It was almost exactly two months ago that Zwift announced they received a whopping $120 million USD in new funding, whereby a large chunk of that would be spent on digital racing – known as esports.
Since then, the company has been on a blitz to draw your, and sponsors’ attention, to the racers and teams which compete each week in their KISS Super League race series. The races are live streamed, with the women competing on Tuesday night, and the men following on Wednesday night.
But just a few weeks after the December announcement, Zwift company partnered with Cycling Australia (the national cycling federation of Australia) to offer a full national championship jersey to the winner of a Zwift event held in conjunction with the federation’s annual national championship events in Ballarat. That’s right, the winner would get the exact same national jersey as any other national level title – this one being for the official event of “Cycling Australia eRacing Criterium Championship”.
Of course, while the event itself might have been entertaining to those in the room – it was more of an example of how far Zwift still had to go when it came to putting on a professional sports event where real-world prizes (such as money or national titles) are at stake. So much so that Zwift has since walked back slightly on their ambitions for 2019, saying instead that remote Zwift racers could act as qualifiers to real-life national finals – rather than national titles altogether (but more on that in a minute).
Still, Zwift’s weekly events as part of the KISS Super League are making good progress upon each iteration, towards at least one element critical to any sporting event: entertainment. But entertainment only works when people trust that the results are believable. And that’s where most of the challenges Zwift still has yet to resolve going forward.
So for today’s post I’m going to dive into many of the elements that lie ahead on Zwift’s self-proclaimed path towards not just more formal national federation level racing in 2020, but their oft-noted goal of getting e-cycling into the 2028 Summer Olympics as an exhibition sport.
So many ways to cheat:
The first elephant in the room to discuss is cheating. Anytime you talk about sports you have to talk about cheating. And it seems like anytime someone talks about Zwift, cheating isn’t far behind either. Today, there are essentially four core ways one can cheat in Zwift – racing or otherwise:
1) Traditional doping (drugs)
2) Mechanical cheating (tweaking trainers/power meters/motors)
3) Weight (setting an incorrect rider weight)
4) Not being who you say you are (specific to racing)
Now, unlike the infamous ‘motor doping’ scandals of the past few years, the technological focus here is more around the trainers themselves.
See, in a race like the KISS Super League, the riders are split in two types of locations: The main event venue, and remotely (such as at home). For those at the main event location that week, as it was, Canyon’s HQ this week – they’re riding on whatever trainers Zwift provides. That’s typically part of a sponsorship deal for the event that week. Whereas those riders at home are riding on whatever smart trainers that they have. They could be a high-end Tacx Neo trainer, or, it could be a lower-end Tacx Vortex.
The difference? Accuracy.
In the case of the Tacx NEO trainer used at the event venue last night, it’s unquestionably the most accurate trainer out there – claimed at +/- 1% (and oft validated as well at that). Whereas that Tacx Vortex is only accurate to +/- 5%. That’s a *massive* difference in this event. And in fact, most trainers at the lower end tend to be least accurate at the most important times: Sprints.
In my testing, I’ve seen numerous cases where trainers that claim +/- 5% accuracy will be off by many hundreds of watts, again, easily the difference between winning or not in a race like this.
Unfortunately, there’s no standardization today on this within Zwift or these events. Zwift doesn’t require anything more than a smart trainer or power meter. Nor does Zwift do any validation of accuracy themselves, or certify any trainers for accuracy.
Zwift today looks at racers in two piles:
1) At the venue
2) Remote (at home/etc)
In the case of at venue events, the riders are provided trainers that they are required by the rules to use (more for promotional reasons than anything). Whereas for remote riders it’s up to that rider to pick their trainer. For example, they could choose a trainer like the CycleOps Hammer 1/2 which is well known to overcommit on sprints in Zwift, thus resulting in higher power numbers than reality during that sprint. In fact, in Zwift racing circles this trainer is popular for that singular reason.
(Minor update: Zwift has clarified that for the KISS Super League series specifically they, and Wahoo, have sent all teams Wahoo KICKR 2018 trainers. However, some teams already have existing sponsors, like Canyon with Tacx, and Wiggins with Elite. In those cases those teams will ride the sponsored trainers, whereas others should be riding the KICKR’s. Zwift argues this makes it more even, but I’d actually argue in this case they chose to send out the one trainer that’s easiest to tweak, out of virtually every trainer on the market. It takes no tools or special equipment/software to do so.)
Still – all riders are actually treated equally in the race. There’s no separate category for people in the venue versus at home. It’s just one race.
But there’s actually a bigger issue than that at play – tweaking calibration of the trainers. All trainers today except the Tacx NEO series require some form of occasional calibration or spin-down routine. It’s during this process that on some trainers you can tweak the calibration to give you a higher power level than reality. In fact, it’s so trivial to do on certain trainers (such as a Wahoo KICKR) that a rider could likely even do it sitting at the race venue in front of everyone without anyone else even realizing it. And that ignores all the riders at home that have all the time and lack of transparency to do as they please.
The key with cheating in this manner at this level is that you don’t need massive gains. You just need a few percent more to win.
These are real problems that have to be solved, and to Zwift’s credit, they do seem to recognize this. When I asked Zwift about this issue, Steve Becket (Zwift VP of Marketing & Customer Acquisition) noted the following:
“You are right to point out that some hardware can be ‘gamed’ to produce higher power meter readings. Across 2019, Zwift will be working with hardware manufacturers to mitigate this from 2020. It is very likely that hardware restrictions for racing will tighten up in 2020, but Zwift needs to acknowledge that hardware manufacturers need time to respond to this feedback as trainers are developed, in the most part, for training – not racing.”
