"Upgrade to next Robo" / Auto teleport if continued pace pushing in front of RP

I don’t like the idea of forcing someone into a PP group they don’t want to be in, just because they’re in front of the Bot. There’s going to be a limit to how fast the Bot can be pulled–that’s part of my own expectations of a Bot’s pace. Bigger group = faster pace. This seems to be a problem that would stop being a problem once someone understands how the groups work. If someone keeps joining a particular Bot group when it’s large and keeps getting upset at the pace going up in those moments…I’d say they need to adjust expectations for reality, rather than having other people kicked out of the group.

Only if they are consistently punching the pace ahead of the bot.

Especially when a couple of them are doing that deliberately and even admitting it (with a sort of “yeah we are doing this because we know it annoys people, so we will do that more because it’s cool”). Well not quite as directly as that but to that effect.

Mind you, they strangely never respond to being invited to join a lap of Alpe du Zwift.

Of all the bot groups, the Constance one is usually the best experience because everyone is going at around the same pace as the bot so it tends to be more natural in the way everything proceeds.

I think this is kind of the only issue.
From other threads, it always felt like people were arguing that they do indeed “expect” an equal effort at any point in time with the RPs, on any route; within limits that is, of course a hilly route will end up with a higher normalized power which should indeed be expected.

The issue however is the variance of speeds day to day / hour to hour with the RPs depending on how many riders are with the bot (or more particularly, in front of it).

Point being, and yes this always seems to end up into a discussion about “lightweight riders”, but it is a fair example to see that while Coco is supposed to be 2.6 w/kg, on things like Tempus, for myself to keep up I end up having to settle into a 3.0w/kg zone, which is pretty far above 3.0; it’s literally above pace of Yumi.

Hence the aiming effect of “forcing” or rather, encouraging, riders to end up with a pace partner that seems closer to what they’re currently achieving.

But yes, all of this topic does indeed fall apart at the realization not all RPs are on the same route at the same time.

It’s just an interesting topic; the fact that someone like myself can ride with Coco at some pace, and then join Yumi, who should be more difficult; but if less riders pulling… ends up actually being a lower effort.

Another point to this that’s been mentioned a few times is to stick the RPs on a Tri bike; which the argument against is “they’re getting the same draft you are.” Which, while true, doesn’t account for everything else. the bots are aiming for power, not speed.

If riders can’t sit in front of a bot to pull it at a higher pace; why can’t the argument be twisted to say “just go up to a faster bot?”

A few years ago when we only had 4 pace partners, or however many, it didn’t make as much sense; but now that we have RPs that are only ~0.3w/kg separation… why not?

Putting the bots on a tri bike makes them more stable in terms of actual pace… isn’t that the whole point, for them to be roughly the same pace day to day?

1 Like

I’m struggling to see why only allowing the RoboPacers to set the pace should be the least bit controversial.


I agree–putting them on TT bikes would be the way to go. And honestly the biggest barrier to that might just be that there’s no Tron TT bike for them to ride, as silly as that sounds.

My objection isn’t about that, but rather about forcing people out of certain groups. I very much wouldn’t appreciate for example joining a PP group because I want to ride with some friends, and being kicked out of that group because I happened to be in front of the PP for a while. If I’m in front and off to the side and the PP is in and out of my draft as we pedal along, am I still doing something wrong and should be kicked out? What if I’m one bike length ahead, because I’m also being pulled along by the 20 people who are stretched out along the road in the actual front of the pack? Don’t like it.

TT bikes would solve the problem. But if Zwift won’t do that, kicking people out of PP groups is a much worse option than people just realizing that group size and route is going to play a role in how fast that PP is going.

They don’t need tron bikes. The new pacers were on a different bike until they “earned” their trons if I remember correctly.

Edit: though there should be a tron TT bike :slight_smile:

Same reason as to why it’s the duty of the group ride leader to set the pace.
Not the bozos dragging folks into and through the fence.

It’s quite literally the same situation; people pulling the RP along at a faster than referenced pace means they’re hugely inconsistent, when the whole objective is to be consistent.

That said I think this has more to do with the lack of Zwift wanting to test these things to find out; or at least doing little study tests in private events to see what people prefer.

I mean, just look at how long it took them to set it so they pushed power up hills and eased on descents? (not that everyone is still fully in agreeance with that either to be fair [but I think that’s more of a function of the route, as the pace pushing is annoying on excessively long climbs])

Also @ Tom, while my OP may have made it sound more ‘forced’, I saw it more as the function we have when warming up before a group ride or race, where it has the “Ready to join the ride? Yes/No” popup; or alternatively the one after where it says “Return to X world?”

Oh yeah, and Aaron is right, that was funny actually when the RPs got reset and they were on the stock Zwift Carbon frames for a few weeks :laughing:
that’s been a while since I last saw that, but it was hilarious.

