My main gripe is the restriction from really exploring and the limited human interaction. I don't like doing fine technical proofs either, more working with ideas and seeing solutions. There's a bit of a push to be structured and planned in research too, and ENFP structure and approach isn't really on the list. My last boss was awesome because he would just say what he wanted and accept the output at the end (normally a description of the whole system so he could understand it). There was none of this progress stuff making sure I was approaching it Si style. It still drove me nuts because I was interacting with ideas and computers too much rather than people though.