Today, we’ll focus on a race format that was unveiled over the weekend as a Zwift Labs event: Beat the Bots. These events are set up on Zwift as group rides for now, but let’s be realistic: we were racing to stay away from the bots, then sprinting to the line to beat out teammates at the end.
And I, for one, welcome our new robot overlords. So let’s dig in!
[adinserter block=”1″]
Beat the Bots Basics
Beat the Bots events, as organized in their first iteration over the past weekend, work like this:
- There are just two groups of riders: the humans (group A) and the bots (group B).
- Humans leave 14 minutes before the bots
- The bots are holding 5.6 W/kg.
- Both groups ride on Watopia’s Big Flat 8 route (31.5km, 109m of elevation).
The goal is simple: to cross the finish line before any bots catch you.
Ride Experience Notes
I did this event yesterday, in the 8am PST timeslot (see Strava activity). I basically treated it like a faster group ride, averaging 234w (2.7 W/kg) and finishing around 4 minutes before the bots. (I say “around” because I have no way of knowing, since those stats aren’t tracked or shown anywhere. But more on that later.)
The experience was fun, and certainly different from a typical Zwift group ride, scratch race, points race, or TT. It was most comparable to a chase race, except we weren’t working to catch any riders up the road. Instead, our whole goal was just to stay away from the bots.
So you end up feeling like you’re working as team with other riders in your group, which is very different from the feeling you get in a typical Zwift race. Of course, once you’re a few hundred meters from the line, all bets are off, and you sprint to the finish with whatever legs you’ve got left!
The biggest thing lacking in the overall experience, I think, was not knowing how far back the bots were. The only way to figure this out was to click through the rider list until you found the bots. It took me around 10 clicks to scroll all the way back. Then you could look at their distance, compare it to your distance, and do the math to figure out how far back they were.
(This isn’t a criticism of Zwift, by the way. I love that they’re doing these “quick and dirty” Zwift Labs events, testing out formats to see what works and what doesn’t before putting more development time into polishing them.)
Here’s some of the feedback being shared in the Zwift Labs Club chat, where Zwift has asked people to post feedback for ZHQ:
Fun For All
How can these events be made more fun for everyone? This is an important question to answer when it comes to any ride format in Zwift, because Zwift needs to remain an inclusive platform where everyone can find motivation and fun, regardless of fitness level.
There are different directions we could go in how these races are set up. Some Zwifters have suggested breaking riders up into standard categories so slower riders (cat D, for example) could be released before faster riders (cat A), giving everyone a group to ride with and different time gaps so everyone is working at approximately the same relative effort level to stay away from the bots.
But I want to flip that idea on its head a bit, because breaking it up into categories like that just makes it into a standard chase race, with the addition of bots being the fastest group that is released last. That doesn’t feel novel enough to me.
What if we categorized the bots instead? All the humans leave en masse, then the bots leave as a group too, but some bots are faster than others. Perhaps something like this:
- A bots: 5.6 W/kg
- B bots: 4.8 W/kg
- C bots: 4.0 W/kg
- D bots: 3.5 W/kg
Basically, you want each group of bots to be stronger than the corresponding category of humans, so if they’re able to catch up to the human racers, humans can then do their best to sit on those bots’ wheels to the finish.
If the race was structured in this way, riders could track their fitness progress by seeing which bots catch them each race. You might get caught by all four bot groups the first time you race, but as you get stronger, you may only get caught by the B and A group… and eventually, maybe you’ll stay away from all the bots!
Why Join a Beat the Bots Race?
Beat the Bots races would be a special experience for two reasons:
- Teamwork: you need to work with the group of riders you’re in to stay away from the bots. This fun team aspect isn’t seen in most Zwift race formats.
- Experiential fitness progression: as mentioned above, you can track how much stronger you’re getting by tracking how long it takes the bots to catch you from one event to the next.
- Work as hard as you want: every rider, regardless of fitness level, can get a good workout in this format.
- A special “virtual-only” format: this is a format that doesn’t really have an outdoor equivalent, and there’s something extra fun about that.
Required Game Changes
What would Zwift need to implement in order to make Beat the Bots work well as a native race format? With the disclaimer that I’m not a game developer or platform engineer, here are a few things I can think of:
- Orientation UI: riders need to know what they’re signing up for, and how the event will work. On-screen messaging should explain this clearly and succinctly.
