RChung

Many people claim that small-small combos have higher losses because the chain must bend more; however, I'm not sure of this. What I am sure of is that for the same gear ratio, the same cadence, and the same power level at the crank, chain tension is lower for big-big combos than small-small combos. Power is transferred from crank to rear hub over the tensioned top run of the chain, not the return run.

genejockey 

They've been suckered by Big Big Ring.

RChung

To address the original original question about max torque, here is a summary from Kautz's classic data on pedal forces. These data show the mean of (I think) 14 "elite" cyclists' crank torque, collected from an instrumented pedal. This is for a single leg so, roughly, the mean power for these riders was about twice what you see here, or around 380 watts. What you'll see is that max torque is roughly four times the average torque over the pedal stroke for one leg, so max torque is roughly twice average torque for both legs.


Although different riders will ride at different powers and cadences, across other riders this rule of thumb appears to hold as roughly true: max torque during the pedal stroke is around twice average torque. If this holds true for world-class sprinters (I think it might, but I'm not sure) then if you know a world-class sprinter's power and cadence you could calculate what the average torque was, then double it to get max torque.

Speaking of world-class sprinters, here's a link to a video of Francois Pervis winding up for a sprint. Notice the bottom bracket.