Take an interviewer who has the benefit of having worked through a problem literally a hundred times, further aided by internal training that represents thousands of iterations. Have them judge someone who hasn't had to solve that particular problem more than once or twice, twenty years ago in a different language (and certainly not on a whiteboard). That interviewer will form an inaccurate opinion of the interviewee's relative ability. Not maybe, and not by a little. Any interview process that does nothing to correct for that is broken.