Interviewers are generally unknowledgable of the questions they are asking (they are pulling it from an internal database). Wasted time trying to convince the interviewers of the algorithms when what I came up with was different than what the official solution is. (eg trying to convince the interviewer that DFS approach works on topological sort) They require complete/functional code, yet they don't provide a proper input to test from (one interviewer presented a 500 line file cut-pasted and wanted to get it into coderpad). They are not interested in how you think/approach to the problem, only if you can get the final answer or not.