Road Bike 46:00 
The Zoo-Pfaffhausen-Witikon loop. I could not forget my low VAM values on Sunday, so I did some more field research on the second climb to Pfaffhausen-Witikn; I rode the first part (roundabout-junction to Pfaffhausen in about the same speed as last Sunday. This part is significantly steeper, about 7.9% on average. And voilà, VAM on that (very short) part was 1450, much higher than the 1220 achieved on the full climb. So the steepness does def. make a difference in VAM values. When I once don't know what to do, i'll try to refine the formula by incorporating gradient and duration of performance.
Btw, Michele Ferrari has apparently also followed the Vuelta stage to Lagos de Covadonga (Monday) and calculated the performances of the leading pack: Since the first part of this climb is rel. steep at 9.1%, rather short (8.6km) and has only a few curves, achieved VAM are among the highest seen in cycling. This year, Mentchov etc achieved 1823 while Sastre even did 1900 (!!) last year...
Ferrari then goes a step ahead and calculates the performance in Watt on that climb per rider and comes up with a 10% higher output by Mentchov compared to Sastre (Sastre is lighter but achieved the same time). Interestingly, Mentchov's performance on the time trial last Saturday was 9.6% superior compared to Sastre's, so about the same difference as on the climb. Ferrari measures the performance in TT by squaring the avg. speed to account for the properties of velocity measures.