PD4.1 causes odd RoboPacer behavior [KNOWN ISSUE - FIX PENDING]

Don’t think racing has taken center strage though.

When will we finally have different race formats natively integrated in Zwift? I.e. elimination races, points races, team races etc…

Give e-cycling a specific identity, just not a bland copycat of outdoor racing.

Sorry for the off-topic rant.

Putting in a 2c here.

I don’t think it’s the draft and blocking etc. that’s the issue at fault. PD4.1 makes the most sense, and is most correct feeling to me.

It’s obvious there are some faults as being discussed here.

But I think it’s tied to how wide packs are allowed to get, and how our avatars are steering.

ZRL on Tuesday to me felt fantastic, genuinely the best racing Zwift has ever offered in my opinion. HOWEVER, it’s because our pack started to thin out, and wasn’t a giant fat blob like we get around RP’s and group rides.

I’m personally under the impression the biggest flaw we’re dealing with is avatars at times not being given the opportunity to pass due to the ability of groups to essentially form walls, and then turn things into wasted watts.

Wasted watts make sense to me; I know they frustrate probably a good 90% of the community; but if I’m behind someone outdoors, I’m going to be on the brakes at times, and that’s quite literally the same effect as wasting watts, ruining the momentum I had… unless I decide to get around that person.

Zwift generally just lacks that ability to get around people at certain times. Steering can do it; but only if the packs aren’t so fat they form a wall.

So personally; while PD4.1 I’ll repeat is the best feeling Zwift I think has ever offered on all accounts; but it certainly has a flaw in respect to how blobs of people are managed.

It would be worth testing a system that forces blobs into skinnier lanes.
In more laymans terms; allowing a better passing lane to always exist.

In the real world, we absolutely waste watts in groups, and at certain times, can’t [safely] pass someone.

Pack speeds are finally correct feeling, and allow people to get away like they should; without front-runners being slingshotted every 10 seconds of previous PD systems. So a newer “lane-system” should prevent folks from instantly turning into possibly other people coming around them; ie be mindful of their surroundings from cutting others off, to catch a draft… ie if someone is coming around a pack at a higher pace, in a race, your objective would be to grab on… but you have to delay yourself to make sure you can safely do so. Our avatars will need that delayed reaction to prevent what was happening in the former PD systems allowing the slingshotting.

I think PD4.1 just suffers from allowing blobs to get wider, more, than they should.


Except that in real life in a large group you can hang in at 20mph+ doing very very low watts on flat ground. That’s not what we want in Zwift but right now at the back of the pack you often have to do more power than the people at the front are doing just to avoid getting dropped. Struggling to not get dropped at the back doing 5.0 while the front was doing sub-4.0 in ZRL the other night wasn’t very fun.

hmmm… I cannot say I’ve personally experienced that ever since 4.1 went live on any races, group rides, etc.

Obviously the draft is shorter than it was during 4.0, so it matters to stay tight to the person in front of you and manage that ranging very well.

But I cannot say that I’ve experienced what you’re referring to unfortunately.

I was riding IRL in a big group today going quickly, I barely had to put in any power - I could soft pedal and stay there with maybe 100w only, sometimes having to tap the brakes to make sure I didn’t get run into the rider ahead. That’s not wasted watts, that’s saved watts.

In Zwift on the other hand I’d be dropped if I did the same. Or at least have to do more than advertised watts/kg just to stay with the group. It’s very unrealistic.

I have a lot of IRL experience riding in huge groups in all sorts of situations- races, whatever, numerous Haute Route events.


I get that; but at the end of the day Zwift is still a training / workout program with a focus on the social aspect of riding.

I 100% agree that the draft in larger groups is nowhere accurate to real world; but if it were… then what purpose would there be to large group rides, or things like the RoboPacers, where some days with 10 people you’d have to run with 170w around Coco for example, and other days you’d fly along and could sit in at 100w as you say… Where the folks up front could be pushing 300w.
Which… is highly unlikely the experience anyone here is looking for; in fact, it’s quite the opposite of what anyone [everyone] is looking for.

There’s a break line between keeping Zwift a workout platform, and making it… too realistic.

Making it too realistic in large groups would decimate the entire purpose of large group riding; because … who buys something like one of those huge Bowflex units; if they only want to do 5lbs curls?

This break from reality has to exist; or else Zwift loses an entire portion of its identity.