And he’s correct. As it stands today, if I were running a racing league where I was dolling out large sums of money or high-value prizes (like a national championship Jersey), there’s frankly only one trainer I would allow and use. Everything else is just too easy to game and cheat with.
Note that there’s an interesting underlying sponsorship catch here too: Individual rider or team sponsorships. For example, last night’s event at Canyon’s HQ has on-venue riders on Tacx trainers. That’s because the Canyon team is sponsored by Tacx. But what happens when a venue is shared by multiple riders/teams (as was the case of the Cycling Australia event?). Well, in that case, it defers to the sponsor for that particular event race. But that gets tricky. Virtually all pro-level riders have team sponsorship with existing trainer companies (primarily Tacx, Elite, and Wahoo). But even more complicated is if there’s an individual with another trainer brand – perhaps Kinetic or JetBlack. And what happens if that trainer doesn’t meet the accuracy requirements set forth for 2020? Is that rider required to ride on a brand that’s a competitor to their own sponsor (which is undoubtedly a violation of most sponsorship clauses)? Lots of outstanding issues still.
However, that still doesn’t solve the next issue of weight. There have been discussions that riders will need to weigh-in on a scale using a webcam to validate their weight. That raises plenty of issues. Most riders will want to wear as little as possible for this – since the lower the weight the faster you’ll go, given a specific wattage.
But that quickly leads to videos of young women standing in bathrooms in nothing but their underwear being sent to an unknown place with unknown security controls and unknown people viewing them. Likely a better solution is to simply require all riders (male and female) to just wear their cycling kit (jersey/bib) when weighing in with a webcam. Still, these sorts of things haven’t seemingly been vetted through yet and tweaking your weight is by far the most common way to cheat in Zwift.
(And weight scale integration without a webcam ignores the reality that simply pushing up on a nearby table/counter while you weigh-in means you can easily shave weight off the reading of a digital scale.)
The weight validation actually slides right into the next category – which is whether the rider competing in a given event in Zwift is actually the rider they say they are. For example, a remote female rider competing remotely in the Zwift KISS series could actually be a male rider – one that more than likely would be stronger, and likely earn more points. This could be solved easily through the usage of simple web cams being required for each rider, but again, gets to the lack of a cohesive platform for how to handle the unique requirements that a professional sporting event has on Zwift.
And finally – that still leaves issues around traditional doping that continues to be a major challenge for both amateur and professional cycling today. Zwift hasn’t even touched this aspect of things, instead mostly deferring to local federations. In some races the rider must be a member of a local federation which may (or may not) do drug testing. No testing is being done on-site at this time for athletes at a given venue.
Analytics and Results:
The next challenge Zwift faces is oddly enough probably the easiest to solve: Results
Today, results are handled in two manners: First, live results are (sometimes) shown within the Zwift app/livestream itself, immediately upon completion. But these results aren’t actually considered the official result of record. Instead, that is deferred over to a site called ZwiftPower. Zwift today considers it a requirement that all KISS World Series riders register on ZwiftPower.
Still, to backup briefly on the live piece, let’s look at last night’s women’s race – there were no instant results as riders crossed the finish line. Instead, you just saw this:
Followed by understandable confusion for a while from the commentator. They even pulled out a static finish line camera replay, but couldn’t overlay any data to support it. No timing data shown at the one point it matters to show it.
To be clear: The entire point of a race is to win the race. Thus, if the broadcast must get one thing right, it’s knowing the winner at the point the winner crosses the line. In other words: You had one job…and didn’t do it.
It wasn’t until some 90 seconds later this was shown:
But the next piece is actually more relevant beyond KISS. See – while Zwift handles results that you see on the screen after a typical community race/event, the official record keeper for most events (including KISS) is actually ZwiftPower. This community supported site tracks results of all scheduled events, both major/minor races and simple group rides alike. The site used also receives funding from Zwift to operate.
The site has two main purposes in life:
A) Tracks race/event results and allows you to do some cool analytics/historical stats
B) Disqualifies riders that appear to be cheating
The first one is great – and is an awesome use of a 3rd party service. Of course, Zwift lacks strong API’s to really let ZwiftPower do the kind of cool stuff that it could potentially do (similar to what we see apps do on Strava’s API’s). But that’s a different discussion for a different day.
It’s the second one that’s more challenging. In this case, ZwiftPower actually acts as the arbitrator of whether or not someone is cheating based on historical results and a black-box algorithm. This runs under the name ZADA (Zwift Anti-Doping Agency), but in reality, it’s got almost nothing to do with Zwift itself. Instead, it’s ZwiftPower’s site that enables ZADA to do result verification. And the only doping they deal with is when power numbers don’t look right (even if they are).
Now, I don’t blame ZwiftPower/ZADA here. They started years ago and were trying to fill a void in amateur racing on Zwift. But the workload simply got too big, and back in early January they disbanded ZADA and stopped performing anti-cheating type activities on any/all results. There were also concerns around the defensibility of ZADA against real-world lawsuits from riders that felt they were wronged. For example, if ZADA flagged a rider as a cheater, that could have real-world implications for a person’s life/career in the ‘hurt but ask questions later’ world of social media today. Especially for something that could be as simple as a crappy and inaccurate trainer or power meter.
Of course, this was at the exact moment that Zwift and the KISS Super League needed it most. So in mid-January Zwift stepped in and funded ZADA to cover the KISS League events.
