I’ve been watching Coco meticulously on Wandering Flats and can see every lap that the field stretches as almost everyone in her group goes in front.
If I turned dynamic pacing on, that wouldn’t happen.
I think I’m pretty comfortable with 10% extra being right on short uphills, but think 20% is needed on downs.
I think it was an arbitrary assessment.
Just think if all those folk stretching in front of coco group would be automatically beamed over to yumi or Jacques or genie group?
They might not stretch out too far ahead any more.
The paces of the groups are so different when everyone stays behind the robot rider. Instead of 44.5km/h average (one day this week on tick tock) the next day was less than 43 average. Same riders, just everyone had cooked legs and didn’t charge at the front.
Average power needed was very different both times too.
My most recent experience with Robopacers was that the increase in power on inclines was insufficient and the decrease on declines was massively overdone.
I guess the individual experience depends somewhat on rider weight, but also on Trainer Difficulty setting. It may also depend on the RP selected. Mostly I’m riding with D pacers. It might be different higher up the food chain.
I have TD set to 50% and I will ride most of the courses I choose without changing gear at all. Obviously this means my power increases far more on inclines than the Robopacers and I really have to hold back. But at 95kg recently (now down to 90 :-)), and on a fast frame and disc wheels, I have to soft pedal on descents to avoid dropping the RP.
I would prefer +20% on inclines and -10% on declines. Better yet, sliding scale with gradient. e.g. 0.2% per degree up and -0.1% per degree down, or something like that.
If people want a completely steady state ride with absolutely no relationship between gradient and effort then i recommend working out with erg. RP rides should be at least a little dynamic, just like real roads outdoors.
Just my opinion. 
With Constance, it’s 315w average on the flat and 345w (4.6w/kg) uphill, with some of the quick guys at the front pushing more than 6.5w/kg uphill and 5.5w/kg on the flat at times, enough to drag the group along much faster. It’s quite amazing they can keep that going for 50-80km but it makes it annoying for the folks trying to hang in with the group.
I definitely don’t need the A grade robopacer going even faster uphill. Maybe give the C and D grade bots 4.0w/kg performance uphill to make it more dynamic if those folk want a bigger challenge. 
I couldn’t imagine if we had a virtual version of the climb in your avatar and the Constance bot pushing 345w all the way up that climb. IRL when I did it we had some fast guys there and they wouldn’t be doing that pace - although admittedly they are doing 800km and 20,000m+ in the week.
Very likely to get a different pacer experience to suit your needs on the day with the Makuri vs Watopia pacers too. I had Constance to myself for 3 hours in Makuri earlier in the week and it turned out like a Yumi ride in peak hour Watopia. 2.9w/kg (190w) and 40.6kph avg.
The groups in France at the moment are better than the ones in Watopia, the groups are slightly smaller and just the way you work in the group seems a bit better than in Watopia.
Not to mention the French routes are more enjoyable. I can guess those will disappear after TdF time, but I hope they don’t.
They move to Scotland next month.
Observed yesterday and day before that occasionally on transition to hills from flat sections Constance robopacer will absolutely launch away like a rocket.
Typically on Watopia KOM going from Downtown direction. Seems not natural.
The group responds with 5.5-7w/kg to not get dropped.
I just finished my first robo-pacer ride with D. Bernie.
I was absolutely struggling to maintain the very strange, seemingly random power/pace. I finally managed to get to the 2.5x multiple, but I must say it took a lot of coasting then sprinting to stay within the pacer zone.
I think something just isn’t right with how this is being implemented. I highly recommend reviewing the design and implementation of the robo-pacer features.
I’m planning on trying the C, then B pacers this week to see if maybe the issue is just poor choice on my part.
I think the issue is when you get mega groups of 100+ then the pack dynamics goes silly.
When robopacers were on France world the behaviour was better with the smaller groups.
I don’t know how experienced you are at modulating your power to stay in the draft, but that has a big impact on the robo pacer experience (and it’s great practice for racing or group riding). It took me a few rides to adapt my riding style to stay with the group and near the pacer without big swings in effort. If you keep at it I think you’ll find it gets easier. If your weight is significantly different from the pacer (75kg) or outside the norm of others in the group, that will also force you to pay closer attention to speed and power. My trainer has some delay in reporting power changes, which means I had to learn how to apply power and back off power in anticipation of the group speed changing and the effects of terrain.
I do real group rides at about 2.3-2.5 w/kg nd have no similar issues.
I completely agree. Zwift is not like a real group ride. The more experience you have with real group rides, the more unlearning you have to do.
Indeed, I don’t really expect a true cycling experience.
I just don’t understand why the power was so variable. Seems wrong.
The pacer’s power is not highly variable - it varies in predictable ways based on the terrain. But speed can vary a lot based on how the group behaves and where you sit in the group. That’s why it’s so important to make small changes in power to maintain your position and avoid changing how the draft affects you. Letting a gap open up means you sprint to get back. Rolling off the front and backing off power means you may be overtaken by a fast-moving group and end up doing another sprint. Trainer responsiveness makes it harder to gauge whether you are going too easy or too hard until you end up doing another sprint to chase on. It does get easier with practice.
I saw the pacer power change between 1.2 and 1.6 w/kg. During the 75 minutes I rode with D. Bernie.
As a percentage that’s a huge variance .
For my real road group that would be like 2.1 to 2.7 . A difference of nearly 50 watts in my case.
Ride with Maria. I suspect you need so little power to ride with Bernie that it’s hard for you to modulate your effort. A couple of decent pedal strokes and you’ll leave Bernie in your wake as she’s well below recovery watts.
So on your real road group ride there’s less than 50 watts between your hardest climbing pace and your easiest downhill pace? Really?
If I remember correctly it was not intended to be realistic otherwise you could coast easily. They wanted to force you to keep working, no rest.
Also, not sure the experience needs to match IRL absolutely exactly. Over time I’ve “learnt” how to adjust power/steering, albeit never stay perfectly in position for the whole ride. But it keeps me distracted enough to see through 45-60 mins of indoor cycling without too much bother.
Is it realistic? I don’t actually know as I rarely scrap round London at 45kmh within 5cm of 100 other cyclists doing the same. For that matter do F1 drivers play their sport on a PS, and do professional footballers play FIFA and complain it’s not realistic?
