Machine Elf and Richard Pearse are arguing that as I walk up the down escalator (or Stairmaster), I’m pushing “down” with all my weight on the steps, and this is driving the steps down against the resistance of the eddy brake – thus equal exercise to walking up some stationary stairs. And they are arguing: If that’s not the case, then what force is driving the steps down against the resistance of the eddy brake? Whereas DSeid and I are saying that’s going to happen even if just park my 160 pounds on one step and stay there.
So where is the discrepancy? Do we have a paradox here? Cognitive dissonance much? 
I spent all last night thinking about this. And recall, I’ve never used a Stairmaster nor any similar device, nor ever even seen one in action but briefly (but I do climb up fixed stairs from time to time). So I’m doing some guesswork here.
One of the scenarios is that of a user continuously walking up while the machine continuously moves down, at equal but opposing rates, such that the user remains vertically stationary. This is the case that we haven’t settled yet. Recall that a few posts up, I questioned whether this scenario actually ever happens.
Here is my speculation:
We’ve established that the Stairmaster has a control to set the resistance of the eddy brake. I’m guessing here that if the resistance is set higher, then when the user steps up to the next step and straightens his leg, the step will descend slowly, taking longer than the time it took for the user to straighten his leg. Thus, the user goes up, then down. When he goes up, he gets up-hill exercise. When he goes back down, that energy is transferred to the brake. The constant-COG-position scenario does not happen here.
If the resistance is set lower, then when the user steps up to the next step and puts his weight on it, the step descends faster. In this case, the user rises less than in the high-resistance case, or possibly not at all, before the machine lowers him. By rising less (in the intermediate case) or not at all (in the lowest-resistance setting), the user gets less up-hill exercise with each step. And, correspondingly, because the brake resistance is set lower, less energy is taken to drive the step downward against the brake. In the limiting case, the user gets NO up-hill exercise, and the energy transferred to the brake is minimized. I think this case must approach the situation of walking on level ground, as I have suggested earlier.