Ultimately though – this doesn’t belong with ZwiftPower. This belongs with Zwift, at least for now. Zwift needs to take responsibility for both results and anti-doping efforts directly. Of course, down the road, a 3rd party organization, à la WADA, should own this, but I don’t think that ZwiftPower really wants to be that organization.
And for results, Zwift.com should *always* be the first stop for finding out if someone won a race. Race organizers should be able to manage results on Zwift.com (or applicable Zwift app) and then DQ people as they see fit. Participants shouldn’t be required to use a 3rd party site that only a small portion of Zwift event participants actually appear to know about. For example, in the recent Tour de Zwift series, only 6,036 out of 24,758 people used ZwiftPower to track their results*. That’s a mere 24%.
(*The easiest number to look at, apples to apples, was riders who completed all 9 stages, which would most closely align with riders most familiar with Zwift. That would actually likely artificially inflate the % even more.)
Which isn’t to say that ZwiftPower doesn’t have a future. I absolutely think they do. I think they could become the VeloViewer of Zwift – doing incredible analytics and stat tracking. But Zwift needs to take governorship first of one of the most important parts of racing: The results.
Of course – there is one part of this that Zwift has been addressing as of late, and it’s definitely shown: The entertainment factor.
And quite frankly – this might actually be the most important one, just as it is for any sport. If a sport isn’t entertaining people won’t watch it. And if people don’t watch it, then sponsors won’t pay to support it. And if all that happens – the sport dies.
In the past, CVR used to run their World Cup events on Zwift, which were pretty well produced. Like anything, one could offer suggestions for improvements, but by and large they were executed quite well. Still, while the races were held within Zwift as a game, it was held totally outside of support from Zwift itself.
Last month though Zwift started stepping up their approach to Zwift as a watchable sport as part of the KISS Super League, which is a weekly race on Tuesday (for women), and Wednesday (for men). These events are held in various locations – last night’s race at the Canyon Headquarters – and are commented live by a team of individuals, including Matt Stephens, formerly from GCN. The events last an hour, and feature riders from lower level pro teams (not WorldTour, but mostly regional teams). These screenshots in this section are all from last night’s event (which you can watch here):
The broadcast incorporates standard elements found in most professional sporting events includes pre-canned video segments about the course/route for the day, as well as commentary about the planned route and the challenges that lie ahead:
As the race progresses the camera focuses on a given rider, and allows you to see the wattage (both watts and w/kg), heart rate and cadence of the rider. Plus elements like speed and distance. This view is for the most part no different than what you’d see within Zwift as a normal user.
In fact – in this respect, the metrics displayed here are generally more instant and complete than you’ll see in most TV productions of a typical WorldTour event, since many teams will restrict elements like power data to only a handful of riders (if any). There’s also a ‘Distance Remaining’ overlay that Zwift has added), as well as laps remaining.
Still, one of the elements that becomes immediately apparent once watching the broadcast is that Zwift seemingly hasn’t done much to cater to camera views used in-game. For example, typically in a pro cycling TV production there will be static and moving TV cameras throughout the course that focus on groups or the entire peloton coming into it.
And while Zwift has some of that via the helicopter view and some of the regular Zwift views where a specific rider passes a temporary camera along the side, we don’t see live cameras at finishing banners for example, or sprint banners – the primary sprints are typically shown in professional cycling. This is the view we get of going just under the sprint banner, which is following a rider rather than staying static at the banner like you’d typically see.
Further, unlike a typical pro cycling event, there’s no leaderboard showing the exact position/splits of the sprint. This is where I’d love to be able to see a direct camera on that sprint/finish-line in real-time and then instantly show the splits as the riders go across it.
Additionally, while Zwift does a good job at ensuring all of the teams have proper team kits in the race, the names shown on the side are as confusing AF. This is because Zwift doesn’t have any way to properly identify teams or groups within the rider list, so it rotates like a Times Square ticker board between the name of the rider and the team.
Of course – this issue isn’t limited to just the big name Zwift races, rather, it frustrates the heck out of everyone – and has for some time. Why we can’t add team names and/or logos after a riders name without cheesy hacks has been a mystery for years. Today, people append a team name in brackets to their name, which isn’t super clear or easy to follow.
Finally – I’d point out that while the stream states it’s in 1080p, it doesn’t appear that the stream quality (especially of the game side) is actually in a proper 1080p. Rather, it appears closer to 720p. While streaming can be challenging, at this level of production and financing it shouldn’t be (keep in mind that YouTube supports live streaming quality up to 4K/60). The reason quality matters here is that when it comes to picking out team rider jerseys, that can get difficult when things are more pixelized.
Those criticisms aside, the overall broadcast quality of these events have dramatically improved each week. If you watched the first events just a month ago – I’d encourage you to go take another swing at things these days. The live in-race commentary by Nathan is spot-on in terms of being both educational as well as on-point. That said, I think any live sporting event can benefit from a two-person commentator team, which often can allow a bit of banter back and forth that is tougher to do in a solo-commentator mode.
Additionally, it’s worth noting that unlike professional sporting events, you can actually support the riders in real-time using the Zwift Companion app (on your smartphone) to give ‘Ride-On’s (which are kinda like kudos) to the riders. So there’s elements here that don’t exist in the real world that are definitely unique to Zwift.
And again – overall the one-hour format that Zwift is doing today with KISS for live events is quite compelling as we’re now a few weeks into them finding their footing.
While the list of challenges might seem long, in reality, most of them aren’t actually all that complex. Instead, like tackling any home improvement project you just need to hunker down and knock them out one item at a time.
