In the January update thread, that is now closed, he said they moved to more focus going from Medium to Low draft, which seems like could be pretty common on a hill. I’ve seen a report (on Discord) of this happening while going up the titans grove KOM.
Yes there is. Those complaining they get dropped uphill because of new pack dynamics are lying 
3 Likes
Sorry for spamming this thread, but I actually have something positive to report 
I did the STAGE 1: COBBLE CRUSHER - R.G.V 17:10 CET again today. I lowered the trainer difficulty even more before the race, and that seemed to do the trick for me. We used ca. 30 sec more than the last time I did the race, so it is not an equal comparison, but it felt easier to move up in the field and also not as scary at the back. Either there have been some tweaks backend, or the trainer difficulty change reduced the auto-braking (more stable power). I also felt that I could get some micro rest at times. The riders didn’t seem to be as stable as I remember them behaving, but I’m not sure if that is just imagination.
1 Like
My Setup is Neo2T with BT to PC at 30% Trainer Difficulty using (Zwifthacks Preferences for this).
I will test TD but this would suggest because of more gear changes your power will be more spiky and the end result could be more Red Numbers 
I also use 40% difficulty, and on a Neo bike… the Neo bike is very different from my previous kickr core, and leads to a different way of pedaling with less acute power spikes (because of the virtual flywheel corresponding to your weight, you have to accelerate on a few pedal revolutions as irl). So maybe you are right and this changes something.
Just raced the cobbled classic race.
Tried to race in the front third for as long as possible to see how it went (this worked till the aqueduct)…
Fairly smooth, managed to stay in the draft, managed to move forwards when I needed to, don’t think I saw a red number all ride. Very little in the way breakaway attempts apart from entering the twisty bit towards the end.
I have the race video if people want it… you can ignore me getting dropped in the last km as I run beforehand 
Coco group was quite easy indeed, I was there yesterday with about 170w average for 45km/h and much less surging than previously.
Constance group only 1w more power needed for PD4. 209 v 210w.
I use 100% trainer difficulty with kickr bike.
Constance group is the best one for testing PD4 because nobody is doing big sprints away, the group speed is very high everywhere and it is on tick tock.
I didn’t notice this effect at all. But uphill (from LAX) and back to the desert on Tick Tock I was usually doing a lot of power (4.6w/kg minimum)
1 Like
Is there still a difference between race dynamics and group ride dynamics?
Wow, the feedback is all over the place!
Is it even useful if it is clearly so wildly subjective?
No idea if it will help, but here goes.
PD4 felt a little different. Hard to fully articulate the differences. They are very subtle differences.
Pack speed is still wildly too fast. Group doing 43kph at 3w/kg. With nobody doing the actual power on the front required to maintain this speed. Churn seemed a little reduced, but still light years from solved. Drafting really didn’t feel increased by any meaningful amount.
Double draft, really? Hell, try quadruple draft, it’s still not aggressive enough.
Movement in the blob seemed slightly better.
Summary, after all that, a very minor change.
A dramatic change is needed to further promote more dynamic racing.
Kudos for all the work. Please keep at it 
Most of all…
Make more aggressive changes, take actual risks. Please don’t listen to the tiny vocal minority here that want things to stay the same. The current Zwift physics still need a fundamental overhaul.
2 Likes
How many raw Watts are your 3 W/kg?
Flat route?
I ride alone most time IRL, but my best Strava segments (drafting in a “group” of 2, flat, no wind) are 35,7 km/h with 2,2 W/kg = 202 W.
Riding alone 36,0 km/h with 252 W = 2,74 W/kg.
Tempus Fugit with Coco in a big group 39,2 km/h with 187 W = 2,04 W/kg - for me it seems to be comparable with my IRL riding.
There’s a billion variables.
Data from a single person is next to useless.
Due to allometric scaling, larger riders generally travel faster for the same w/kg than smaller riders. So, n=1 comparisons are difficult. There are many cycling calculators that estimate speed vs power etc. Many are very accurate. Try entering 43kph into any of them…
43kph average speed on a road bike without wind assistance requires very high power.
In my case about 350w or 5.5w/kg.
Zwift is obviously not real life, and doesn’t need to be. However, one of the goals of this update was to slow down the ridiculous pack speeds. It has had an effect, however it needs to be far greater.
Did my first race using pd4 yesterday. Felt absolutely fine. Red numbers regularly flashed as I approached the front of the group and checked my speed so I didn’t roll through. One instance where it felt like I got braked out the back and had to push back in but overall fine.
