Try writing a letter to a sympathetic person (real or imaginary) who you trust, but who knows nothing of your specific circumstance, which explains your dilemma in detail, and explains why it’s a problem — why the obvious solutions won’t work. Try systematically listing all possible solutions and explaining exactly why they won’t work.

While you do this, or when you read it afterwards, does some part of it seem too vague, or dubious, or wrong?

The bystander effect, especially in an emergency situation, is one of those things that causes me a bit of visceral horror/terror.

Three times I’ve had an occasion to call 911 as a bystander (car crash, house fire, and gang beating). Even though calling to report such things is extremely low cost to me, each time I’ve second guessed picking up the phone, thinking things like, “dozens of people have already driven by this same scene. Surely one of them has called by now and my report is redundant.” Each time I’ve reminded myself “bystander effect” and called anyway. And when I asked if officers were already on the way, I’ve been told each time that my call was the first notice they had.

Even just within “what you think about”, I think you also need to think about many different things, not just the things you think are most important. One reason is cross-fertilization, but another one is a kind of burnout — not emotional or energetic, but creative. This is also related to “not being in a rut”, but I’m not sure it’s exactly the same.

I didn’t find this to be the case in mathematics. I usually found diminishing marginal returns to the number of hours of math I did each day, after which I had time to work on other projects like founding the Center for Applied Rationality (http://rationality.org/), and I still managed to get my top choice of postdoc offers.

I don’t think this was particularly rare or exceptional, either… many of my friends in math grad school who managed to get good academic jobs afterward seemed to spend a lot of spare time playing board games or video games of various sorts, eating extended meals together, or engaged in hobbies like music, dancing, biking, or climbing. I think they experienced these activities as “needing a break”, and I found that working with friends on non-technical projects can often (though not always) serve equally well as a break from technical work. In other words, I think many academia-bound math grad students have the capacity work on other significant projects during grad school without sacrificing much from the quality of their research.

Laboratory sciences are a different game, however, where there seem to be a lot of physical tasks that need doing at specific times, and so my friends in those didn’t seem to have as much free time or flexibility as those in math, statistics, economics, or theoretical computer sciences.