As noted earlier, Zwift seems to have slightly course corrected on how they’re categorizing racing in 2019. Keeping in mind that Zwift’s original Cycling Australia championship event on January 4th, 2019 awarded jersey/titles to two remote riders. Remote riders being people not at the race venue. The company has since shifted away from that method for the rest of the year. The official guidance from Zwift is now as follows:
“Covering (1) Racing Remotely and (2) Racing at Live Venues, we’re working towards two categories describing a governing body’s association with an esports event:
- Racing Remotely:
- Zwift’s position here is a National Governing Body should only “Endorse” these events in 2019, i.e. Federations endorse that Zwift’s cultivating a Zwift racing community with the aspiration of developing real world national championships/finals.
- These events can act as qualifiers to in real life National finals and will be subject to assessments by ZADA
- Racing at Live Venues:
- Zwift’s position here is a National Governing Body “Sanctions” events run by Zwift or produces these events by collaborating with Zwift.
- The upcoming British Cycling eRacing Championships is something you should cast your eye on. The rules and regulations are in keeping with high level competition and we will share this document once all parties have signed-off.
The aforementioned British Cycling eRacing Championship event will hold its first qualifier this Sunday morning – February 24th, though information and details on the championships beyond that is surprisingly scarce. Of course, like all these events, these are paid partnerships between Zwift and the governing body in question. The goal from the governing body’s side being to get younger people into racing by pulling in those with an interest in Zwift, while the goal from Zwift’s side being to legitimize Zwift as a valid platform to compete on professionally.
As we saw this year, Zwift has already started pulling in lower level pro cycling teams to compete on Zwift, with the eventual goal of getting not just individual pro riders – but full WorldTour level teams onboard as well to compete. The company isn’t there yet, but it won’t take too much more advancement on the event side, plus a scoop of that $120 million in cash to solidify higher end team and rider involvement.
Still, it’s good to see Zwift admit that the platform isn’t fully ready today in 2019. In fact, Zwift’s Steve Becket went further on this in a recent e-mail discussion to me:
“Esports from 2020: It’s mind boggling (and scary) how much work needs to be done to create an Esports platform from 2020 – which will orientate around a truly gamified racing experience.
So, creating KSL [KISS Super League] & KL was to meet these business demands. We aren’t launching a full fledged racing product – far from it – which is why we’re positioning KSL (and Zwift racing with pro athletes) around being a demonstration sport in 2019 to industry, media and consumer audiences in the cycling space, i.e. to help form a picture around what Esports could look like in future.”
And I completely agree with that position. Thus while it may seem like I’m negative on Zwift esports, I’m actually quite the opposite. I think there’s a fascinating opportunity here, not just from an entertainment standpoint – but also to shed new light within the sports technology realm. And I think last night’s women’s race was the first time to date where I felt like the production quality was to the point of being captivating and entertaining.
That said, I just think that there’s some work to be done before cycling federations start handing out national championship jerseys to riders in events that wouldn’t pass the same muster out on the real road.
With that, certainly interested in hearing your thoughts below – and thanks for reading!
Hey – I think you nailed it all in this post, SPOT on.
One thing I find interesting is your statement about Zwift Stepping into to fund ZADA. As someone with a bit of insider information, that wasn’t the case I was told. I think this post will give some light to the situation of CEVAZ stepping down. link to facebook.com
Correct, here’s the details on the Zwift funding. I didn’t include that exact quote in my piece above, but the specific quote is below:
/// from quotable Zwift e-mail as I was doing background on this piece – Feb 1st, 2019 ///
“• The community-run service disbanded because workloads became untenable. This (quite rightly) created greater scrutiny on Zwift Racing at a time when Zwift was promoting the launch of KSL.
• Zwift engaged with the folks behind ZADA and yesterday agreed to fund its re-instatement. The bottom line is Zwift felt duty bound to do so but needed a little time to get aligned on a plan of action.
• ZADA will cover KSL and most importantly the launch of KISS League where the participant pool is larger and less familiar.
• Beyond KSL and KL were are determining how best to scale this service to best scrutinize and legitimize the racing community.”
Great reporting! Very fascinating for sure and glad to hear that some verification is being completed, would be interesting to see more transparency as I am sure that would garner more support for eSports and the results being taken seriously. I’m not sure that all the members that were apart of the community-run ZADA were engaged with, so just some.
It is incredibly easy to write an ANT FE-C or BLE proxy that sits between the trainer and receiver (I do not use Zwift but actually created one for a different purpose). And that proxy can play with the power reported to receiver in very creative ways. This will render any “race certified” trainers meaningless unless they will start implementing proprietary protocols which they night as well do. In this case we are back to square on in market fragmentation.
Great analysis. Yes, there’s a wide gap between having a platform that somewhat allows racing, and generating compelling entertainment. The ZwiftPower/Zwift thing is indeed one big part of the gap; second is the absence of a live API to allow broadcasters to overlay race-oriented data (leaders, breakaway-pack distance, etc) and to select broadcast-friendly eyepoints (camera locations). Working with eSport streaming experts such as Twitch would be a good idea for Zwift, rather than trying to invent the wheel all by themselves.
And all that ignores the equipment/cheating part of the equation, which probably requires limiting serious races to on-premise participants only, where the mechanical, weight and identity issues can be managed (and even the traditional doping as well).
I’m a big fan of racing bikes, and I think Zwift is a great platform for allowing people who would never line up at a real start line to race (especially if you accept that the results will be affected by technological limitations and cheating); however, I don’t think I’d ever sit down and actually watch a Zwift race.