Maybe the meta will change with new pd. Favourite pacer now too hard- drop down to the next one down (use gravel bike if too easy). Racing favouring a particular rider weight or power phenotype- implement results based categories and everyone will find the appropriate level for them
Zwift is super flat so it’s hard for me to compare most of my group rides with it, but occasionally I go home to North Florida, which is almost as flat as Zwift and jump in the Wednesday Night Worlds style fast group ride there which is basically a mix of racing each other and pacelining—very fast, large group (20 or so), drafting is important.
We did the main loop at 25.7 mph (41.36 kph). I didn’t have a power meter at the time, so I’m treated to Strava’s estimate, which is really an estimate of what I would have to do solo — 347 watts. But I can’t do close to that many watts. The actual figure was probably no more than 250 (3.2 wkg) since now that I’ve got a power meter I know that tends to be standard for me for a hard feeling groip ride. Draft does make a huge deal to how fast a group can go. That route is 42 ft climbed per mile (its Florida) RGV in Zwift is even flatter, for comparison, 27 ft climbed per mile.
I think if you factor in the fact that Zwift roads are perfect newly poured tarmac and you’re riding the equivalent of a very expensive aero bike and always riding in the drops, 43 kph isn’t that ridiculous.
2 Likes
Now that the churn at the front is reduced and the speed is still about the same, it seems like this was not the only factor causing the high pack speeds. There seems to be a more fundamental problem with physics in Zwift. When we also know that the draft is increased by 50% and should be much stronger, something is not right.
No, big pack speeds irl are really fast… my speed irl for 4 wkg with a big fastpack is the same as in Zwift…
2 Likes
It should actually be even faster in Zwift then, since there is no wind in Zwift?
I am speaking of a loop, so the wind really doesn’t change things much : if you have in coming from front in a direction you have a tailwind on the other side…
An example, with 30 km at a good speed, with low watts (big and fast pack)
1 Like
Wow. I guess you spend all your time on Tempus Fugit?
No, nearly all the race courses except for actual climbs are flat. For instance, the Watopia “Hilly Route”, last time I rode it, was 4 laps, 23.10 miles, 1430 ft of climbing. That’s 61 ft per mile, on something called a hilly route. Where I live our flattest routes are 65 ft per mile, but the average is more like 75ft, and it doesn’t feel hilly until its over 85ft, and if you go ride in a place with rolling hills like Atlanta, GA you’ll do 100 ft per mile of climbing, or an actual really hilly area like Knoxville, you’ll do 125ft / mile. Unless you are doing the big climbs in Zwift, most Zwift races are ________/\____/\____ i.e. basically pancake flat with little short climbs thrown in.
The Glasgow routes are the first race courses in Zwift (apart from the actual climbs like the Alpe) that are really not flat.
Downtown dolphin: 16 ft / mile
Volcano Circuit: 25 ft / mile
Whatever Makuri route 3R raced on Apr 1: 39 ft / mile
Innsbruckring: 42 ft / mile
Road to Ruins: 48 ft / mile (You can get more than this in Florida, and it has the “Jungle Climb”)
Three Little Sisters: 55 ft / mile
(EDIT:
I looked up the Glasgow routes, and I was incorrect – they are flat too, just not as flat:
Rolling Highlands: 37 ft / mile
Glasgow Crit Course: 55 ft / mile
The Muckle Yin: 61 ft / mile
And finally a hilly one:
The City and the Sgurr (I probably ended after the descent for a cool-down, so it is probably more climbing than this if you stop at the top): 85 ft / mile
)
The last ride I did in Florida, where the highest point in the state is 345 feet above sea level, and the max gradient of any hill was literally 100 ft long at 7% (the climb it appears on is the only strava climb segment on the group ride, which is 0.33 miles long at 4%) and all other “hills” were below 5% maximum gradient and average gradients below 2.5%. That ride, was 42 ft / mile of climbing, which is about mid-range for Zwift. Zwift is just not hilly. (Edit: except for Yorkshire and NY as pointed out below).
My outdoor evening rides last week that was just around my town: 101 ft / mile for one, 84 ft / mile for the other, neither hit any actual categorized climbs. (Edit: Last time I was in Atlanta I did a downtown ride (i.e. not going out to the mountains, just riding in the city itself) that did 113 ft / mile. I can’t find my last Knoxville road ride, but I remember it being 126 ft / mile.)
3 Likes