I agree that you want to see someone thinking on their feet and tackling a new problem. But you can get that without making the whole interview a mystery. I can tell you "we will be giving you some C# code with concurrency bugs and asking you to find them", and it's still going to be a challenge when i do.
I have recently been doing interviews where we talk to candidates about past projects they've worked on, and ask them to go through some of the technical decisions they made. We don't warn them ahead of time that we're going to do this. Often, candidates have interesting projects they did a few years ago, and they just don't remember them in enough detail to go through them with us. This tells us nothing useful about the candidate. I think the interview would be far more useful if we had prompted them to revise their old projects, and i don't think this would generate false positives, because we can still distinguish bullshit from actual understanding (although of course i think that!).
> Also the most valuable question you can ask as a part of the non-technical interview: "So do you have any questions for us"? If the answer is "no", you are not looking at a good candidate.
I've heard people say this, but i don't see why it would be true. Is there reasoning behind this, or does it just feel smart?