So I did Mount Goode today. All good:

I have TrainerRoad controling the Kickr, which is in Erg mode. I also have Zwift ‘watching’ me and simulating what my power output would look like in the ‘real world’ of Watopia. It’s interesting to compare the two datasets.

In TrainerRoad, I get

which is all fine and good. I assume the speed is calculated by knowing the wheel size and the power I’m putting out at the specified cadence. I’m doing 0 mph, of course, so the speed is a calculated parameter. The Zwift data is identical in all measured data, of course:

Although I think the smoothing algorithms must be different, but the data is basically the same. Now Zwift of course calculates the speed quite differently, as it accounts for the gradient of the road, and at the same power I could be doing 1 mph puffing up a hill or 40 mph coming down it. So it must assume that the motor (me) is putting out a fixed power (which I am) and then using some infinitely-variable gearbox to match the known power and cadence to road speed. Fair enough.

Now my perceived effort is the same, of course: I’m putting out X watts on a stationary bike. In real life, I perceive hills as much harder than flats, and flats as harder than descents. I tend to pedal harder uphill to maintain speed, and I’m using TrainerRoad to train me out of that: I should be maintaining a fixed power, and so letting my road speed vary though use of the gears.

In real life, my riding becomes very wobbly below 3mph, and when riding uphill in Zwift runners were going past me. Imagine a steep hill in real life, and you’re granny-gearing it at a constant power. At what point do you just give up and push?

Also, in the simulation I’m maintaining a cadence of 100 rpm while climbing a hill. I can’t see me with a sufficiently low gear to be spinning at 100rpm going uphill at 1mph.

Basically, the question is: how slow can you go without falling off?