A gentle nudge popup could be a good idea, for sure. (So long as it was a popup that would go away eventually without having to click on it.) There definitely are people who are pulling the pace on purpose, and there are also people (I’d bet) who are doing it and aren’t even thinking about it or realizing it. An appropriately worded message could do a lot.

No draft was done before but everyone complained loudly:

Ooooh Coco is too slow now!

They couldn’t get their 60km in 90 minutes on Tempus.

That was back in the times of sticky draft.

When, back in the days of PD3 and before?
If so I can definitely see that being a huge issue back then, when a blob of D’s could outpace a solo A rider lol

Also I’m assuming if it were back then, that’s when there only were what, 4, maybe 5 pace partners anyway?
So that test definitely seems like it would’ve gone poorly.

I think with all of the changes we’ve had since then should warrant a re-test of the concept.
Slower packs in general, and now 9 RPs?

In theory all it would likely mean is folks used to being with Coco are likely to be up with Yumi. (maybe; this may not be true for everyone).

Just to put on record;
Coco has been on Tempus this week.
Riding with her last night for 20m as a warmup required 2.78w/kg for me, seemed like a “smaller” group (probably 70-125ish folks, and not nearly as many were in front of Coco).
I had done a 20 minute warmup ride on Monday as well; it required 2.93w/kg, with a huge group (200+ I’m sure, majority of the field in front of Coco)

I’d be curious if it’s closer for heavier folks, again I have no idea how much of a weight bias problem this is versus anything else.

Regardless; those two data points should prove how much a RP can be affected; that’s a pretty significant difference IMO for the same route; I know folks have pointed this out before, but there’s just one more data point.

The result was the 2.5w/kg Coco bot had no draft and the mega coco group had to cruise along at 1.5-1.8w/kg to go with the very slow bot.

Back then the group pacing was a heck of a lot easier (even with draft for the bot turned on) - especially for a light weight rider like me. I would do around 2.5w/kg in that group. Now I have to do 3.0w.

Anquetil (A grade bot) was not as overpowered as Constance BUT Anquetil was 65kg and rode the mountain routes. There was always a consistent 6-8 people in the Anquetil group. You’ll need to take my word for it because I didn’t get any screenshots. I would ride with that group daily for 75km and get around 1000-1200m+.

Now it’s the same amount of people but with the bot being much heavier, much faster 315-345w and on very boring routes so I don’t ride with it anymore. The 4.6w/kg uphill is too much as well.

It is More enjoyable riding alone on ADZ or Ventoux. You’ll find me on ADZ daily.

Pacer bots have been a mess since around February of 2022. Previous to mid-February I could ride with the OG Coco at a steady Z2 pace (for me). I have a ride on Feb 18th with a note in Strava “Not Z2 ride… Thought it felt a little quicker than previous C Bot rides and it was!” I averaged 3.1W/kg and 150bpm HR for that ride which was considerably higher than previous CC rides.

Ever since Zwift has been tinkering with pace partners, now robo-pacers, and my experience with them has degraded.

I did a ride with the new Coco last week on Neoyoko all-nighter and averaged 2.8W/kg. I did a ride with Maria Monday on Tick Tock and averaged 2.5W/kg. Both I averaged above the advertised pace. I was attempting to stay close to the bot and maintain a steady pace.

I don’t know what the answers are, but the pacers need to own their respective rides and not get pulled by pace pushers…


I did 90 minutes with Maria on Tuesday and averaged exactly 2.2 (163 watts) at 74kg. :man_shrugging:t2:

How big was that group David? Or was it fairly small with Maria mostly able to stay in front?

Sven: ever since they made them variable I’ve not been a fan either. I still much preferred the static pace they offered before. I do feel like I could join Coco, and was able to have an extremely steady ride power wise.
Ever since that change it’s always felt like things are far less stable than what I remember them being.
I definitely get the advantages of the variable paces (folks do like to push a bit more on hills as one would naturally outdoors).

Definitely has ended up a different [pedal] strokes for different folks.

Granted something I just realized; the reason why ZHQ may not want to play with these things now through the winter months is exactly that; so many more riders; I don’t think a lot of folks that only ride on Zwift during the winter months expect things to change or “testing” phases like those of us who ride all year have seen throughout the years / particularly changes through the summer months…

About 100. She was usually somewhere in the middle. My NP was only 2 watts above my AP, so I guess I did a half-decent job of avoiding too much surgjng.

So yeah that sounds about like my experience with Coco yesterday; where rider numbers were quite low compared to my ride on Monday.
Which to be fair was also separated by ~2 hours time wise.

I didn’t actually check my average vs NP, shouldn’t hurt though since I’ve already got two comparable rides mentioned anyway. Both of these were in the 22-24 minute range for the record as well, 1 mile distance difference however, but that should be fairly minor. Both contained one full lap of Tempus at the very least.