- Bot tracking UI: the event would be more engaging and motivating if you know where the bots are on course, how fast they’re closing the gap, etc.
- Results Screens: a race results screen at the end of the event would be nice, and perhaps the existing results screen could be modified to show which riders got caught by which bots.
- Post-ride chat: since you’ve worked with “teammates” throughout the event, it would be nice if the game kept you in “event mode” after you cross the finish line, so you can chat with other riders before returning to the open Zwift world.
Your Thoughts
Have you tried to “Beat the Bots” yet? What did you think? Share below!
I’m glad someone has finally posted about this! Did my first attempt on Monday (AU) and we simply didn’t see the bots at all..!
Tried searching for them on the companion app but didn’t know what to even look for.
Felt very much like co-op mode in a videogame. With some refinement, I can see a lot of potential for this type of race/event.
I`m a D Category and tried Beat the bots. I rode like mad (for me) expecting to be passed at some point by Cat A and B bots which is what I thought Zwift had done but no nothing. Felt like a letdown and the lack of direct feedback on bot advance was also disappointing. Addressing these 2 issues would make the game format much more interesting for repeat experiences.
It seems like this is being “community designed” into a group ride which already existed or a chase race… which already existed. Chase races also have no visibility (Natively) to the packs.
IndieVelo has really good bot racing with race craft. I wonder how long it will take for the community to suggest each feature on IV so Zwift can copy it without copying it.
I think Zwift are having to respond quickly to keep people’s attention. With the recent price increase, quite a few are testing the alternatives (like myself) and IV is a very real threat with some interesting advancements already. It’s great that Zwift see the need to test things out and hopefully it will benefit everyone as a result with ‘new’ features.
I would like chase races so D first than C etc wth 3 minutes intervals and the chase bots say 2 minutes after their category so D bot just 1 minute before C group
Why not just do a chase race then? Why add the bots?
Each cat beat their own bots and higher cat beat the lower cat bots
It was just a thought
what if the chase lasted until the last human has been caught? More incentive to work together longer. I doubt any B or A riders have seen the bots with such a huge head start. I’ve done 4 and we always end around 42’. The bots would have to 28’ for 31.5Km. A hillier course might introduce some decision making, do we push or keep the blob together longer?
One more idea for Zwift: a last man standing ride. Set course to complete in a set time. Reduce the set time by a pre determined amount for each re-start. For example, 5km loop. Start on 10’ then 9’30, 9’, and so on. Mass start of course.
Last human caught sounds fun! Could also copy from the format of Wings for Life World Run, where the chaser accelerates every half hour (or so) and so gets faster the longer the run goes…
Be cool when on the same route you had to beat the bot and also your hologram of your best time on that route. Double the way to measure progress.
Would be cool if these things looked like zombies and once they catch you, either you turn into one too and get the keep-together elastic effect that you get in group workouts and you get to rip around to the finish eating other players
But the time gaps need to be seriously addressed to even make it feel like you’re being chased.
Nice article and great idea for this series!
I’m all for Zwift trying things and improving them on the fly!
Not sure if you’d be interested, but would love for you to have an on the record chat with a top Zwift exec to get a sense of where Zwift is going in the future. It’s fun to try reading the tea leaves, but many of us would love to hear from Zwift in a more direct manner. And you’d be the perfect person to do this. Just a thought…
I was a bit confused about what was happening – a big group went ahead and went hard, I wanted a medium-effort Z2 ride so didn’t go hard initially and ended up in a bunch of about 20 riders. We mostly stayed together apart from on the climbs where a couple would push hard and split up the group a bit. We finished a couple of minutes before the bots, averaging about 2.7 w/kg.
Would definitely have liked more information, especially on what was going on and where the bots were!
I’m all for developing novel formats and elevating races. But so far what I am reading here and in the previous episode of your series, is that these events are making Zwifts technical debt to host such events painfully clear. Which honestly leaves me somewhat confused. Things like a UI for time gaps between groups have existed in the feature requests from at least 2017, but racers were always dismissed as the unimportant minority bunch. That can of course be a corporates’ position, but it seems a bit late to change this position now that all these limitations have long been solved and implemented properly on an alternative platform. I wish Zwift well, but I can’t help but wonder if the time and effort (and resources) wouldn’t be better spent by collaborating a bit more.
