unterhausen

I often get dropped at the top of the hill. Has anyone noticed if they are allowed to go over their nominal w/kg, or do I just suck at riding a bike?

Also, I thought the xp multipliers showed up while they are active, but I still only get 32xp per mile. Is there any way to tell if zwift is being honest about that? I am far behind my peers on zwift level, need all the help catching up I can get.
I first noticed this the other day, every pace partner was named N. Valid. But then I rode with another pace partner more recently, and it didn't use the multiplier when I rode with them either.

zacster

The robopacers go .1 or .2 over their nominal rating on hills, both up and down. The multipliers apply to drops, not XP. I was doing a long easy ride and really racked up the drops when I stayed with one pacer the entire time. Every 5 minutes the mulitplier increases, 1, 1.2, 1.4, 1.6, 1.8, 2, 2.5 and then stays at 2.5 as long as you maintain the pace. XP will get you gear, but drops will get you bikes and wheels. I have enough drops now to purchase 3 or 4 top frame/wheel combos. But unless you race in Zwift, does it really matter? I "bought" a gravel bike the other day because I was going to be on a gravel road but it didn't matter to my ride. I was still riding at the same watts and burning the same calories, and it didn't matter if the generated scenery went faster or slower. What did matter to me though was that I was fast on the sprints, coming in in the top 5 on two of them, and a third I was eighth, all times out of around 250 riders. Not bad for a 69 year old. There's no way though that I could beat the top scores as they are always way faster, so much so that I think some cheating occurs on these.

PeteHski

I did a hard group ride yesterday and eventually got dropped halfway up the Volcano climb. I only lost about 10 seconds, but it was enough to lose contact with the group over the top. I was using my usual Canyon Aeroad, but this reminds me that I might have just about hung on using my lightest Zwift climbing bike - Canyon Ultimate CFR with Meilenstein wheels. 😂

unterhausen

I definitely am a neophyte on zwift not thinking about drops. I guess I just have false memories of seeing it affecting xp.

The problem I have is going too hard up the hills and getting too far in front and waiting, only to be dropped over the top of the hill. It makes sense they increase the power up hills and down. I rode the other day even though I was sick, so I went with the lowest power robopacer. It was really difficult to not get too far ahead. I recently upgraded to the fastest bike and wheels I could buy, and I like riding better. Just feels a little less like riding through molasses

spelger

i don't like what the robo pacers look like now. i much prefer the old style much better.

PeteHski

Obviously your avatar moves a fraction faster along the virtual road, but I don’t see how it would affect your trainer feel.

Redbullet

No, they are advertised to ride at "increasing power by up to 10% uphill and decreasing up to 20% when descending". It means that, overall, for a ride with climbs and descents, that starts and finishes at the same altitude, you should input slightly smaller W/Kg than advertised robopacer's power (if your weight is 75 kg, like the robopacer).

However, the process is far to be accurate: I have similar weight with the robopacers (only 2kg heavier) and, even if I draft 100% of the ride, I still have to input 5-20% more W/Kg than the robopacer, otherwise I'm dropped. Over a 1 hour ride I input in average 10-13% more W/Kg only to stay in the pack.

rm -rf

Pace Partner strategy
Zwift Insider says the bot adds 10% W/kg on climbs, up to -20% on downhills, trying to mimic real life group rides.
I wonder if the bots are drafting faster on climbs if the group size is reasonably large? Then my own drafting could be critical to stay with a fast group.

Lately, it seems that the bot has some kind of increased drafting -- on flats, it sometimes goes flying through the steady paced bunched up real riders. Weird, but not normally a big problem.
I try not to ride in the last few riders. It's too easy to get gapped off just slightly, and then it's extra watts to get back in the full draft.