I have limited time to watch races, and given that there is always traditional racing available to watch on NBC Sports Gold, FloBike, YouTube and other sources, I’d much rather watch people racing bikes in the real world. There are so many more variables when racing on the road with people next to each other over terrain that constantly changes in varying weather conditions. I just can’t see how a Zwift race could come close to simulating that sort of an environment.
and what about height doping?
link to twitter.com
Interesting detail. I wonder how they modeled that, because when folded into proper road bike posture (most body parts close to horizontal, were extra length has little aerodynamic influence), a tall and lanky build might actually be more aerodynamic than a short and stocky rider of the same overall weight.
Chris Pritchard just did a Youtube vid on this topic. You gain a significant advantage by lowering your height in Zwift.
Zwiftpower doesn’t appear to check changes made to height in races either.
Being on Zwift for past few months i find the biggest thing that needs fixing is the constant texting of sexist comments. I can appreciate that its a reflection of larger issues, and pretty darn reflective of the standard mid night crit or Saturday group ride mentality. And I appreciate we cant’t fix stupid, but if there is one thing that makes me want to turn off Zwift and turn on Rouvy its this.
+1 to this. Most times someone posting a political view will get immediate feedback to stop. But sexist and/or sexual comments go on with no pushback. I kind of agree with the viewpoint “if you wouldn’t say that at your grandmother’s dinner table, why do think you should say it here”?
Is that perhaps a group ride thing? Or people riding outside of events?
I spend 10-12 hours a week on Zwift (since 2015), primarily warming up, racing and cooling down.
Can’t recall ever seeing a sexist comment. When in a race few people have the energy to type much and we don’t see comments from non-event people etc.
This was a really interesting comment and has made me get a bit ranty. Although sport has come on a long way in the past 11 or so years, there is still so much sexism floating around. From guys who just will not accept a woman might overtake them in the slow lane of the pool to the men who do a pointless sprint/overtake/slowdown on the hills on the bike, to the fact that in the UK at least women’s cross country races are still shorter than the mens (and let’s not even start with the TdF/Ride London etc). In the UK, women were not allowed to race over the same course as the men in the Boat Race (a huge sport/cultural event even to those with no interest in rowing) until 2015, and that was only because the CEO of the company that sponsored the race overall (a woman) called them out on it. That being said, it sometimes works in our favour as we had prime seats for the Women’s Cricket World Cup final for very little cash due to initially limited demand…
I could go on- the Ride100 with its 80-20 gender split. Kona. What Serena Williams has to put up with. Getting yelled at by men in cars when we run in shorts.
So call it out guys, on Zwift and elsewhere. Because women do and we often get told we have no sense of humour, or a police chief tells us not to run alone. And sometimes, ya know, I just want to ride and train and not deal with nonsense. And while most of the time I do just get on with it, if you stop and think about it properly, in the context of your comment, it’s pretty bloody depressing.
I will say this for Ray. He’s been great about calling out Kona etc. And it’s great that you all recognise this. But it’s part of a wider issue that we should all be talking about.
link to theguardian.com
link to theguardian.com
link to runnersworld.com
link to theguardian.com
This is an excellent analysis and very well written. Very insightful! You ‘da Man!
Being a Clydesdale rider I volunteer to for absolutely zero reward / recognition act as a tester or spoiler or hacker for everyone to verify if they can “catch” me in any event. I can produce stupid watts for sprints and with a little imaginative weight doping and kickr tweaking i’ll bet I can rile up some racer types…
I win because it gets me on the bike even more. They win because it helps them learn how to deal with the cheaters
Good article Ray. I used to race on Zwift, then I got a Hammer and thought I’d be having more fun on zwift and racing, but erg mode actually drove me to TR. Nonetheless, I have opinions of racing!
Re: cheating. I think one thing that eventually needs to happen to make racing more legit is to find a way from proactively keeping cheaters from racing. I’m not sure how, but I’m sure the means exist to develop a front-end gatekeeping. This post-hoc stuff with zwiftpower doesn’t change the fact that cheaters can dictate the pace of an event and influence decisive events.
Not that I want to know exactly how it’s done, but how on earth can someone miscalibrate a direct drive trainer?
Thanks for the article. The one thing that I wish Zwift would introduce would be some type of FTP normalization based on the altitude of the rider (See Joel Friel piece: link to joefrielsblog.com).
It seems like using rider location and a simple rule-of-thumb adjustment would help to level the playing field a bit for races.
Even though it sounds like everybody is accepting that venue racing will have a legitimacy advantage over remote racing, there is another minor issue with remote racing that I haven’t seen discussed very often – local elevation. It dawned on me during some random weeknight race that I was at a severe disadvantage against many of the other people in the race (some of whom I pretty handily beat in IRL racing) because I live at 6000ft and they live at sea level. Usually, for Zwift racing, I’m just looking for a good workout that doesn’t involve intervals – but I might have a hard time committing to racing or trying to take it at all seriously knowing that I was perpetually at a disadvantage of up to 0.5 w/kg. Everybody responds to elevation changes (up and down) differently, so it’s not as easy as a correction factor. I don’t have any solutions and realize it’s not the biggest issue, but something that I hope promoters and developers keep in mind if they do want to develop remote racing further.
but this is happening in current non zwift races and it generally accepted as a allowed. so i suspect it wont be different when it comes to esports cycling
I’m not sure I follow – what non-Zwift races are you referring to?
