[ankurdhama]

I think it depends on the interviewer intentions. If the intention is to judge whether the candidate knows such and such algorithm (aka they have used or read about it) then it is obviously stupid. On the other hand it could be about how you approach the problem, how you start with a simplest possible solution and then engage in a thought process (along with conversation about what you are thinking) to improve the solution. In this case the final solution doesn't matter, what matters is your thought process.