http://lesswrong.com/lw/3gj/efficient_charity_do_unto_others... "Just as there is only one best charity, there is only one best way to donate to that charity. Whether you volunteer versus donate money versus raise awareness is your own choice, but that choice has consequences. If a high-powered lawyer who makes $1,000 an hour chooses to take an hour off to help clean up litter on the beach, he's wasted the opportunit…
Sure, that's a selfish choice to some extent, but I think it's a reasonable one. Some people choose to live in one city over another just because they like it, or to work at Google rather than Wall Street because they like what they're building there more. Some people will pass up a job for one that pays 20% less, because it's building an exciting new app they want to help succeed. Wanting to work on a social-justice project just seems like a variation of that; it's quite common and natural for people to want their day job to be something they care about, if they have the option.
I think the argument as given also assumes that charities are currently operating optimally, and there's no room for improvement or disruption. If you applied the efficiency argument to startups in general, it would argue that you should work a well-paying job and pay other people to do startups. When would it ever be beneficial to do one yourself? Only if you think the market is not already efficient, and your personal intervention as a founder could change things more than working your current job and just funding startups would.