Great synopsis about the challenges. And while they look numerous, I think it is important to remember just how far Zwift has come! Back in my day [beats cane on the floor for emphasis] , waaaay back in 2014 I think, Zwift racing was really just a dream. We had no start pens, no Zwift Power, heck…we didn’t have any results or mini-map. And we were happy … okay maybe not so much. My point is that we’ve come a loooooong way in just a short time. It is my belief that Zwift will be up to these challenges.
To speak specifically to one of the challenges (Accuracy of the Smart Trainer), I would challenge Zwift and Zwiftpower for that manner to provide a list of trainers and powermeters that are “approved” for racing. I’d set the limits of the accuracy at 1-2% for everything (climbing, sprinting, accelerating, etc…). Then, racers can choose which one to use (much like they choose an UCI approved bike). If you want to run on a sponsored trainer, then go ahead, but your results don’t count unless you are on a Zwift Racing approved trainer.
To speak specifically to maybe another challenge: I think that Zwift is missing a massive opportunity to communicate to its audience while we are lined up in the pens. I mean, we are all just sitting there! Use that moment to tell us more about the race (give us the profile), some of the bigger stuff going on in Zwift, etc… Yeah it will be tricky to know which audience to gear it towards (beginner vs veteran), but I would do both.
Indeed, the idea of having a racing approved unit is something Zwift discussed for 2020, from Zwift:
“it is very likely that hardware restrictions for racing will tighten up in 2020”
That said, it sounds like some aspect of a list exists today already internally. Zwift reached out after the post with a bit more clarity there, but they’re getting some more details settled first with how they might plan to publish such a list (as you can imagine, that might set off a bit of a mess).
Zwift is, unfortunately, no different than the real world. It’s full of narcissistic virtue signalers claiming the high ground, busybodies and easily morally offended people that won’t leave us alone. Then you have the WKG Gestapo, people constantly b*tching about fliers like they are being rapped out of their subscription or something, allusion of cheating, and whiners with 1st World problems and nonsense BS in general.
Stop whining and just use the stupid service. If you are not happy, mind the door on your way out, and give your money to somebody else.
I think you missed the point of the article: This isn’t about people wanting to enjoy the service of Zwift – keep on keepin’ on.
It’s about Zwift partnering with national federations and awarding national championship jerseys (the same ones awarded to real-life races) without controls that match those.
I got grievance politics bingo with this post!
I understand, but on every single point you have mentioned, I see people (a very small, but annoying group) whining about it through Zwift, and running it for the rest of us.
On that magical garden of national federations and awards. If Zwift wants to go into that fine, that’s their business, but it just concerns yet another very small minority of Zwifters. It doesn’t do one thing for the rest of us, the experience we have with Zwift, or willingness to send them 15$/month instead of somebody else. I just see it as an annoyance and polluting the experience, so just put a ring-fence around it. Call it “ZWIFT Unicorns” or “ZWIFT Neutered” or something, put them unicorns in there, along with a ever growing book of compliance rules, regulations, approved material, expected behaviors, controls and so forth, and don’t create negative externalities for the rest of us.
If they don’t, I will end-up sending my money somewhere else.
Nobody at Zwift is saying they’re going to dork with anything for folks that just want to ride in circles around Zwift sans-race.
They are saying that if you want to race though, then yeah, there’s going to be some controls around that. I guess I don’t understand that there are people who want to race but don’t want any controls.
It’s actually a really interesting sub-topic: Zwift seem to be convinced that they improve their brand with racing, but too much focus on in-game competition could also ruin it. Once the dominant brand association is “it’s full of cheaters” there is no turning back.
I think that they once knew that, because until recently they stoically refused to get involved in the racing scene. This sudden change of mind has all the appearance of a hail-mary pivot to get that one last funding round.
Fantastic report and analysis of Zwift and the possible future of eSports. It is interesting that there has been a couple of comments about whether people would watch an erace when there are IRL races available. I personally watch quite a few live streams of Zwift races and there are a number of differences watch esports and IRL.
Some generic questions:
1. IRL, How many of the riders do you know personally?
2. Can you interact with IRL races? apart from cheering from the side line.
3. Do you have the finances to enter numerous IRL races?
4. Can you race at any time of the day in an IRL?
These are only a few questions for IRL but heres the thing about Zwift.
1. Zwift, How many riders do you know personally? generally loads.
2.You can interact in every race on Zwift in a number of ways.
a. Riding in the race
b. Just watch as you would IRL but see it all by following as a spectator or by viewing one of the live streams.
c. Listening to the riders chat on your own team Discord.
d. Give race update to your team mates even when not racing.
e. Give live feedback to team mates whilst racing, what’s happening at the fron or rear.
3. Finance, all you need is your monthly subscription.
4. You can race at any time of the day, any course, any duration, hard, easy.
You could go on and on with the pros of Zwift and yes we can bang on about the weight doping and all the other negative things about Zwift but everyone has to remember this is a game and only a tool to improve on your general fitness and well being.
It is also a great social platform and many teams IRL have been created, many friendships started and so much more.
In my opinion Zwift has a massive mountain to climb before it gets where it wants to be in esports regarding all the negative things that go on around the platform. Eventually for the general population to be completely interactive in any of the races and the fairest way possible all smart trainers will need to be calibrated in exactly the same way with the same tolerance of error, possibly even remote calibration from a central hub. Weight doping , I don’t think will be resolved with regards to Video weigh ins before a race. a tart would be to have all trainers fitted with body load cells with bluetooth capability so that every time a rider gets on the trainer there weight, fully clothed is monitored and sent to the Zwift API. (yes a rider could still take the pressure of by holding on to something but its a start)
“And I think last night’s women’s race was the first time to date where I felt like the production quality was to the point of being captivating and entertaining.”
