I was sort of wondering about the human power aspect myself. From spring through fall, I bike the ~16 miles between my house and the lab several times a week. I can propel myself at an average speed of something over 16 MPH (counting all the time spent accelerating away from stop lights), and on level ground can maintain about 25 MPH. I can do this without eating noticably more than I do in the winter. So I can't help but wonder just how much energy I use, and how efficient a biological converter is.
"...refining petroleum requires less energy than is released by that petroleum, making the process over 100% efficient..."
That depends on how you define efficiency. I'd think of it as the crude oil containing X amount of energy per barrel, with the resulting gasoline & byproducts containing something less than 100% of X. Then when you run the gasoline through an IC engine, you only get about 30% of it in a useable form.
The idea behind fuel cells is that the conversion efficiency is limited by chemistry, rather than theromdynamics, and so can be much higher (90% rather than 30%?). So fuel cells aren't inherently bad - what's dumb is running them on hydrogen, with all the associated drawbacks.