Drafting is so critical. It takes way more watts to be solo riding near the group, either ahead or off the back.

~~~

The pop up warning, "Let {bot's name} catch up" sometimes displays when the bot and pack are only a few bike lengths back, and closing. If I coast here, I can be spit off the back too easily. There's a few seconds lag before my increased or decreased watts show up on screen, too. Very annoying. A side, rear, or overhead view lets me see the situation much better than the usual front view, where all of a sudden the group goes flying by and I have to chase.

For "a little too fast for me" bot groups, I like to use the overhead view, the "9" on the keyboard (which is in reach while I'm riding.) I can see where riders are, both ahead and behind, and if they are closing or slowing. That overhead view is not available in the companion app on Android, so there I'd likely use the Rear view or Angled rear view display. I want to react before the trailing riders pass by me!

For easy paced rides, the "0" (zero) controllable viewpoint is interesting. Use the arrow keys to pan the viewpoint 360 degrees or move it upward. use the "+" and "-" keys to get closer or farther from the riders.

This is good for interesting video clips or saved photos too. Pressing the video icon on the companion app saves a 15 second video -- and the icon press is the end of the 15 seconds, so there's no guessing of when to start "filming". I sometimes hit the "H" key or the Hiding icon on the app to hide all the numbers displays and just watch the scenery. (I need the displays if the ride is hard, so I can react quickly.)

~~~
Teleporting!
This works great. I can warm up with a very easy paced bot group, then use the Teleport icon on the companion app (the "two map markers with a wavy line" icon). It pops up a list of all the bot groups, and I can pick one, then jump directly there, with the usual 10 second pedal assist.

Now I usually warm up at an easy pace, then jump to my preferred group. I've done "intervals" by teleporting to a too-fast-for-me group to see how long I can hang in there, then back to the usual bot group. The whole session is now one recording -- good. I can even just pick a route and free ride to warm up -- the bot teleport still works, as long as I'm in the same map, either Watopia or Makuri.

Got gapped off the group when you weren't paying attention? Just teleport back into it, no need to do that huge chase effort to close the big gap.
~~~
I see that now I can be about 15 seconds ahead or behind the bot, and still be part of the group. It used to drop me off the group at a much smaller separation.
~~~

Redbullet

I think that "drafting" is not a real feature in Zwift. For example: if your weight is equal with robopacers' weight (75 Kg) and the robopacer's power is 1.8 W/Kg, then the robopacer's power is 75*1.8=135 W. If you draft this robopacer, then you should get at least 20% allowance of your power (as in real life). It means that you should be able do draft in these conditions with 135*80%=108 W or less. However, you need a little more than 135 W to draft this robopacer. The trick is that they keep the robopacer always inside the pack, so you will suppose that he is drafting as well, thus, he has the same drafting allowance as you have. But even the riders who lead the pack, input the same power, so there is no real drafting overall, it is as simple as all the riders input the same power (maybe adjusted for their weight on climbs).


In my experience, Zwift "drafting" seems to be an over simplified algorithm, respectively: increasing the speed of all participants who input similar power and form a pack; and if you have a slower internet transmission, you might need to input bigger power because the server might find you out of the pack (although you seem to be in the pack on your screen). Overall, this is a poor simulation, very far from the real life.

zacster

The big difference between Zwift drafting and real drafting is that you don't feel the effect. It is up to you to back off a bit when you get into the draft. IRL, you'll feel the wind resistance change. If suddenly your trainer was signaled that you are in a draft and went easier it would just feel a bit more natural. Instead all Zwift does is move you up or down the drops to show you in the draft. My goal isn't speed but power though so drafting would actually work against my goal. If I can power myself up to draft an individual I'm more likely to want to just power past them which is what Zwift will do. Of course riding a pack is different as the effect is much greater. The big problem I have in the robopacks is backing off to match their pace, I'm always jumping in front.

And .1 is 10%, .2 is 20%, which is what I said. I just didn't remember which went with uphill or downhill.

PeteHski

Did you set your display to show instantaneous power rather than the default 3 sec averaging?

unterhausen

The increased interval where you are considered to be riding with a pace partner was part of a recent upgrade. You can get really far back now. Getting too far in front seems to be a smaller distance, but I have never changed the view to look back and see how far I was in front when I dropped out of the group. I have found that there are problems near intersections with the new code.

spelger

i've been ~30 meters in front with no problems. you do get a warning when you are getting close to being too far out front.

unterhausen

You can be really far behind now and still be counted as riding with them. There is a GP Lama where he tested it. It's a lot more than 30 meters.
Tbh, I'm not sure they advertised that you could be further in front with the update, maybe it's just that you can be further behind.

rm -rf

Testing just now:
There's the warning " Catch up to {bot name}" and "xx meters" for the closing distance. This appears between 25 meters and 50 meters away, but I don't lose the group based on distance--instead, it seems to be based on the time gap.