It also helped that it was just a better race. That Canyon/SRAM rider in the live venue, burying herself — and failing — to close down a gap was the first really compelling thing I’ve seen so far in this initial iteration of pro-racing on Zwift. You got to see her face and the effort etched into it, in a way maybe we haven’t seen in televised bike racing since helmets became mandatory and everybody started to wear sunglasses. You got to see her face. Sports where you can see people’s faces will always have a leg up in terms of monetizing star power and gaining wide-scale acceptance.
Rider verification is one thing, but I think the same verification and openness requirements should be made for the platform itself. Zwift is a proprietary black-box game. We have to trust that they’ve modelled cycling physics accurately and fairly. We have no idea if they strive for accuracy, or intentionally tweak the physics to try to balance the playing field in an attempt to improve the gameplay experience. The fact that their default drafting model (until “double draft”) was not actually representative of the real world indicates that they probably do tweak things for balance.
As an example, we know rider height affects drag in Zwift. It feels to me that Zwift’s drag penalty for taller riders is disproportionately large, and playing with numbers in a few cycling physics calculators supports this. If that is the case, riders like Conor Dunne (2.04m) riding in the KSL are going to be unfairly penalised without anyone knowing. At least in a real race if a Commissaire makes a questionable decision it’s clear for all to see!
To me, it actually speakers to a larger issue I didn’t touch on in this post: Why should any given private company carry the torch for a platform in something (down the road) like the Olympics?
Virtually all Olympic sports today are based on sports that aren’t 100% tied to a given commercial product. Whereas when you go into the esports realm, or towards Zwift racing becoming an Olympic sport as Zwift hopes – that commercial product is the Olympic experiance – or, to a lesser degree as it was last month – the UCI sanctioned event experience.
I don’t have a solution for that – and frankly, I’m not sure there will ever be a real solution for that. But, I think it’s a discussion worth having eventually.
“To me, it actually speakers to a larger issue I didn’t touch on in this post: Why should any given private company carry the torch for a platform in something (down the road) like the Olympics?”
Knowing a few people that have actually participated in some of those, and being involved in the journey of a young hopeful, it’s quite a different experience and journey that being on something like Zwift.
Does Zwift believe it can attract more business by making people dream they can be just like Olympians? Does it believe it has a higher purpose?
Regarding the commercial aspects, besides being able to sustain their sporting activities, some actually want nothing to do with it and do not fancy making a living out of it. It’s 1st and foremost a passion, and about personal achievements.
Conflating esports and the Olympics is mindboggling at best. I hope Zwift and its investors are not falling into that trap, the crash in the wall of reality (and their wallets) will be hard.
To me, it seems like Zwift could function more as the venue provider for esports in the Olympics (or National Championships). So, for example, the Olympics or UCI or whoever, might contract with Zwift to provide the venue for their esports race for the current year. After that, it would be open to Zwift competitors who meet whatever standards are established. That way, the Olympics would not be tied 100% to a specific commercial product.
> So, for example, the Olympics or UCI or whoever, might contract with Zwift to provide the venue for their esports race
I think that you have this backwards: the drive for esports comes from the platform provider. Zwift desperately wants stationary trainer racing to exists, everybody else is just reluctantly tagging along because they don’t want to miss out when Zwift is throwing 120 million USD at the project.
This is a timely article Ray. Well done. I just saw this video from Chris Pritchard which adds another wrinkle to the integrity of Zwift racing- height.
link to youtu.be
Maybe one day remote stationary indoor racing will be ready for prime time. Just not today…
Great post, Ray! As someone who started racing – and started the first women’s only race series – in Zwift over two years ago, I can say that Zwift racing has grown exponentially in number of participants and teams, but little in technology. To date, the initiatives and tools created to support Zwift racing came from third parties and the community, as the examples you pointed out (ZADA and Zwift Power). The technology side is just as important as getting tons of racers, pro-teams, and publicity, so I hope Zwift will invest accordingly.
Zwift wants to be a sport, but I’d rather play a fun game while exercising. Somebody has to come up with a fun version of Zwift. (finding stuff, crashing into other bikes, biking off a cliff, etc.)
While only a few smart trainers (as mentioned) are reasonably accurate, Zwift can also use any Ant+ power meter.
Requiring either one of a small number of smart trainers OR any trainer with power coming from an actual power meter should be an acceptable solution as the best power meters should be more accurate than any smart trainers. It might be reasonable to have a list of power meters that are accurate enough.
“While the list of challenges might seem long, in reality, most of them aren’t actually all that complex. Instead, like tackling any home improvement project you just need to hunker down and knock them out one item at a time.”
So how’s that pizza oven list coming along? Sorry, I couldn’t resist! ?
One thing you don’t hear mentioned much in relation to Zwift racing is lag. Having played video games for a while, I can’t stress enough how important is to have a lag free game. I don’t use Zwift, but I can’t imagine a scenario where their servers are good enough to ensure there is no lag for all their riders worldwide. If I’m in America and another guy is in Australia, there is no way we are on an equal playing field if the host server is on the East Coast of America. Especially in bike racing where you can win or lose a sprint by fractions of a second, it would be so infuriating to race for an hour only to lose because one person lived closer to the server. Maybe I’m completely wrong and Zwift has figured out a way to eliminate lag.
That happened to me and bummed me out of Zwift racing.
I “won” a small potatoes Zwift race about 2-3 years ago -as in I came first in the sprint to the finish line.
