We haven’t heard much from Zwift Co-Founder Jon Mayfield lately, as public interactions on game updates are now being handled by other members of the Game Team, including Wes Salmon.
But Jon and his team have been busy, working on various projects deep in the bowels of Zwift’s super-secret R&D lair. (Actually, thanks to Covid, I would guess they’re all working from their homes. But the lair sounds much cooler.)
One project recently pushed out to the live codebase is Pack Dynamics 3.0 (we’ll call it PD3 for short). I’ve had the opportunity to participate in a few test rides using PD3, and wanted to give you all a look at what’s coming. Because it’s exciting!
First: The Current Situation
While most Zwifters have grown used to the game’s pack dynamics, the way our avatars move on the road lacks realism in various ways. Watch a snippet of a recent race and see how the pack looks to you:
See anything odd?
I see a lot of side-to-side movement within a tight pack, which isn’t something that happens IRL. Outdoors, riders hold their lines pretty well in a group, because there’s not much space to move right or left, and it’s a dangerous move if overlapping wheels are involved.
I also see the bodies of riders moving through other riders, which of course knocks the realism down a notch.
Lastly, I’d say the pack looks too tight. Even a pro peloton can’t fit that many riders into a tight space!
PD3 seems to address all of these issues quite nicely.
See Pack Dynamics 3.0 In Action
Zwiftcast creator Simon Schofield participated in the same two test rides as I did, and put together a quick video with his impressions of PD3. I think he sums it up quite nicely:
Less Lateral Movement
You’ll notice the jerky side-to-side movements visible in my race video are nearly non-existent in PD3. It feels and looks much more realistic… like the pack is a school of fishes rather than a scurry of squirrels.
Less Sticky Draft
While Zwifters generally recognize why the sticky draft exists, there’s also nearly universal disdain for it. We like sticky draft when it helps us hold a wheel, but we also hate feeling like it’s harder to get around another rider than it should be.
I found the draft to feel much more natural and not sticky at all in PD3. I could ride up behind a rider and hold my position on their back wheel by modulating my power. And if I wanted to go around them, bumping up my power moved me smoothly around, without feeling like I was wasting watts trying to power out of the sticky draft.
One caveat, though: both of the test rides I’ve done were at a rather low pace (2-2.5 w/kg). It may feel quite different at race pace.
More Realistic
Overall, the look of the pack, and the feel of moving forward and backward in it, is much more realistic with PD3. Riders hold their lines well, and, at least in the two test rides I’ve done, there is more space between riders.
Will the increased space making drafting more of a challenge, especially when it matters at high speeds in races? That remains to be seen because, again, we haven’t tested PD3 in a race situation.
Better Predictive Positioning
PD3 also includes an upgrade that isn’t apparent to the individual rider, but matters a lot when it comes to racing and broadcasting. That is much improved predictive positioning of other riders.
The positions of other riders on your screen are calculated using a combination of client-side predictive positioning (your Zwift software is predicting where each rider is going, and rendering their positions accordingly on your screen) and server-side adjustments, where rider positions are broadcast to a central server then your game client sees them and adjusts riders positions on your screen.
This means that one Zwifter’s view of the pack could be quite different from another Zwifter’s, especially if server connectivity was slow or poor. Most of the time this doesn’t really matter – but when it comes to racing, and especially pack sprints, it matters a lot!
In the past few years, Zwift has been tightening up how well this works, so rider positions have become more and more uniform between Zwifters. But PD3 tightens this up much more.
When Will It Be Released?
Zwift is being cautious in rolling out PD3, and will continue running test events before a global PD3 rollout. The current PD3 has a few bugs which will be fixed before any larger public release, and a lot more private testing will take place before public test events are available.
The broad release plan is:
- Holding more private test events
- Hold public test events
- Enable PD3 on an entire map
- Enable PD3 across the entire game
Zwift isn’t giving a PD3 release timeline, but it’s certainly not “just around the corner.” My best guess: sometime near the middle of Q4 2021.
Watch More
Want to see more PD3 in action? Here are the full recordings from my two test rides:
Test Ride 1, Makuri Islands
Test Ride 2, London
Your Thoughts
How does PD3 look to you? Share your thoughts below!
I don’t know what flavor of PD they were using last week during the WTRL TTT’s, but it was by far the worst iteration. Riders would shoot through at 4.0W/Kg, but struggle to get to the front intentionally at 6.5, then slide off the back at 4.5 while others in the group could cruise at 3.8. It was an absolute mess, so I’m really hopeful it was not what we can look forward to over the colder months.
How will this work for “The Blob”?
I would like to see more realism added by allowing use of the whole width of the road. At least in races and events where you’re the only group on screen. There’s no yellow line at the highest level events, it would be nice to see that added into Zwift, especially for the events that get massive participation.
Keen to understand how it will work with TTT. PD2 definitely changed team dynamics (anecdotally not positively) so this will be an interesting conundrum for zwift to solve – what might be better for slow pack rides might not work for fast races or TTT for instance. Ultimately there’s been a lack of investment in TTT (eg start process is a joke given TT taxi rank already exists) which given TTTers are some of the more hardcore dedicated evangelical zwifters is “disappointing”. If PD3 doesn’t work for team time trials could they run PD1 for those races only?