(and no, braking is 100% wasted energy full stop, under any and all circumstances for any type of vehicle; that’s a ridiculous claim to say otherwise; unless you want to talk about energy reclamation in hybrids and the sort [and even that is not the same, as that’s converting kinetic to potential energy by using electric motors as generators… which is technically not “braking”…])


No, we are still here listening to feedback and complaints.

So yes, If we can improve the experience of riding in inside the pack for everyone we’ll do it.

The pack dynamics topic is another one of those damn if we do, damn if we don’t. If we do nothing about it, we get complaints that we don’t care about racing, etc. etc.
If we do change things, specially in this area, we can’t automatically do a knowledge transfer on how to best ride it :confused: It will have to be adjusted or learned.

With that being said, I’ll leave again my 3 main tips to better improve the experience:

  • if you want to move up in the group, a short / sharp acceleration of 3 seconds (or more) is much more efficient than slowly ramping up the power and try to grind through the group.
  • if you want to keep your position in the group, just relax as much as you can until you start to move back from the rider(s) in front of you.
  • camera 3 is your friend :wink:
1 Like

Hi David, I like 4.1 and think it’s working fine generally. no real issues with it

but I agree with other people that robo pacers specifically are acting weird. no idea how to explain what I think the problem is as I dont log a huge amount of time with them but it was pretty immediately obvious to me when i did since i was on tempus fugit (like what, 2ft of elevation) with about 100 other people present and could not hold a steady 175w without either drifting 50m ahead of the bot or the bot shooting 50m ahead of me pretty much at random. maybe it’s something to do with how the robo pacer AI copes with a system designed around human behaviour

1 Like

I’ve been thinking about this suggestion in my last few days. The idea of “just ease off until you start to fall back” means you have to surge to recover your position and is unpleasant in a pace partner group and potentially catastrophic in a race, especially since in a race you can’t expect to just say “ok I’ll do 5-10 more watts than what I almost got dropped at” because the group isn’t going to stay at a consistent pace.

Right now PD4.1 has the blocking effect of PD2 sticky draft but without the assistance of pulling you along.

1 Like

Latest iteration of Pack Dynamics feels way off base. It should not be necessary to increase power output by 20% to slowly move up through a pack or riders.

1 Like

Only one example of a recent RoboPacer chat.

To be clear: I wasn’t stirring the pot. Multiple riders have been asking what’s wrong with the game. Most people are not racers, nor on the forum, or followers of Zwift related social media - they just hop on to ride and sense something’s off.

I expect you’re going to see more of this as we approach fall and winter indoor season in the northern hemisphere.

Why is it so hard to accept that in trying to make something “more realistic” it’s now also “less fun?” Both of these things can be true.

@DavidP Can the Zwift code support different versions of Pack Dynamics at the same time? (PD4.1 for racing, PD4.0 for RoboPacers)

There’s no AI, the bot is just another rider with a mostly-steady power output. Yesterday riding I was drifting away from a bot doing 0.5wkg less than a bot because it was being blocked and I was in the draft of someone going faster, and in the same ride I was getting dropped by the bot doing about 0.5wkg more than the bot because I was being blocked and the bot was in the draft of someone going faster.

Problems manifesting in the pace partners are also showing up in large-field races.

1 Like

Same problem with a Pace Partner B 2 days ago (3.7w/kg on the flat and uphill 4.0w/kg)
(over 50kms with 524md+) Usually it’s not really complicated to follow him, sometimes we even manage to pass in front of him. But here it was different, I had the impression of following a Partner A pace following sudden changes in pace… but remaining at 3.7w/kg while he was going faster…

1 Like

David ,
Thanks for getting back. I’ve been with zwift since the Blue men out numbered actually rides on watopia and they were the pace partners, so I’ve been around the block a few times. (~73,000 miles) and was a huge racer. I did the PD4.1 Kiss test events also. So I’m not someone who is just starting out. The main issue is with pacer bots since I ride with them 5-6 days a week and what people here are saying shouldn’t fall on deaf ears or made to work around something that’s not natural or vaguely representative IRL………

1 Like

this doesn’t work in a small fast group. i can try to keep the pace line but i’m dependent on what happens in front of me. if one of the riders in front of me can’t hold the exact pace of the bot a gap will open and everyone will be forced to chase. if you try to stay right behind the bot, people with steering going left and right - just for fun or because it’s a disfunctional beta feature - will push you out of the pace line… combined with the halfwheelers pulling the bots and the ridiculously exaggerated uphill efforts caused by the dynamic pacing it makes for a very frustrating experience, especially on longer rides. i completely agree with the opinion above and in other places in the forum that zwift just doesn’t care anymore about the feedback from its users. i’ve personally had enough for now and decided to take a break from zwift as it was no longer fun and i felt it had negative impact on my training.

