The upper limit of any heat engine (turning heat into useful work) is called the Carnot efficiency and is limited by the temperature differential. It's usually not more than 70%, though for a Stirling engine, a few percent would be considered excellent efficiency.

A Stirling engine easily needs to radiate 20 times as much heat as it produces as power.

> for a Stirling engine, a few percent would be considered excellent efficiency.

Wikipedia says "Stirling engines have a high efficiency compared to internal combustion engines, being able to reach 50% efficiency," which seems to contradict that. What's the discrepancy?

Yeah, even thermoelectric piles (Peltier effect devices) can easily get ~~double digit efficiencies. Not sure what that guy is talking about~~.

Thermoelectric piles are basically the worst form of generating electric power with heat (it has 0 moving parts, so it's used in various places), and it's still not that bad.

edit: Never mind, apparently they are worse than what I thought they were, at 5~8%. But they are still the least efficient way to turn heat into electricity, and yet they can achieve what that guy said.

Yes. Though cooling sails are relatively feasible. Assume you can cool to ~250K: the back of my envelope and a Stefan-Boltzmann constant cribbed blindly from wikipedia tells me that such a sail would radiate ~200W/m^2, which is very doable for a spacecraft with a 1kW power budget.