One thing I’ve always wondered about sticky draft is why any “stickiness” is applied when you’re you’re approaching a rider from behind. Why doesn’t it only apply like a “bungee cord attached to the rider in front’s saddle” when you’re in danger of falling back; i.e. when you’re starting to go slower than a rider in front of you?
The way RGT handles it is when you’re riding, there’s two numbers shown, one for the power at which you’ll be dropped, which is the solo power minus the draft benefit, and one for the power required to pass. In between, you stick to the wheel, and any power above the minimum us wasted Imagine I’m tuning Zwift…. if stick draft is infinite, you cannot pass riders ahead. If it’s zero, then I can move from the back to the front of the pack in a frictionless slingshot. Obviously reality is in between. You want it tuned to adequately reward… Read more »
It really looks smoother and not so exhausting for the eyes. It would be interesting to take part in these public test events (step 2) to feel this new experience by myself. And additionally perhaps in a Pace Partner Pack. Thanks Eric for this first appetizer.
Jon and the lair should team up with WTRL and test PD3 with the TTT enthusiasts on Thursdays. That would give great feedback from riders already very detailed in pack dynamics instead of mostly random beta-events.
Can’t come soon enough. After the two Acadamy Road workouts, I learned to stare at the floor instead of the screen, for the constant and drastic side-to-side swerving motion of several cyclists in front of me was just too tiring to watch.
This is why I hate group workouts, do them on your own and you will have a better experience.
The overhead helicopter view works best here. Bizarrely it is not available when I cycle through on Companion, but it’s the eighth view when I do it on Apple TV.
Does it make riding feel anymore realistic or are the races still threshold workouts with Supramax intervals?
That remains to be seen. I’m really curious how PD3 will feel at race pace!
The drafting doesn’t seem realistic to me. Why do they model the draft to be so much less than real life? Just to give people a better workout?
The descending algorithm is screwed up too. Pack Dynamics needs to drastically increase the draft effect to approach real life. Racing on Zwift is just an FTP willy-waving contest at the moment.
the draft in Zwift is pretty significant, for example, its pretty easy to average 40kph for a lap of Tempus with Coco and 30-40 riders, and the same lap solo at the same effort, I would be lucky to average 30-32kph… Admittedly you have to learn how do draft effectively in Zwift and it can be trickier than IRL (though there is no shortage of those that are terrible at it IRL as well)
Absolutely the right question. I don’t care what the screen looks like, I just want the drafting to be realistic so the racing can be actually interesting instead of a W/kg dragrace every dang time.
will be interested to see if this changes things at all. currently when riding with pace partners, i notice 2 issues: 1) if you drift to the front for even 1 second, you instantly get catapulted to the back of the group, this is annoying because sometimes you have to get out of the saddle just to not get dropped and 2) if you fall off the back for any reason, you gotta push literally 5w/kg to get back on (to a 2.5w/kg pace partner). this seems a bit extreme to me!
I warmed up for ZA with a pace partner yesterday, and felt that the pack dynamics had changed. Definitely more fluid movement and no jerky steps to the side.
Far as I know, Pack Dynamics haven’t been changed across the game for a few months now.
I often ride with my best buddy Diesel and felt no changes in the PD until today. I agree with ShakenBake that it would be perfect in PD3 that the positive- or negative-momentum-effect for winning or losing positions in a bigger pack would be decreased. I think the group in Eric’s test ride wasn’t big enough to feel this possible momentum change … or ?
I’ve always wondered if it is known how pack position is determined, front to back of a pack? Is raw watts, w/kg, connection lag or something else the main determinant?
sticky draft is tricky…but where is the car with my sticky bottle? now there is a feature I need — haha!
thanks, Eric and Zwift appreciate all your work!
Less sticky draft is a big bonus! I can’t count the number of times it has cost me precious watts in a race by having to burn extra matches to rip the velcro from the other rider that’s shoving me to the side of the road and get back in the pack draft. As for pack dynamics overall, I’m amazed it’s taken Zwift so long to get it figured out. Must be nearly as difficult as getting a Tron with a red glow I guess.
Yeah either that or the one where a rider in front of you lets a gap appear between you and the pack and you have to go to some insane W/kg to get past them and by then the pack gap is out to 10+ metres and you either burn up to catch or you’re relegated to dropping behind them and eventually getting swallowed into the next pack…and then another rider blocks you and leaves a gap…and repeat!
What I hate the most is when you’re in a pack and suddently your avatar decides to go left or right and you get dropped because there is no draft there ! I hope they change that !
Or sometimes you’re near the front of the pack, doing more than 4W/kg on the flat, and suddently half of the pack come by and they’re doing 3.5 or less, how is that possible ?
Like it. But pack speed is still 2-3km/h to fast. Speed in TTT and soloride is good. A fix would make racing alot more fun for non-sprinters.