Tried this over the weekend and it was fun and different. Definitely a workout. I like the idea you offered here and other ideas I saw like having us chase the bots moving at different wattages to see which ones we can catch before the race ends. Somehow I am more motivated when I can see in front if me who I am trying to catch versus trying to guess or calculate on the move when a bot might catch me (assuming that info is available for viewing).
Good idea chasing Bots with different watts better than being chased adds more incentive for me as well
Intriguingly, when signing up on the app there were 2 groups. The B group said for bots only and a clear message “do not sign up for this group if you are a Zwifter”, so naturally I signed up for that group. It registered me as being in the bot group. I de-registered but wonder what would have happened had I tried to join as a bot (and also why the bots need to sign up at all…do. they have agency??)
But aside from that, it sounds like an interesting format if they can iron out the wrinkles you mention.
I was watching one of the Beat the Bot races, I think it was actually the one Eric was in. I signed in as “A” and then never pedaled, that let me select the bot pack as my rider view. There were 6 riders in the “B” pen, only one of whom actually left with the bots, his goal was to see how long he could hang on riding with a group of Big-Mig-Bots. The interesting thing, one of the bots got dropped by the pack was was 30 seconds off the group when they caught the first rider. The sad thing is, when they caught the rider nothing happened.
I can’t try these events because of the fixed schedule zwift have decided to run them, but conceptually I don’t understand why the bots should be a pack. If they had multiple bots at different power levels on TT bikes then instantly becomes a challenge of how many bots you can beat rather than an all-or-nothing situation.
Also, they should be able to lock humans out of the bot pen by adapting the women’s only pen mechanism that already exists.
Is there already a mechanism to make beacon-riders more visible in events? Then giving the bots beacons would then work to show where they are better.
I think the event was too easy for higher riders. In the group I was in, we had already covered 1/3 of the course by the time the bots were released, and so the enthusiasm just died down as some geek told us we only needed to hold around 3.5w/kg to the end having averaged around over 4.5 up to there. But it definitely seemed about right for those further back!
The bots were too far off the map for almost everyone. The time gap was too large. Their speed was unusually high. Something like volcano laps would be more visible.
It was more fun to be the bot and join the bots. I only lasted for eight minutes, as it was an extreme pace to hold. But it was a blast, I must say. Do (not) Join was great. Could they slow them to 5 w/kg next time?
Did same race as you and paced with you before group behind caught up to us (or we fell back. . . 😉). I enjoyed it. Like many said already would be good to know where the bots were.
I think after the initial group of bots deploy, the bots then leave at intervals. I was passed by multiple groups of bots but still had bots behind me when I finished.
A few bots fall out of the draft on a slight climb and get distanced by as much as you saw. 😀
Did it yesterday and enjoyed the ride. averaging about 2.7 watts per kg. No real complaint other than what was noted. Need a easy way to track the bots progress. I didn’t actually know if we beat them? I assumed we did since we were never passed. Be nice if there was some sort of recognition at the end as well. Like “you beat the Bots” or something. We had a lot of chat in the peloton about keeping the pace down as we got a little strung out. Didn’t know we should race each other at the end. I rode more like a TTT or ITT keeping pace steady and group together
Simply a bad format and a bad concept. Are we introducing bots because of the plummeting numbers due to subscription hikes and summer cancellations? Certainly seems like it. This game needs a new CEO.
Eric Min’s resignation should have been immediate following the hardware fiasco, that much is clear. However, for us users, it would not make a big difference. As one person mentioned on the Zwift forum, the company is facing a significant technological debt. In order to tackle this debt, they must start by rewriting the core engine. But that requires a substantial amount of money and time, both of which they lack due to investing in the hardware venture. Maintaining a complex code is also expensive. What would you do if you knew that making quick changes was not an option? If it were up to me, I would work on some PR strategies and reach out to all channels, even those so-called fan pages, to promote a positive outlook for the future. It is important to give people something to talk about, something to engage with, and build up some anticipation.
This is a general problem of all races in Zwift. Unless you have Sauce you have no idea what the time/distance difference between breakaway and the pack is, or between various groups.
There’s no real incentive to work together as a single group, as all you need is stay away from the bots. Now if the bots would speed up as soon as they catch someone, this would give an incentive to keep others in for longer, as dropping them does require you to finish faster. And if you do get dropped, you would work as hard as possible to give the bots the bonus speed as late as possible.
If Zwift would add in some sort of meaningful reward (centimeters of yarn to be used in a shop, or bricks for your club’s virtual club house) that would apply to the group as a whole based on the number of riders finishing before the bots, this would give even more incentive to work together.