1 Like

Thanks, but the tips are no different to handling PD4 well with RP’s.
However, PD4.1 in relative sense is harder and more volatile.
So: (1) the short/sharp acceleration needs to be higher and for longer (4.1 is way more sticky in large packs); (2) you have to maintain a relatively higher power when “relaxing”, and (3) you start to slip back sooner/ more easily than PD4.
And, the RP surging is more pronounced which only exacerbates behaviour (1),(2),(3) above…

1 Like

Personally I think we are in that unique phase of everyone still adapting and learning the Pack Dynamics…

If you feel you have to hold a relatively higher power at the moment, it is most likely because everyone else has not ‘learnt to relax’ yet. When everyone ‘gets it’ - which will be a gradual process - I’m sure things will settle down again. You can control your power, but you cant control what everyone else is doing, which has a knock on effect.

I love what PD4.1 has done for racing. I also ride casually (with RoboPacer’s, lead group rides and take part in other rides), and personally I don’t think PD4.1 has had a negative impact on any of these areas.

Learn, adapt experiment and have fun! :+1:


There is something obviously missing since it’s been 23 days since this thread was started. If it takes people that long to “learn to ride on PD4.1” and still having problems. Once the winter season starts here in the states, there will even more people not learning PD4.1…


I disagree; or at least, I wouldn’t say something is “missing” at least.

When PD4.1 went live globally, I was having to constantly explain to our riders how PD4.1 worked and what to look for, how to react, etc. Basically, how to treat it for about that first week and a half, as our group riders were very clearly struggling to get used to the new system, with lots of splits in weird spots (primarily false flats leading up to small descents).

Now that we are indeed 1 month post 4.1 public, things have changed dramatically; the group rides I am part of, people are doing a FAR better job of riding together, and working together than they were those first weeks. Most people are definitely learning and used to it now, that much is very clear and obvious from what I’ve watched.

That said, if people choose not to learn, and blame it on the system instead of themselves willing to adapt… then yes, they’re going to have a bad time.

As Jon said; people need to learn to adapt; if you aren’t willing, then the problem lies with [you], not the system. Is the system perfect? No, and nobody is saying it is, but it is very clearly far superior in a majority of aspects than the former systems.

I’ve said it once and I’ll say it again: People here seem to act like Robo Pacers are meant to be some perfect image solution in their minds, allowing them to ride a perfect constant Zone at any given moment. They aren’t, and they never will be due to the massive variance in people riding with RP’s, varying routes, etc.

If you want perfect zone riding, set up a workout and run ERG. Robo pacers are never going to be a perfect pace for everyone, at any given moment each day.

On flat, (generally more popular) courses, more, heavier, high wattage people will join the bots, and pull the packs to [consistently] higher numbers.
On stranger, more hilly routes, there will be far less of a populous around the RPs, and the pace will be highly variable.

Once folks get it out of their heads that RP’s pace is unaltering, can be biased heavily, by the size of the groups around them and the style of riding, this topic will fade away and/or people will realize the RP might not be the solution to their “problem.”

RP’s are designed for big-pack riding majority of the hours through the day. That’s the only part about them that should be considered constant.
Beyond that… every day, and every hour, should feel different than the day/hour prior; and the data should [and does] only prove that.

There may be a solution however to this.
Something we do during our group rides is tighten the fence, or pull back our front-riders to settle down the pace if they begin pushing.

If the RP’s could potentially have the “brain” to realize that their moving pace is higher than that of which it should be, and slow down / speed up to bias its own power to aim for a specific speed… some of this issue could potentially go away.

That does however greatly increase the CPU demands of whatever is running them; more than “increase watts on inclines”

1 Like

Not to be confused for being a beginner here again, I know how it work, as in the video…


(If you can figure it out)

I’m just sharing the feelings from other riders, especially on the flat courses with larger groups.