I feel like giving people a decent coding project is, at the moment anyway, the best approach. If I'm interviewing someone I really just want to see these thing

I feel like giving people a decent coding project is, at the moment anyway, the best approach. If I'm interviewing someone I really just want to see these things: what their code is like, what their final result looks like (if there's room for creativity -and there should be), how they got there (revision history). And then in the interview portion I just want to get a sense of their personality and their decision making process on the project. It's not perfect but it gives a sense of if they can finish a likely task, and if they would be easy or hard to work with. Btw, none of that involves puzzles or trying to guess their IQ. If your company is like google in 1998 maybe you care about that, but chances are you need someone to build something relatively straightforward, not invent PageRank. Honestly having a super genius on board might be bad if you're not solving a truly hard problem - they might get bored and check out. It's a valid complaint that these projects can take time, but job searching is pretty time consuming anyway, and joining the wrong company or hiring the wrong person is way more so.