PD5: Steering - Braking and Zwift serversf

Thank you for the careful testing of PD4.1. In a test race with only four Ds nothing weird happened, so you have definitely fixed some bugs since the first test.

But here’s a plug for PD 5.0:

I’m sure Zwift is checking out indieVelo’s dynamics. They are extremely nice. To be fair, I’ve only experienced it with races with bots, and while the bots are dynamic, their power may not oscillate as much as it does with people. Also, I don’t know if their system would scale to group rides with hundreds of people.

That said, especially observing how steering and braking is now impacting everything with drafting in Zwift, I think Zwift needs to seriously consider a major revamp to do whatever indieVelo does that makes their pack dynamics so smooth and prevents riders from going through each other.

Zwift needs:

  • Slowing at corners, based on radius, speed, and pack size
  • Blocking of bodies so avatars can’t go through each other
  • Smoother pack dynamics so avatars don’t move around so much
  • Steering that is somewhat sensitive to speed of riding, so it behaves more normally
  • A visual display of how much draft a rider is experiencing
  • When someone doesn’t have steering, their avatar should be willing to, and better about, getting in the draft
  • Possibly: more trainer responsiveness to being in/out of draft to give better “feel” of the draft
  • Possibly: once all the above works, wind.

My 2c!


That wouldn’t be pack dynamics 5, it would be Zwift 2.0. Fundamentally architected in different ways.

1 Like

Do most people actually want wind? When I saw the pitch for indieVelo and wind was one of the big selling points it seemed odd to me. I get why it can make things more realistic and add another variable to strategy, but there are many other things I’d want ahead of simulating the frustration of riding into the wind :slight_smile:

Edit: I haven’t had the chance to try out PD4.1, but it sounds pretty promising.

O yes. Wind will make the tactics on the same course different every time the course is used and the wind come from a different direction.


The value of wind is for racing. The same course with different wind amount and direction becomes a different challenge. It adds more strategy to the game, which can be fun for racing.


The cornering and steering things that you have said would be so cool but at the moment there isnt any platform that has made cornering like irl and steering doesnt correlate to the speed which would prevent riders from going from side to side like crazy.

1 Like

Maybe that will change soon :wink:


Yes pls to wind. Attacking when the wind flips from head to tail/cross would make flat courses very, very interesting. Especially if it was from variable directions and random.


This is probably the #1 thing to solve.

Zwift racing can never be taken seriously until it is.

1 Like

Really disappointing on a major showcase event.

1 Like

Once again, this is derailing a bit from the Pack Dynamics debate, but I think it’s worth addressing a few topics that have been raised here.

Wind and auto-braking in corners:
Sure those would be nice additions (in theory), specially when thinking about the dynamics they cause in real life racing. But keep in mind that those have a much higher impact in real life because of rider position in a group, where riders can’t easily move up in a group if they are blocked. In Zwift, moving up can be solved with a 5 or 10 second sprint. I know, there are situations where riders are on the limit and that is not possible, but my point is that perhaps things like wind and auto-braking in corners is being a bit overrated at the moment in the context of virtual racing.

Server authoritative model:
Also another one that is being a bit overrated. For sure, it’s a good thing that everyone sees the same thing. Would it definitely solve problems like in the video above or other past examples, not necessarily. Because of network latency, even if all are seeing the same thing, they can all be seeing the same “wrong” information.
Let me give you an example. One of our competitors that has a server authoritative model, that recently held the 1 million dollar championship, I know of at least one case in one of the races where both the broadcast and the racers all saw the same image of a rider crossing the finish line first but was not the actual winner. When the differences are short, because of network latency what you will see graphically might not be exactly what is the true result.
With this I’m not saying that server authoritative is bad, not at all, it’s probably better, but I see it as a “marginal gain” in the context of Zwift racing.

With that, what I really wanted you to be aware of, is that currently there are more important things for the Zwift racing ecosystem than those above that have been hyped lately.
For example: finishing the work on the ranking based categories (scoring), performance verification for the masses or better tools for event organizers, etc.

(Also would like to make clear that this is my personal view. I’m not saying that Zwift won’t implement things like wind, etc. It’s all a matter of allocating limited resources)


All seems perfectly reasonable… Wind is currently in favour as one alternative has introduced it.

I do question if braking isn’t coming, why enable it by default on the zwift play and not be able to disable it? It’s kind of annoying/frustrating - it’s literally only a hindrance at present.

1 Like

I agree with you on the server authoritative model while it is probably better to watch all seeing the same thing it does not mean it’s accurate so hard to compare. Would people rather all see the same result even if it’s wrong or the wrong result on screen. You could argue for both but maybe server authoritative is better to spectate assuming you don’t get much rubberbanding.

latency happens all the time the big FPS games have not solved it whether you do client or server authoritative it’s never going to be perfect.

Although for these big events that are running on the same LAN the occurrences of latency should be almost zero. I assume despite being located in the same place they are still relying on Zwifts Datacentre…maybe worth an offline model for these headline events to minimise risk of latency issues.

anyway that’s for discussion elsewhere.

1 Like

Be careful not to assume that because something was typed by an authoritative figure, it is correct.

Most likely an option to configure it/disable it will come out very soon.

From my experience you only accidentally trigger the brake in races once. After that you are much more careful :sweat_smile:

Still about previous concerns of steering in racing events.

A set of fixes and improvements will come out (probably in 1.44) that will make the steering experience much better inside groups.

Also for the concerns about riders that “abuse” steering to easily breakaway, I’m working to get some changes specific to PD 4.1, that penalize riders that do so. We want to make sure things are balanced.


Wouldn’t it be better if my avatar would be looking for better draft if someone is steering in front…than to penalize someone that does it.

Example i’m on the side of the Group and i want to break away…if i have steering ON my rider will stay in that line. But with steering OFF my rider will move to the middle of a road by it self. So why would you penalize someone?

1 Like

Wind: When ever we start talking about wind there is always the notion that you will need steering to make it like IRL with ECHELONS all over the road.

What I am referring to when mentioning wind is riding into a head wind on a flat road will make that road harder almost like a climb that is not always there. Planning attacks so that you can take full advantage of the wind situation.

1 Like

Yes, that is the ideal solution, however much more complex to implement for an optimal experience.
Imagine several riders with steering going from side to side at the front of a group, it would turn the pack into a weird “conga line” :joy:

I prefer to get something out first that is not perfect, but it’s configurable, and can prevent the spread of a potential problem, until we work towards the ideal solution.

Edit: Why move the rider following to the middle of the road instead of following the steering rider in front? :thinking:

1 Like

How do you balance the want for steering and the need for it, against a balanced system?

Otherwise, if all things are equal, what’s the point of manual steering over auto steering?