I did a very similar calculation in an article for the oildrum a few years back.

I agree that in order to compare apples to apples, you have to measure human output, not input. I used the wattmeter on Floyd Landis' bike (the Tour de France winner accused of androgen use), which registered 230 watts average for the whole riding time of the Tour. I divided by 2 for us less-in-shape people, and then assumed you could put out that power for 6 hours a day on weekdays (not counting breaks).

On the barrel-of-oil side, I used 20 gallons of gasoline from a 42 gallon barrel. Once again, you have to measure useful power output, not input calories, so I assumed it was being burnt in a 25% efficient engine in a car or piece of oil-powered equipment.

That works out to one barrel equals one year of human work. Of course, a human with a brain (which, incidentally, runs on 5-10 watts) could use that power output more strategically -- e.g., by pulling individual weeds instead of a plow -- but sometimes, you just need the power output straight up -- like when a load of concrete is hoisted up to the top of a building, or old concrete is crushed into gravel.

In the article, I also noted the difficult situation with non-renewable helium, given that I mostly do MRI for a living.

Marty