Pack Dynamics 4 Release [April 2023]

Collision detection will still stop people from riding through each other. Whether they can steer or not. If you have steering and you steer into the wind you need a certain output to move up. How is that different from a model that knows all positions and your power/speed that will know whether its enough to move up given the constrained space or not? The latter will be more optimal like our position on the bike is and arguably more realistic as with all different setups and lags pack shaped would likely be a mess. Or due to inexperience - and randomness people tend to create consciously in games - very inefficient. When people know their watts/speeds and know their position in the pack (as well as size) with a little experience in this game it won’t be rocketscience to know required output to move up (or prevent from being pushed back).

There is a fight to be first at the decent to be able to get some rest with super tuck. The rest of the decent is a fight to be at the front. There is no rest at the back anymore because of this behavior.

1 Like

Yes, but IRL factors in spatial presence i.e. potential collision hazard(s) and position. Two key elements not replicated properly.

As a hardcore racer with strengths and weaknesses — I am unfortunately not an all-rounder cyclist — I am more disadvantaged from the Zwift physics engine than compared to other virtual platforms + IRL physics. I’d love to hear of your competitive experiences on Wahoo x RGT and whether they were “fun.” PD 4.0 outlined a goal to make the Zwift physics more “realistic,” so what are you misunderstanding that “Zwift is not simulating IRL?” I do not follow.

I train to my max on Zwift, but I strongly feel “cheated” by the mechanics because of doing incredibly high power output via a race-winning move and get bridged by a chasing peloton not going full gas. No; I don’t send it for a 5++ minute effort often. Elite racers agree with my stance. Without being a sprinter, it feels discouraging I cannot hold off the peloton for just enough time, unless it’s a points race.

For reference, I’m talking around 8WPKG per 2 minutes to around 7WPKG per 4 minutes at 70+ KG.

Personally, I don’t believe we should accept sub-standard things in life if we pay money for it, especially as this is a recurring monthly subscription. If another virtual racing platform attracted enough subscribers with perfect physics emulation, I will abandon Zwift.

1 Like

Sorry, If you read that somewhere it was either a miscommunication from myself or someone else.
We outlined the goal of PD4 making race dynamics more realistic, not necessarily the physics. The goal was to give a bit more chances of success for breakaways to avoid the very defensive style of racing of waiting for the final sprint in a majority of races.


I think had that been part of the comms from the outset, it would have changed the tone of feedback.

That perfectly explains why the back of the pack feels a more difficult place for some, from the statement above it feels like that is the intention of the change to improve race dynamics. I’m perfectly fine with that, I just think a lot of the conversation has centred around how it feels, and not its intent.

If the intention is to animate racing and change the style of races that’s a good thing…

Came second here: Cycling news - Peter Sagan takes sprint victory over Geraint Thomas in Shanghai Critérium - Eurosport

A really simple fix for this would be to increase the default CdA for all riders by, for example, 5%. Everything else works exactly the same as before, there’s just a slight slowing.

There is no correct answer to the CdA a virtual avatar should assume, so there is also no issue increasing it across the park slightly.

Note that comments like it’s too fast/slow versus real life are really meaningless, as anyone that has ridden across a variety of destinations, road surfaces, air densities etc will attest too. I can point to irl rides on the flat that demonstrate that zwift is both too fast and too slow versus real life.


It looks like it can be hard to get approval for CdA changes:

Continuing the discussion from Pack Dynamics Test Events (December 2022):

This would also affect the speed of solo riders, which I think is very close to IRL speeds. My speed at a given power on flat terrain on Zwift is very close to my speed at that same power on flat IRL. It is the pack speed that deviates from IRL pack speeds due to Zwift not being able to simulate real world drafting in a pack effectively. What’s the point of making everyone slower, when it’s the group speed that’s the problem?


Guys… let’s change our way of thinking : in fact for some people like me solo ride speeds et group ride speeds are identical irl and in Zwift.

The problem in Zwift is that most races are way too short.
A 20km race is just a bing sprint… if you want breakaways to succeed you have to stop trying to simulate the reality.
The only real solution would be something that has shortly been tried (we had a few races like that two years ago I think) : bonuses you accumulate or charge, and you could then use to have some extra power or extra speed when trying to breakaway.

Zwift cannot simulate real life and offer short races and be fun at the same time without being a game, with bonuses and stuff like that I think.


In all honesty, not a lot of people liked that concept, which is probably why it never made out of futureworks.

This topic is going way off track.

Move the Off topic chat to a new thread. (Crossing finger I did not break something… :woozy_face: )

Please keep this to PD4.


Did the DBR race on Sand and Sequoias just now. Lots of complaints during the ride about the pack going too hard. Lots of people blaming PD4. I’m not so sure though, I think it was a pretty stacked set of folks in the ride (at least the B pen seemed pretty strong, and the race had a very low race ranking indicating a pretty solid field). I was dropped at about the 15 min mark which is before I was expecting to get dropped (actually before the Titans KOM), the first 15 mins were mostly flat, my average power was 270W (3.5w/kg) for this time, and I felt if I kept that up I wouldn’t have enough power to keep with the front pack on the inevitable KOM surge anyhow. So I let them go, of course in retrospect I feel I wimped out and should have pushed harder.

That said, a lot of complaints were “that’s it, I’m tapping out, this is too fast” from a few people sitting in the front group. Meanwhile I’d been dropped a long while back from that group, and the second pack was way less of a sufferfest.

I think a lot of people just have an expectation that they need to stick with the front group, and if that doesn’t happen they’re going to be upset, call it a pack dynamics issue, and possibly rage quit, or just continue complaining.

How someone who is at the end of their matchbox can continue to type while racing is beyond me. I can’t even type when I’m doing ok in a race :slight_smile:


I think people are reacting to the change in behavior and not that it is hard. This is the same complaints we see in the RoboPacer feedback threads.

1 Like

Not in this case, the feedback was specifically “this is too hard”.

…and many don’t even recognize that there was a change in PD.

And also many don’t really care, even if they knew :wink: :rofl:

1 Like

[quote=“Johannes [FUSION COLOQUICK], post:179, topic:605018, username:Johannes_K”]
Well, the primary goal of PD4, as I understand it, was to "Make breakaways have a slightly better chance of sticking if the attackers work well together.”,

Only if behavior doesn’t change. If the pack behavior is to go much harder than before, then it is fine that breakaways don’t get away. The problem with PD3 was not that breakaways got chased down, but that a break doing 5wkg got chased down by a group doing 3wkg with nobody having to pull the group. If PD4 requires that the blob actually has someone working to chase down the break and this results in more people working in the blob, then PD4 has done its job.


But it doesn’t, as long as a sufficient number of people are consistently pushing 3wkg and keep the churn going.


It would be nice if someone can actually get a good video of this.

Wow this stream is very helpful today.
Look at Bjerkseth he get a 2sec lead then hold +5w/kg and the pack is +3w/kg and not catching and he opens the gap to 3sec then the pack start to work.