Economics. The altruistic motive to help defend the blockchain, in a crisis or otherwise, cannot continue beyond the willingness or resources of the altruistic individual. Excepting such a crisis, the userbase willing to mine at an ongoing loss is always going to be a vanishingly small percentage.
I tend to think that dedicated miners will have more computational power than the entire rest of the Internet, which would make the chance of volunteer takeover low. But if this is not the case, I can easily imagine a network dominated by volunteers. There are a lot of people who compute for @home projects, and it's in the interest of fee-charging pools to spread propaganda about how mining helps "keep your money secure". It's not even altruistic from the perspective of these miners, since they believe it will help a system they rely on.
Beyond efficiency, volunteer mining frequently also centralizes control with pools and gives some power to people (the miners) who don't really understand Bitcoin. It moves us away from my picture of the optimal end-game: a few hundred private, competent, for-profit miners operating outside of a pool.