Just checking: is the 80–90 W figure low enough that it does not lead to riders at the back of a stretched pack getting interpreted as being at the front of the pack?
Power drops below 5% of your last 3 second power (this would be 15w if your last 3s were at 300w), or power drops 5% below your last 3 second power (this would be 285w if your last 3s were at 300w)? Both options seem unreasonable.
it’s the second option. I didn’t typed it correctly.
why both seem unreasonable? what would be the best option to signal you are slowing/keeping your pace?
That shouldn’t happen because of the distance rule. Only if you are very close to the rider in front you and start to overtake you can get brakes.
At the back you are either being distanced or getting closer to the pack in front of you and in that case your draft savings are increasing and not decreasing.
Sounds like a great set of criteria to test. How come you didn’t choose 0 as the second value? Is to catch it that slight bit faster? The reason I ask is that potentially you could be autobraked when there is still a draftable rider in front (albeit lower draft) for example at the transition to a climb. Checking if it goes to 0 would ensure the rider is in the wind. Looking forward to testing this new approach!
It’s not random. If you accelerate in to a pack from the back in clear space you go through the middle. Every time. What could be random is to which side another rider gets pushed.
1 Like
Has any discussion been had about making the descending physics model more accurate? This seems germane to trying to improve races. Descents in Zwift are not only a neutralizer, they are a nullifier for anyone Under 65 kg. A rider that could put over a minute into a chase group on a climb gets rewarded by being swarmed by that group within less than a minute after that group crests the climb. A group, mind you, that can coast on past a rider throwing down 5 W/kg on a descent to try to stay away.
This certainly isn’t an issue faced by the meaty part of the bell curve on Zwift, but it is problematic, especially at the more elite categories. Unless finish lines are at mountain tops, lighter riders are required to put out unrealistic efforts compared to others in the field. While apples to apples, a heavier rider will go faster on a descent than a lighter one, the model on Zwift disproportionately favors the heavier rider. Moreover, smaller riders often have a better CdA that neutralizes some of the weight advantages.
The proof is seen out in the real world all the time. I know that I can hold my own, if not pass, men that have no less than 15 kgs on me on substantial descents. But my anecdotal account is an n of one; a solid look at the pro peloton, where sprinters and time trialists are being left behind by climbers and domestiques on descents, routinely demonstrates weight is not the determining factor.
3 Likes
I agree downhills are total nightmare in zwift. Riders doing unrealistic speeds…but that is the part of the pack dynamics and supertucks speeds in races. Hopefully breaking in corners will also help but not sure what’s the progress on that.
4 Likes
Weight isn’t a determining factor for descents in the pro peloton, you’re right. The main factor is descending skill.
However, there’s currently no such thing as descending skill in Zwift.
I agree that perhaps things are skewed a bit too far in terms of weight for Zwift descending, but I don’t think there’s an easy answer to rebalancing this.
1 Like
That’s really my point. The only determining factor on Zwift is weight. And yes, descending skill is paramount, in Grand Tour races especially, but there are also many examples of descending where it’s not about skill and more about CdA of the rider - how small they can get. The heavier riders with a great tuck aren’t flying away from the lighter riders, even when descents are ultimately a straight runout. I call out the pro level specifically because it eliminates the confounding variable of a generic fear of speed many amateurs have (even when no real skill is required).
2 Likes
A bigger reduction (50%?) or reduce to a low absolute value, e.g. 50% of your zFTP. 5% is noise.
This won’t work on fast descents as everyone will supertuck (last 3 sec. power = 0W)? Could it be changed to «or your power is 0W» to allow for slower descending in large groups?
There’s no reason why this shouldn’t be improved with PD4 too, if the churn is reduced. There’s already corner speed braking in the game that hasn’t been enabled yet.
I completely agree with you.
I’ll be happy to see an appreciable improvement. What is not clear is whether this would impact the dynamics outside of the bunch. The solo rider should not be caught as quickly by a heavier rider, mainly when additional power is being applied (if the lead rider is coasting, that is a different animal altogether). Braking in the corners will attenuate that in theory, but the impact, would that be proportional to rider size (reverse linear relationship) or fixed? The benefit might be minimal.
Honestly, I would be happy just to see terminal speeds for all riders ratcheted back on the descents. No one on Zwift competently descends at 80+ KPH - ON SWITCHBACKS! 
Yes, it’s to stop any braking a bit faster. The intervals will be configurable so it’s not a big deal for now.
4 Likes
Before we look at descending issues, lets test it with this new formula for drafting. I have a feeling based on the above breakdown that we may see packs descending slower.
We need to test one iteration at a time, testing multiple new things at the same time makes it too hard to quantify accurately…
2 Likes
I popped another feature request up the other day that can also with help this…
2 Likes
A bigger reduction would not be enough according to initial tests, the amount of braking would be very mild in most situations and would barely make a difference. Anyway this value will be configurable so it’s not a big deal if initially is not perfect.
FTP or zFTP are not usable. The first can be manually set and the second can even not be available (e.g: someone that hasn’t ridden recently). So once and for all FTP and zFTP are not part of the equation.
2 Likes
I think it’s a fair statement to stagger out changes. I wasn’t suggesting it be addressed with PD4; however, I think addressing the physics model around group descending, unto itself is problematic. At a -8% grade, for example, the riders should be considered individually, not as a drafting pack. With the braking, I think Zwift might be backing into a stopgap, but it doesn’t address the root cause of the problem.
It’s been a very long time that this has plagued the platform. It seems like racing is the only mechanism to get positive change made, but I really believe this needs to be addressed more ground-up.