I was in the velodrome yesterday and was cruising around at abut 55kph, pushing around 480 watts or so. However I then started to push around 850 watts and the speed only went up to around 80 kph. Dropping back to 40kph I was using only 250 watts. So this is definitely not a linear progression in a power/speed graph. This would indicate that there is an increase in resistance (in this case wind resistance) the faster you go. Otherwise if I double the power, I would double my speed.
Don't forget that Power = Force x Velocity
Or, Velocity = Power/force
Now, force = mass x acceleration
So, velocity = power/(mass x acceleration)
If there were no adjustments to resistance, this would mean that the accelerative force would remain the same. The mass would also be the same (although some of us can sweat out a fair amount!). Therefore you would have a linear graph of increased power being directly proportional to the resultant velocity, which I do not think is the case.
It may be that the additional resistance added due to higher velocity is too little at the top end. The power/speed graph should be exponential and it is quite possible that this curve is too shallow, which I think is what
@bridgy is implying. It may be that they need to adjust their algorithm slightly in this respect which would be more representative of the greater effect of wind resistance.
Make sense?