You should pick the thing where you are the most above average in expected success, because that's where the marginal gain to the world from you doing it is highest.
You're assuming perfect equivalence between success and value creation. That is a very shaky assumption in any realistic human society. Success may stem from creating value and trading it, but also from winning zero- or negative-sum games such as signaling arms races and rent seeking contests, and even the created value can be offset by the negative externalities you generate. Humans are amazingly capable of rationalizing away such unpleasant observations when it comes to themselves and people they like and admire, as well as exaggerating them when it comes to those they dislike, so evaluating any concrete success scenario accurately is a very difficult problem.
(Of course, if you're altruistic, you'd also care about your positive externalities for which you capture no benefit, and similar caveats apply there too.)
By "success" I meant successfully doing useful research/getting money to charity/spreading rationality. Which I think is closer to "value creation" or "positive externalities" than what you seem to be calling "success," namely personal welfare.
All I meant was that if he would create an above average amount of value in one area, but even more in another, relative to the rest of the labor pool, he should go with the one where he's best by comparison.
I have been having some difficulty deciding what to do with my life. I'm not really ashamed or surprised, because the problem seems extremely difficult and worth getting right. I still don't really know. Here are some of my options, though I wouldn't be surprised if what I end up doing is not on the list. I thought I would share to elicit advice, to give some context to some of my recent remarks, and maybe to provide comparison for people in similar situations.
1. Research
I could do research in a university, or in a private research lab. Many fields have at least a few questions that seem important and interesting.
A. Theoretical Computer Science. I could work on collaborative learning, recommendation and reputation systems, distributed protocols, or anything else that might assist collaboration in the future.
B. AI / Computational Cognitive Neuroscience. I could work on algorithms for inference and planning, to increase the probability that we develop comprehensible or human-like AI before something horrible happens (like developing incomprehensible strong AI).
C. Neuroscience. I know almost nothing about this field, but technology for measuring and interfacing with the brain appears to be important and developing rapidly. I don't know how hard it would be to start working in the field, but I suspect that the connection to computer science is strong enough at MIT that it wouldn't be impossible.
2. Start a company
This seems way harder, but I am sufficiently arrogant and risk-neutral that I consider it a reasonable option. In particular, this is what I would do if I decided that making money as efficiently as possible and giving it away
A. Tech company. If I had to guess based on the currently available evidence, I would guess that this is the way to maximize my expected earnings.
B. Online Education. I would like to take a shot at designing materials for online education of smart, significantly underserved, high school students.
3. Cooperate with Other Rationalists
A. Work for a rational charity. Self-explanatory. Probably worse than earning a lot of money and giving away.
B. Start a rational charity. Probably worse than supporting an existing charity.
C. Work for a rational start-up. Can't really arrange this one; but optimistically it could happen if you were prepared (someone else does 2).
4. Be a Hobbyist
I could also simply try and earn a living as quickly as possible (rather than making as much excess as possible, or having a more structured way of doing good) and do work on the side. I don't think this is a good idea.