, so now you do". Get rid of it forever.
Also, any set of questions designed not to determine someone's programming ability but instead to act as a proxy for something else, like "went to Stanford", goes out the window first chance I get.
This is not implementing van Emde Boa search structure, but simple stuff any entry level programmer must comprehend for a general programming posistion.
If you think the only way to have a coding exercise is to pose a whiteboard problem, then my advice is get out of this industry, now.
But my point is that, as a team you need to be able to express ideas in a format that other developers can reason with you in a live setting. Nothing beats a whiteboard in that regard. You can use a projector and a computer, but what if you need to draw a graph or something else?
It's not a tall order to ask them to be able to do this. And please don't forget that I'm not asking people to implement RB search trees, but usually a simple loop and an if statement. Reverse a list using a for loop, transform this string into a palindrome, is this string a palindrone, remove all duplicates in a list - stuff like that.
Asking coders with 10 years of experience to write 10 lines of pseudoish code is wrong!! Try avoiding a coding exercise in interviews for coding positions at all costs. In the worst case make it a take home exercise. This is the only way to save this industry!!! Got it!
Also there's the self confidence issue, common for many people, even while not a lot of people speak out about it. Ask someone with low self confidence to stand next to a white board and "implement" solutions to random problems, and he/she will freak out. At the same time, these people might very well excel in a more nurturing, friendly environment. Yet at the same time, people who are actually self confident and able to solve problems without much context on a whiteboard, might start running in circles once faced with unfamiliar problems that do not directly map to the algorithms they already know.
I found that just talking about code or problems with someone, can tell you lots of things about the ability of a person. Also, most likely, the candidate will already have some code put online, and god forbid companies are too lazy to look at it, rather than stroking their own ego by asking out of context questions that they already know the answer to.
I don't think a company interviewing a candidate needs to be holding their hand during the interview process. I understand that someone might perform a lot worse when whiteboarding, but an interviewer should be taking that into account.
Google sees about half of their employees come from at least a second interview attempt, implying that qualified people fail at least 1/2 the time. Maybe you're claiming that their process is terrible? Or that most companies are significantly better at hiring than them? As the only data point I can find, though, it seems pretty damning.
> And if there are only three examples of rejections
It's unclear to me why only rejections count. 7 for 7 is a lot more concerning than 3 for 3.
FTR, out of the 6 interview processes I've gone through I can think of 2 that don't correlate in this way (one in each direction). So I think I have some basis for suspicion.