Okay well… so much for that being interesting; both are literally 1 watt away average to normalized.
159/160w on Monday
172/173w on Wednesday

Another thing I’m curious is how much that scales out; for heavier riders / those at higher watts, will that begin to expand? Is it even possible that say someone at 100kg would see a ~0.2w/kg difference one day from the other? ie: my 14w difference; would that expand out to a 20+ watt difference for someone heavy; and would it be that drastic? I guess this is more health and power capabilities focused of a question but still curious.

Anyways, as far as Robo Partners goes; that’s a 8% variability between days / different times; for something that should be … well, a robot going at the same speed.

8% seems a bit of a stretch for something that should be equal, no?

I’m sure there are worse examples too that might push that out to 10% or greater too; especially if you end up with a fresh bot and nobody around. Either way, I don’t think Zwift ever intended for the RPs to be this different to keep up with.

But that just sounds like an issue of recalibrating expectations. Yes, if the PP is unaffected by draft and the people in the bunch are affected, you’ll now be putting in less effort than the bot, unless you’re at the front. So we could just let that affect our decision of which bot to ride with.

At some point you’re going to run up against the fundamental issue of it being a bot–unlike a real person, it cannot adjust for what the people around it are doing. So it’s going to plow ahead at it’s set w/kg, and it can either be set to be influenced by others or not. And if we don’t like having to account for the disparity between its w/kg (not drafting) and our w/kg (drafting), we can either A) choose a TT bike ourselves and then our w/kg will be similarly unaffected, or B) go back to letting it draft, and recalibrate there by using our judgment to take group size into account when we choose a bot.

So here’s a question: aren’t you in that group being affected by the draft of the group in the same way that the Bot is?

If a Bot is listed at 3 w/kg, and you go in expecting 3 wk/g, but the group is big and so the Bot gets pulled along at faster speeds…why aren’t you being pulled along as well? In other words, if the Bot maintains that 3 w/kg but is getting pulled to faster speeds, shouldn’t you be able to maintain the same 3 w/kg and get pulled by the bunch to the same degree?

Am I missing some fact here about how the Bots work? I know they vary wattage by grade, but are they also varying wattage by speed too? I didn’t think they did? I thought that a 3 w/kg Bot would do (roughly) 3 w/kg on a flat if it was all by itself, and 3 w/kg on a flat if there are 100 people with it. The speed will go up in the latter case, but I thought the w/kg stayed the same, regardless of bunch size or speed. And if that’s the case, why isn’t your 3 w/kg being affected the same way?


What’s different with the bots is that they sit at a set power, while humans have to change gears. We don’t just maintain same cadence at 315w irrespective of the terrain.

When it gets steeper, we change gears and when we do, our pace varies. I think that could be part of how the bots can zoom forward at times at the start of hills.

That’s what I’m trying to figure out.

The only differences are the number of riders around / in front of the bots respectively.

On the same course, thus same terrain; indeed, how should I be seeing such a large variance in power when trying to sit besides a bot.
Doesn’t make a ton of sense does it?

The only changing factor here is just that though; the size of the blob. (and their position I suppose, but that’s hard to really ‘calculate’ / make notes of)

As for the “don’t you get equal draft benefit”; yes, but apparently no.
This is what I’m curious about to hear from heavier riders though; if they experience 10% variance in their rides with bots like my numbers have shown; or is it more stable for them?

I think the reason for the draft situation is that my examples were on Tempus. So while in theory, sitting mid-pack it really should be equal power… right?
The going theory has always been that “6” riders back, there is no more draft benefit offered in Zwift. But I’m not entirely sure this is the case.

Alternatively; when the group was larger, the front riders were just pushing that much harder, dragging more people along, also pushing much higher power, and what we ended up with is a higher ground speed.

Which… actually let’s test that theory… (goes back and looks at Connect stats again)

So in theory, while draft benefit is the same, obviously to go faster, we have to put down significantly more power to overcome drag (we all know this).
Draft benefits we know have their limits in Zwift (in the real world you can get behind a car or truck and go 100 mph… we also know we definitely cannot do this in Zwift).

So stats wise, speed difference is there, but less than I expected.
24.6 vs 24.9mph
Max speed (which would be that descent in the Tempus oasis), matched at 28.4mph

Frankly Tom… it’s a great question LOL :rofl:

0.3mph is definitely nowhere near 14 watts

And no; the bots indeed only alter power on climbs and descent, beyond that they hold a steady watts no matter what. They don’t care about their ground speed (assuming they even have the ability to consume that information directly from Zwift in the first place)

I’m almost curious if there’s a way to record one of the bot’s data for a day, and see how much it varies even in speed throughout a day.
That would be a rather interesting test; maybe not exactly telling of how or why, but it’s clear there is definitely more to draft and pulling in Zwift than some people like to think.

If anyone else is getting slightly more confused by all of this; it’s okay because I am too.