I did this event Sunday and Monday, and I rarely return to an event, but Sunday was quite fun so gave it another go Monday.
On my first one Sunday, I got in a slower group and broke away on the last 3Km. when the bots started to show on the map – of course failed to keep them behind me, but that was great fun, both riding with the group and getting caught.
Monday I was in a bigger group, that suited my current power, and we managed to keep the bots behind.
With some better visibility of where the bots are ( maybe a timer on the HUD ), I think this has potential for some great group riders, with room for chatting, working together ans/or just push like crazy.
BòóX
I think Zwift should get their app working properly first. Also they should not try to fob me off by referring me to forums etc. I don’t have time to read through all that crap.. My app did work correctly and now it does not. It won’t show any events I enter . Get it ? Great ! Thanks. Get the basics right before trying to build something extra.
This actually sounds like alot of fun. Just like using groups on Sauce, this would be far more interesting if you can monitor the bots so you know where they arenandnapply effort accordingly.
Also mass start is a bit of a hit and miss thing… It sound great in theory but what happens is riders get dropped spread out so instead of a D or C field working together to keep in front of their bot, they are all seperated and have to solo against the bot unless they get lucky to form a group.
Like all racing, it would be best if there wasn’t to many events so the participation rate per event is high enough to make it work.
Join the bots 😛https://www.youtube.com/live/FN7hHJ06hn8?si=KYGfolez-8TeRJeX
Never tried it on account of the power difference. Having power graded bots would be much better – but make them at the top end of each power level so we have to work together to beat them
I thought it would be cool to have the bot pace a little more varied, and perhaps have bot comments annoucing their pace for everyone to see (“ok we’re going to push it to to 6w/kg for the next 3 min).
Obviously there would have to be an overall w/kg target for each bot group, but having them ride in a more unpredictable pattern might make the chase more interesting.
And overall have them start much sooner with an overall lower w/kg than they currently have. Maybe they even pass the humans at some point on a climb but get reeled back in. Not just a pure chase race.
Great suggestions. You echoed my thoughts as well. I thoroughly enjoyed the group ride but I was running away (or shall I say, riding) without knowing where the bots are. Being clueless does not help. But we had a great group averaging at 3.3w/kg till the end. I would’ve loved to see my score. I mean by how much gap did I beat the bots. No post ride results are disappointing. Hope Zwift will consider this feedback.
I am posting this late because I was traveling last week, and I’m just getting back to Zwift, so I know the conversation has moved on… But, here goes my $0.02:
I only did one of these workouts last weekend. The blob I was in was caught about 4 km from the finish. However, being caught was a bit anti-climactic. This is a video game and we’re being chased by the Robot Overlords!!!
Recommendations:
Having those caught turned into Zombie Pacers and then « switch teams » to be on the chasing side (with rubber band to keep everyone in) would slowly increase the mass and speed of the pursuing pack, meaning everyone is in the event until the end and it’s a last person riding challenge
Really seems like Zwift is crowd sourcing ideas with these events. 5.6w/kg bots that start 14 minutes behind is so pointless that you wonder why they bothered but maybe that’s the point? By being so completely pointless it encourages people to suggest all sorts of ideas for how an event like this should actually operate.
It’s like the meme that if you want an answer to a question on the internet you post the question first and then deliberately post a completely wrong answer as a reply. People on the internet won’t rush to answer questions but they will rush to tell you how wrong your answer is.
reflecting on a lot of the comments, time gap too big, anticlimactic when caught, too easy for some, too hard for others, keep it fun for everyone.
Here is the suggestion: reduce the time gap to 1’ per 10km of “flat” courses, reduce the single chasing bot power to 5.3w/kg (just a guess, no hard science.) When someone gets caught, they transform into a “zombie bot” (adding to the pack of bots at 5.3w/kg) which should increase bot pack size and speed progressively. The people who have been caught and turned into bots can keep riding along with other “zombie bots” but they don’t have to push the 5.3 w/kg as they are linked to the original chasing bot.
I really like your idea about categorizing the bots. That would be brilliant!
I rode it too – we never saw a bot. It was a group ride, without a leader and with someone from ZHQ pootling around. It could be really good, if finessed into a hare and hounds type event and with shorter gaps between the bots and the riders. One thought though, would it be easier to just wait, jump onto the back of the bots and then out-sprint them?