Comment by MontyCarloHall

>Here’s what they found: The participants being watched scored half compared to those who were alone.

I once had a coding interview where I was given a laptop with no internet access and 30 minutes by myself to solve a couple questions a step or two above FizzBuzz, in pseudocode. The interviewer then came back into the room and we discussed my solutions, using them as a jumping-off point into a more open-ended discussion (e.g. "why did you use a hash table here? can you think of another approach?" which then led to a general discussion on the suitability of hash tables versus other approaches for related problems.)

It was one of the best technical interviews I've had. I'm surprised more places don't do this. It's a win-win-win: it relieves stress on the interviewee, gives the interviewer back half an hour of valuable time, and makes for a much more productive assessment of the candidate — it's a lot easier for a candidate to explain code they've already written, and thus a lot easier to an interviewer to assess the candidate's thought process.