I wanted to see the bot rider with it's number of seconds gap on the right sidebar, but there's too many riders in most groups.

So I jumped to the 4.2 w/kg, with zero other riders! The gap was growing fast as I was soft pedaling, the 25-50 meter gap warning arrived within a few seconds. It looks like it cuts off at 20 seconds behind, with the bot way out of sight.

But a similar test with a slow paced 1.1 w/kg group, the group drop out seems to be more like 30 seconds ahead of me.
The 1.1 w/kg group was going down a 2% grade when I tried to catch up again. I was doing 250+ watts, and barely closing the gap!

Eric F 

With a bot group, you can ride at whatever effort you want. To stay with the group, however, you'll need to stay within a fairly close range to the advertised pace.

One of the things I will often do is to ride with the group for a while, until I get warmed up, and then try to form a splinter group off the front with a few folks who want to ride a bit harder. Most often this happens going over the top of hills, and keeping the pressure on for the descent.

zacster

You can also teleport to a faster or slower group and continue the ride without break. It'll only let you go to another group within your world. I've been doing that on easy days where I warm at and progress up through the groups until I'm riding at a comfortable pace.

Eric F 

I'm aware of it, but haven't used that feature yet. There's an urgency to being in a small splinter group that is entertaining. Everyone needs to stay on the gas, or you get swallowed up again.

unterhausen

I have had that experience. Riding with a 1.1 w/kg group isn't easy because it takes too little effort. But then get behind for some reason and it feels like they are in an all-out sprint. But it's not easy to catch up to a group IRL either, I think this is one of the things that zwift simulates well. I have a lot of experience getting dropped and having to catch up, and it's all too realistic.
I have gotten dropped by losing too much time taking a drink, and the effort to catch a 1.5w/kg group was enough that I was wiped out and didn't feel like riding with them anymore. I should teleport back, but I only rarely think about it. Or I should take a coffee break to avoid getting dropped.

Redbullet

Robopacers are advertised as increasing up to 10% uphill and decreasing 20% downhill, see "A Note About Dynamic Pacing" at:
https://zwiftinsider.com/robopacers/

However, it looks that they don't behave as advertised, but somehow as you said: rather increase the pace and never decrease. Therefore, I always end up with a bigger average input, although I stay in the pack and I am very close to their weight.

Eric F 

I haven't had that experience, and I ride with bot groups pretty frequently. I tend towards flat to flat-ish courses, and on 0% grades, the bots ride at advertised pace. I also see the increase/decrease on hills.

Redbullet

That sentence is a very sharp and true description of Zwift "drafting".
In more detail, everybody in the pack rides faster that solo for similar power, including those who lead. But you don't really draft, because you don't really get an edge over the leaders of the pack while riding in the pack in Zwift.

Eric F 

Accurate. This is in an effect that I hope they can improve. That said, I suspect it's fairly complex to do it relatively accurately, which would include a variation of effect depending on the number of riders involved.

Redbullet

Displayed power of the robopacers is sometimes misleading. For me it is very easy to observe because my weight is very colse to 75Kg (as robopacers). For example, It happens many times to me to watch the robopacer overtaking me me in the pack with 1.8 W/kg, although I input 2-2.1 W/kg; I have to make a surge to 2.5 or above to catch him. Some other times they look reasonably calibrated around 1.8. Overall, I always end the ride in the pack with an average of 10% W/Kg above robopacer's power.

Redbullet

Well... managing expectations... I'm afraid it will never change because (I assume) this is by design: I guess that everything is managed by the application on the server and it is impossible to have the server managing the situation / drafting of each of those thousands riders individually. If part of controls were running on your computer, then your computer had enough resources to properly manage accurate drafting for you... The same for all the riders.