It’s not just your imagination. I’ve been studying this phenomenon for some time, including two 50km rides with C. Cadence just this week, during which rides I never departed from the 15m “drops zone”. Even though skills DO build in keeping in the PP peloton, after months of riding during which I developed the necessary techniques to stay in zone, I still find I must vary my power output through 100-110 watts in order to do so. Invariably, my weighted average power at the end of such rides comes in just below Ms. Cadence’s 2.5W/kg (I’m taller and heavier than she). But I ride with B. Brevet, too, and the same thing happens That’s how “averages” can be so misleading. You may have noticed that sometimes the PP will sometimes be located at the front of the peloton, in the middle, and at the tail end, and that these relative positions will vary continuously. The behavior was quite confounding to me at the beginning, and though still confounding now, I’ve developed hypotheses to help me explain the weird behavior necessary to stay with the PP:
The size of the peloton (and network/processing latency issues with processing large groups);
The relative skill level of the peloton constituents;
The size and frequency of encounters with other, “non-joined” groups and whether they be faster riders overtaking the peloton, or a slower group being overtaken by the PP peloton;
The terrain geometry and equipment selected by the rider mix.
The Pace Partners are definitely being affected by the so-called “game physics”, including the equally bizarre “sticky-draft” effect. Riders overtaking and blowing through the PP peloton can have strange consequences. Likewise, lower-skilled cyclists who are unable (or just not paying attention) to maintain a steady power output, have a tendency to rubber-band through the peloton, which then causes the PP to be so influenced. When one or more riders in the PP peloton suddenly pull ahead of the Pace Partner, this has the effect of accelerating the PP herself! She is then slingshot to the front of the peloton, where she is now breaking the wind and tends to slow again. Rinse Repeat.
You might find it a fascinating experience to join a Pace Partner ride using a Time-Trial (TT) bike! However, given what you’ve said above, I would recommend entering the group below your FTP level at first. I’ve suggested that Zwift should consider running the Pace Partner such that the game “physics” do not apply, in order to obtain a more consistent ride experience. That said, I’ve come to develop a appreciation for the random power changes I must employ to stay attached to the PP Peloton, and armed with knowledge of what to expect, now use these rides as a crucial component of my structured training.
I’ve had such abysmal results with so many group rides (races???) in Zwift (those which simply and consistently ignore their own group-ride pacing descriptions), that I’ve been keen to see the PP programs be dramatically expanded. Clearly this is quite possible, given that Zwift deployed costumed Pace-Partners on various courses for Halloween! So, PPs with no game physics. Pace Partners on every route. Individual (PR) Pace Partners. And more. These are all VERY doable…
You might have noticed that Zwift has taken my suggestions for improving the relative position animation of the PP in-game, which is extremely helpful (KUDOS TO ZWIFT!!!). I often use the drone view (keyboard 9 view) so that I know not only where the PP is, but also her rate of closure (or distancing) on my position. Knowing how these Pace Partners behave is an important tool in the training kit, and I just wish the scenery would soon change. Clearly, whether they’re on hills or not is still NOT going to change the dynamic you describe that already occurs on the flats. Look at my power stochastics from last night! As you can see, I’m in my second full round of x2.0 drops multipliers - never having been dropped or blown out the front…