One more proof:Interviewing on whiteboard does not work

This is fascinating, but it's nothing to do with whiteboards, and everything to do with individuals making technical assessments of other people. The title of the article is:

    You will randomly bomb technical
    interviews. Everyone does. Here's
    the data.

That's more accurate that the one given here.

I have the same problem with these analyses from interview.io every time I see them: they are making many of their inferences by fitting linear models to data that is non-linear. Is getting a 4 twice as good as getting a 2? If not, don't use a linear model!