And after going through the article, I have a number of disagreements.

The author writes that the human leg is not capable of turning cranks at constant velocity, and that the freewheel "decouples" to allow this at the bottom of the peddle stroke. This cannot be correct as it is possible to feel when this happens as the result of a crummy pedal stroke -- the rider notices that the crank has to turn against no resistance for a fraction of a degree as the cranks "catch up" to the freewheel.

Moreover, even if this argument IS correct, there would still be no power or speed advantage from riding fixed. In fact there would likely even be power loss as power is robbed from the drivetrain in order to carry the leg throught the "dead spot" in each pedal stroke.

The author then goes on to argue that the fixed gear allows a 1/3 longer power stroke and provides 50% more torque and acceleration. This is disproved not only by results, but also by hard numbers. Those familiar with Powermeter numbers from road and track racers know that they are essentially the same with individual athletes.

The author, IMO, is looking for complex answers where simple ones suffice: Why are fixed riders sometimes faster on hills? Because they have to be. If you've got lots of gears you have a decision to make: do I grunt or do I spin? But if you've got one gear you're either going to turn it or you aren't.