No, what I’m saying, is that as far as I could understand so far, the phantom draft is only shown on Sauce because the updates are not as frequent as they should and as they are really inside Zwift.
I haven’t yet seen any accelerations that are out of what they should be. I think there is a mix of cases, where the RP accelerates due to drafting someone in front and slight negative slopes AND/OR riders slowing down because they let off the gas and can also be slowdown (potentially) by PD4 system, and it gives the impression that the RP is speeding away.
I may try some testing in ERG to validate (while trying to find a group as small as possible). Would try to sit behind the bot and not overtake at all. Theoretically constant watts behind shouldn’t overtake so bot accelerations if nobody is in front would then have to be explained by small gradient changes (but then chasing in ERG you should catch those too albeit with a small delay as you cross the gradient later).
fwiw the refresh rate standard in Sauce 1x per second. obviously latency of game plays a part, but just so you’re aware if you aren’t already. this would explain the refresh difference between my stats on sauce versus the pacer bot (where pacer has a delay in data feeding to sauce)
In the video I’ve seen above, the user has it set at every 2 seconds…
Theoretically constant watts behind shouldn’t overtake
Yes they should. If you are doing constant watts behind someone while drafting and don’t reduce your power you should overtake the person in front assuming the power + draft is enough for the speed increase.
If front is constant on flat ground and you are behind at a constant power (albeit lower than the rider in front) and that power is low enough to not induce overtaking then you should be able to sit behind. THat’s basically how draft should work in theory. Sorry I wasn’t clearer the 2nd rider watts would be lower not the same as the front rider. If they were the same then I agree the 2nd would overtake and there would be natural churn
So if this latency thing is really the case why would it be so bad where it’s multiple seconds where we notice the delay?
I suspect this is possible from before too; but due to PD4 dynamics it wasn’t as noticeable, especially with the faster pacers. It’s probably impossible to see with the pacers who have huge groups where the pacer never overtakes the group and loses the draft so the speed is consistent because the draft almost never changes even taking latency into account
I’d think it’s feasible and i’d be happy with tha as answer. Fascinating if so. I raised the question out of curiosity and what I saw in game and thought it felt unnatural.
This kind of goes against what I mentioned though with the auto-braking. I realize you say “it’s the same”, but … in this exact case it shouldn’t be. It should take a pretty drastic effort to push through a group, and it … does.
However, there’s quite a few people here including myself who have seen a robo fly through a several hundred person blob like a snowplow.
For what it’s worth, this would be REALLY easy to have evidence of if that video recording feature finally made it to the rest of us haha!
You can change the refresh rate within the application. Increments are 1s apart and lowest is 1s.
I did actually have it at 2s for the first part of the video but at 19:45 you can see me reduce it to 0 (1 is not the lowest). For the clip I showed, it was running at the 0s setting - basically displaying the data the instant it comes in.
Fair enough. Even with Sauce ui refreshing every second, there seems to be some network lag for the draft information that takes a while to update. And that also for other riders not only the RP.
I did a ride with Constance on ticktok april 6th, and again today, april 14th, on tempus fugit. Both with PD4.
(FYI, my zwift updated april 12, from 1.33.6 to 1.34.)
Before todays ride, i didnt know about this discussion, so i expected about the same ride when i joined.
I was immediately surprised were cruising 46kph in the flat parts. By “flat parts” i mean every part Zwift showes as 0% (yes i know that could be +/- 0,8%).
Both this rides was with a group of 7-10 riders most of the time.
On ticktok Constance did most of the pulling i the group. Today she was at the front of the group at least 98% of the time, so cant blame other riders for the high speed.
April 6th we avg 44,2-44,3kph on the 8km flat part.
April 14th we avg 45,8-46,0kph on the same 8km. (in the same direction, group-size, and only Constance pulling the blob, at 4.2w/kg)
Today we almost at no point did below 45,0kph, wich is well above the avg from last ride.
I also observed Constance just ride of the front of the blob several times, with her 4.2w/kg. (wich i can see more have mentioned above)
For 50km, i was trying to understand what was going on, but this 2 rides did not feel the same at all…
It felt like Constance was on a invisible TT-bike, or had someone in front of her pulling at well above her 4.2w/kg…
No problem riding a bit faster, but something MUST be different from last week 
1 Like
I ride with someone who sits on ERG (4.4-4.6) and the experience is nowhere near the same even factoring in height/weight differences. I don’t get nearly the same surges with my buddy that I do with Constance, even taking terrain into consideration (we’ve done Tempus regularly too and don’t get this effect). With Constance it feels like something else like actual game data lag for the Robo Pacer.
Thanks for another report. That just adds a bit more to the mystery.
- The pacers game build is the same for around 2 months, so no change there compared to last week.
- PD4 is also in effect since the 4th of April, so that also rules out the change. Also no parameters were changed in the PD4 configuration.
- Some other reports claim the problem to exist for some time (months?)
- You mentioned that your Zwift updated to version 1.34 in the meantime, but no changes were made to the physics code between 1.33 and 1.34
I always consider other riders data in Sauce to be 1-2s old and not a reflection of what’s happening right this second. Especially with all the other variables going on in the background as you mentioned. The only reason I give that clip any credence as to what might be going on is that there is a good 10 seconds where she is clearly riding in clean air but also reporting significant draft/watt savings. If it were only a second or two or there were other riders in the vicinity, I would discount it.
I hope you can figure out what’s going on, certainly appreciate the time you’ve put into it. I don’t think slight declines etc are enough to explain it.
1 Like
Agree that the game-update on my computer should not have anything to do with how the RP rides
Here are the two rides for comparison:
strava com/activities/8840669697/overview
strava com/activities/8889352941/overview
To ride the tempus 17,6km TT solo at 46-47kph, you would have to avg +/- 5,5w/kg. (4,2avg on a TT-bike would give you 43,x kph avg.)
When the blob is only 7-10riders, on roadbikes, and just Constance pulling at front, 46kph avg seems strange at 4.2w/kg. The ZI-PD4-speed-test, shows a solo rider at 300w will do the same speed as a 4 rider TTT with front at 300w, and 3 riders at the wheel, only drafting. Kinda like when Constance solo pulls 5-6-7-8riders trying to draft easy.
Good luck on your mystery 
3 Likes
This is right.
What’s strange is the first week PD4 went out Constance was on Tick Tock and my average power over 50km was 210w - I was able to stay with her without difficultly. 1 watt difference between PD3 and PD4 on same course and same group.
It’s only this week something has changed and Constance is going much faster, I had 229w average on Tempus.
1 Like
It seems all pace partners require you to put the same w/kg or .1-.3 lower w/kg to stay in the group in pd4. It use to be you could draft a group and be 1-1.5 w/kg lower in the group…
I think my pace in a pace partner ride had always been around 0.1 lower at Coco/Maria level.
1 Like
Every rider on a route will effect the speed of the RP…so if they passed more riders the speed would be higher because of more draft. The only way to test RP is in an isolated event.
1 Like
My ride with Coco yesterday evening at 20:40CEST with 90 in Group…i’m 80kg at 186cm and i did 2,4375W/kg with S-Works Venge + DT Swiss Disc
Dual recording: ZwiftPower - Login
Could definitely get by with Constance at 3-3.4 before. Now it’s 3.6-3.8. Looked back from February last time on Tempus for comparison. 70kg / 170cm. Might need to work on your drafting 
2 Likes