I found time to do three short rides with C. Cadence across Sand and Sequoias on Saturday evening, doing a hard close of the Zwift app between each activity:
- With Sterzo Smart steering engaged
- With steering deactivated (no steering icon visble)
- With Sterzo Smart steering engaged again, but not moving the handlebars
There’s a perceptible difference in how the avatar moves in relation to other riders (none of whom were using steering devices on this occasion) when with and without steering. Definitely jerkier movement with steering.
I reckon I reproduced the effects you have in your screenshots.
When the group was large, dense, and filling the available width of the road, my avatar under control of the Sterzo Smart was generally pushed off to one side of the road and advanced while being angled in towards the bunch, despite me not turning the bars. A bit ugly.
When riding in a similarly large bunch without steering, my avatar faced straight ahead, gliding gracefully and often gravitating to the middle of the group.
There were times when using steering that I found that I was able to melt into the centre of the bunch quite smoothly, but that required some subtle manoeuvring and only worked when the group wasn’t at capacity side-to-side.
On the third ride, where I tried to ride with steering activated but not physically steer, after a while I think that the avatar started to move less jerkily. I think it was more easily pulled into the churn just like when steering is absent. But that could all just be confirmation bias on my part.
All this to say that I don’t think that there’s a problem with steering in PD 3.0 per se. It’s certainly not inferior to steering in PD 2.5 IMO (or whatever we have been using until now). I suppose the limitations of the steering function when layered into the Pack Dynamics algorithm mean that there are compromises in the fluidity of avatar movement that we will have to live with.