One problem I’ve encountered is trying to find the right tempo to pace other riders, both human and AI. It seems you make an effort to catch them, then end up blowing past them and it’s hard to adjust your speed to match their effort.
I’ve found that it’s much easier to pace by changing gears than by trying to slow your cadence briefly. I find that when I try to change cadence, there is a lag in the effort and the rider I am trying to pace just opens a gap.
As you approach another rider and feel yourself about to overtake them, click back up a gear and maintain the same cadence. You should be able to slip back in behind them easily and then adjust gears to maintain pace all the while trying to keep cadence steady.
This technique works best for me, too, Greg. However, I have also noticed that the AI riders sure do blow out the back on climbs, only to mash hard on the downhills and come roaring back past. Being a rather large man, I think I understand their technique!
I think that the right approach will be to fix the issue properly. I know it is probably difficult since we are dealing with variable accumulated lags in here (network, ANT+, power averaging etc) and this interference can and will introduce feedback that can range from negative to positive and positive feedback is definitely no good.
However I also use BKool software (their 3D and internet racing part) and while not completely free from peculiarities somehow it is way easier to pace in there. So somehow they did manage to nail the subject.
I have found pacing in Zwift virtually impossible. Today I had more than one occasion where I was taking it pretty easy and an AI would pull up next to me. I would then have to almost completely stop pedaling to get them in front of me, but once there, they would zoom ahead. I could double (or nearly double) my output and not be able to close the gap. That just doesn’t come close to reflecting reality. I also did a studio Computrainer race once, and the drafting was much easier to dial in, and the “reward” from drafting was quite noticeable (as in real life).
I would be very interested to hear comments from the Zwift folks on this topic.
Some of the AIs are definitely stand-offish. They will pedal past you, tease you to jump on then drop you like a hot potato. I still see this but I do manage to find a few that along that way that I can ride with for a bit. Typically I drop them at any hill. They must weigh about 225 as they all seem to slow greatly on any incline then speed away on the descent. I also wonder if they get friendlier the longer you ride.
I have less issue with the absolute power output (not that I can hold 350W for very long) because I find that some AIs are stronger than others. What I find most frustrating is that their output seems to vary by a factor of more than 2 or 3 in a very short time frame. I would think that you should be able to draft an AI for a complete lap without too much trouble, and I have never come anywhere close to being able to do that.
Haven’t tried with real humans, yet, but I would like to.
Kostya - have you done some group rides? Is drafting a human just as difficult as drafting the AIs?
Finally got the drafting somewhat down today. I had to do view number 9, over top to get everything working right. I even dropped a couple AIs on the climb. It was a little tough on the climb to get in and stay in the draft, but that’s fairly realistic. On the flat it wasn’t too bad.
Pacing humans just takes a little practice. It’s not that hard, you just have to get the feel for it. Use my tip of downshifting to keep from over taking.
I agree pacing is rather tricky…I don’t really understand the “close the gap” widget really…does that mean I’m drafting? When it says “ok” I’m usually so close to the person that I have no option other than to overtake them…so the drafting or not question is there for me as well. I’ve noticed that if you have your speakers turned up, whenever you are getting a boost (like drafting or a downhill) there is a wind noise that comes over the speakers. This seems like a nice audible way to tell the strength of the boost that is being given due to going downhill or drafting, but a visual method would be good as well…