[eanzenberg]

My analytical thought process on DS interviews:

- Signal is still quite low among noise, even with long multiple interviews, take-home homework, coding challenges, etc. Most relevant data is still hidden and takes months-years to come out.

- Companies seek to minimize false-positives much more than minimizing true-negatives.

- It's a numbers game from both ends because the probabilities are low, due to above 2 points.

[FireBeyond]

Woe betide the company that asks for a take home project, especially one that isn't just 50 lines of an algorithm (I'm talking one that I recall that asked for parsing Apache logs from a stream, displaying moving averages for URLs, aggregates, having high water mark "alerts" and resetting when rolling averages dropped below that, some unit tests and docs...

And doesn't bother to respond after you submit. I'm certainly not the world's best coder, but my solution met all those requirements in a reasonably elegant manner, and took several hours (10-20?) to make comprehensive.

That gets you on my shit list, and people I know get to hear about it, too.

[eanzenberg]

Same FireBeyond. Even worse, I've gotten feedback on a take-home project where the solutions had errors in it and I had the correct response! This, btw, was from one of the huge tech-companies in SF. Since it was the recruiter who went through the solutions, I couldn't nor cared to correct their mistakes :)