Short version: the power required for me to stay with Constance is oddly low.
I’m a 65kg 185cm tall rider who rides over 5000 miles per year on Zwift, frequently with Robopacers. I find the principle of allometric scaling holds true: at 65kg, I need a higher wkg to keep up with the Robopacers, which are 75kg. This is true for Coco, Maria and Miguel, where I do most of my Robopacer riding. Whenever I ride with Constance, however, my wkg is much lower. Today I rode 20 minutes with Constance (and Magnus Cort Nielsen!) on a relatively flat course (Tick Tock, I think) with an average power of 219w, or 3.37wkg. I’ve experienced something similar each time I’ve ridden with Constance, though today’s 20 minutes was admittedly the longest I’ve done. Based on rider physics, I would have guessed that my power should have been 4.3wkg. Any theories why my wkg is so low? I realize the sample size is too small. I’ll try again next week.
Pay attention to where the RP is compared to the other riders.
If the RP is in the wind and you’re in the draft and you’re on a level road then you’ll need to be doing maybe 100 watts less than if you were out of the draft.
I’ve jumped from Coco to Yumi and found myself needing less power because I’ve gone from huge blob to small blob. Similarly, riding in Makuri with Coco usually leads to lower average power than riding with Coco in Watopia.
It is because you are not affected so much by the Zwift “pace dynamics” messing with your actual pace. The speed and behaviour in the small group of 15 in Constance group is much more natural than the massive autobraking effect in the Coco and Yumi 250 rider mega-groups (which are also very high speed).
I was with Constance group yesterday and had average power for the ride about 220w. Day before about 230w with less kilograms than you. I can fairly frequently ride with Constance group as I have the strength these days to do so and to push the pace and tow the robopacer along.
The Constance group is the best one to be with at the moment because it is least impacted by the stupid pace-dynamics changes which I have never liked and still dislike.
You’ll also find that the upper PP are in the wind with very few or no one in front of them pulling them along.
Ah, but the RP can’t be towed along faster, they only go at a set pace. (repeat, repeat). 
You make a good point, aside from when Mr or Mrs Dot joins the group and sits on 4.7w/kg for the entire ride (at 120bpm heart rate) there are very few others who can easily just sit in front of Constance and stay there (which has the effect of speeding up the group). That’s why the pace is so much better, it’s as advertised.
The old Amelia Anquetil bot from the old days at 4.2/kg steady and 65kg was also a really great workout. No matter what the gradient, it was 4.2w/kg. And that bot rode proper hills - even Epic KOM and Radio tower! There was always a good turn out of riders. That used to be my daily ride.
On the topic of towing along pace partners faster (which we have been told cannot happen), we got a chance with Constance today on Big Ring course to see that.
A European Z-power rider is sitting at the front with these power figures:
Around 140-145bpm
Power between 330-400w the whole time
Surges of 25-30sec at 700w+
Over a time period of 1.5 hours
When he did another sprint and dropped Constance and the group of pretty fast riders, our speed went from 45km/h+ back to 40-41km/h.
My power average also dropped by about 10 watts. Pretty big change.
It did ruin the pace, people were getting dropped all the time and I’m fairly sure one guy had to keep teleporting back as a result. Not very nice for him.
Have we? By whom?
Haven’t you been here for long enough to see?
Maybe I’ve subconsciously erased it from my memory for my own sanity.
It seems obvious that if you stick a rider (or group of riders) riding at 40kmh in front of a RoboPacer doing, say, 195 watts and as a result the RP is getting a draft benefit to the tune of about 100 watts then the whole group is going to go significantly faster than it otherwise would.
Less of an issue at lower speeds (less draft benefit)
More of an issue at higher speeds (more draft benefit)
Is obvious, no?
I’m pretty sure that we’ve never been told that a oacer can’t be pulled along faster by other riders, only that it can’t put out more power than it’s been programmed to do according to its dynamic pacing.
This is not true.
Who officially said that?
The set pace is defined by their power output.
Pace partners are enabled to draft and can be drafted like ordinary zwift riders and therefore can be pulled along by a faster/more powerful rider in a draft.
Read from 270 onwards:
Even a post claiming that if there are heaps in front of the robopacer, it doesn’t make a difference to your power. Actually it does, a lot.
If everyone stays behind Constance then the average power will be lower. If a group of people start pushing the pace and dragging the robot faster, the power requirements goes up a lot. Then when they clear off, it goes down.
In that topic one person really hit the nail on the head but still people disagreed.
So your comments I have cited are ironic?
You can test this on the temp fugit course with each pace partner. I’ve been in a ride on that course that cleared that course faster with Genie than with Constant. It’s has always been that way since getting rid of Amelia Anquetil and the other constant PP
Why not to place pacer bots on TT bikes. They do not experience draft, however any other rider on an non TT bike can draft the bot.
See Drafting in Zwift: Power Savings, Tips, and Tricks | Zwift Insider.
This would be an excellent idea. It has been raised numerous times but will probably never happen unless an angry mob comes along beating down the doors at Zwift HQ.
Other solution is to get a ZPower trainer and start burning up the Constance group with 6.5w/kg for 100km. Since that seems popular and nobody seems to care about it, might as well join them. 
And use a proper trainer for workouts only.
Would it? Let’s say the RP does 195 watts @ 2.5wkg and there’s a 50 rider blob.
On the flats (at similar weight) I’d maybe be needing about 150 watts in the draft. Uphill I’d need maybe 210 watts. Downhill I’d perhaps be needing about 100 watts.
We should try it out, Chris. I’ll draft. You lead.
Yep. This has been suggested repeatedly and nobody from Zwift will even respond to it.
Since the TT bike idea has been totally rubbished above, the other more forceful way is robopacers and the riders that join them ride in a parallel world to normal riders - you only see them if you join or teleport.
And the robopacer has an invisible barrier in front of it that nobody can pass, or if they do they get more resistance.
Currently the robopacers are just too influenced by flyers towing them along.
Another method would be the robopacers have a maximum speed they will not exceed for their pace level.
It would be like me riding IRL and saying if you want to go faster then you can, but I will maintain a particular speed and won’t go faster. I’ll stay with the rest of the group.
not my area of expertise as i rarely (if ever) ride with the robo pacers, but why would flyer only drag the robo pacer and not the other riders too?