Then I go to see the results in ZwiftPower and a) I didn’t win, by about a second. and b) the order of those results was different than those on-screen. I got pretty upset, but then realized how silly all this was. I still ride lots indoors, but mostly not on Zwift anymore because of these competitive-issues. My head doesn’t deal well with them, and so mostly watch a movie for long-interval training or Sufferfest for short, intense efforts.
Not a dig on Zwift – things are going to work out differently for different people! I got too irked at the 6 W/kg riders doing laps in 15 minutes and so on. I’ll get my ass kicked outside instead 🙂
Check out the ZwifTransparency group on FB, folks are volunteering to declare their real weight, in a tasteful manner. Their weight is also checked with a control weight of their choosing. It’s no ZADA but even that had many hitches and holes that “gamers” could find as well.
Crazy idea…. trainer mat with two weight sensors (one under the front wheel block, one under the trainer itself).
It is incredibly easy to write an ANT FE-C or BLE proxy that sits between the trainer and receiver (I do not use Zwift but actually created one for a different purpose). And that proxy can play with the power reported to receiver in very creative ways. This will render any “race certified” trainers meaningless unless they will start implementing proprietary protocols which they night as well do. In this case we are back to square on in market fragmentation.
The screenshot with the red arrow in it, where you’re pointing out the team names in brackets. The inset picture is of women, which means it was the women’s race. The large screen shows the riders as male. Maybe Zwift should work on that too.
Yeah, I think that may be due to three things:
A) The pixelization is crap, because the in-game upload feed is crap, so it’s hard to make out female defining features
B) The athletes themselves undoubtedly make their own avatar, including hair styles – so enough short cut hairstyles on a pixelized screen makes it tough
C) Just a weird angle at that exact moment.
I get the appeal of being able to race on Zwift – but will people want to watch it? I think that’s a tough ask, considering how few people watch real-world cycling races.
I’ve never used Zwift. Do they do something to simulate drafting? If not, why bother with teams? Isn’t it in that case just a simultaneous ITT?
Yes, they do have a drafting algorithm :O)
In a way, hasn’t virtual cheating already existed on Strava for some time now?
Strava really needs to do a better job with their algorithms, because it isn’t challenging to pick out eBikes. I see KOMs/QOMs somewhat regularly to challenge because Strava doesn’t filter the obvious out on its own. When a local climb KOM that belongs to Chris Horner pushing 500W gets bumped by another rider pushing 1000W at a relatively low HR, cadence, and average ride pace, something about that is just plain wrong :-(.
I’d just like to see certified solo efforts on strava. Why even bother when any flat or downhill segment is owned by a pack of draft animals or the Person with the KOM wheel sucked the whole trip conserving energy until they could jump free for the title.
Zwift needs to remove all algorithms for realism. No wind , no negative elevations, no height, nothing that could possible be changed for an advantage. . Just a certified weight that puts you into a weight class and you must submit runs on a qualifying course that are then averaged so that on race day you are racing against people with the same weight and wattage range. This could then qualify you for regionals etc but at some point all further races are public and live as a group with certified equipment. Which means the platform would then cater to track specialists as it just became all about watts over time. Hour record racing, velodrome like …so it’s now boring as hell and no one participates or watches.
So many questions………and issues… it’s mind bottling
the dull answer of how this was solved earlier was to use standard bikes like they use in Keirin races in Japan, but then where is all the fun.
I like this idea and have been thinking about using the blockchain technologies for this purpose, haven’t figured it out fully. but it will be great when it comes out
Indeed good point.
As you mentioned and demonstrated yourself, anomaly detection (as used in financial fraud detection or cyber security) based on statistical methods is something that could be done rather “easily”. It’s a question of money nowadays with machine learning. What is usually implemented however, is “quick and dirty” idiotic cut-off limits and controls not taking into account any context and/or historical data and frustrating everybody.
That being said, when obvious outliers like you have mentioned are not removed by Strava (for example), we have a long way to go.
Even a very, very soft implementation would be a great improvement in Strava: “This effort looks unrealistically high. Do you really want to include it in leader boards? [yes, I am exceptionally strong and/or willfully cheating] [no]”. Bonus points for suggesting auto-trim of track parts done in a car/train/ambulance.
I am convinced that the majority of flagging-worthy rides is not deliberate cheating but merely an accidental upload where someone did not override the convenient auto-upload after an e-bike trip or forgot to stop their Garmin when putting the MTB back on the car at the trailhead. The nice thing about this would be that false positives would not be much of a problem, people could actually enjoy a false positive: “woah, we were so fast today Strava asked me if I was riding an ebike!”
Or maybe this feature already exists and I just never saw it because my uploads are clean? (and the existing motorized KOM pollution is just the remainder of false negatives and willful “rank anyways” cheaters?)
Totally agree. Some of the false KOMs are so unbelievably false… Highway speeds!!
The replies, however, to your post, are plain wrong. Bike riding is about drafting. You don’t like that someone drafted till the last second? That’s on you. You let them do it. You need to be clever about how you play them. It’s why the strongest person never wins a bike race.
You don’t like that? Then racing a bicycle may not be for you. It’s a mind game.
PS #doubledraft doesn’t go far enough to the 30% gains you get from real world drafting. Bring on #tripledraft.
Exists. MPA from Xert.
One option no one has mentioned is the use of handicaps to normalize setups, similar to sailboat racing. Although no one is ever happy with the corrected results and some boats have a definite advantage in specific courses and conditions, it allows maximum participation and is still great fun and generally a pretty good indicator of skill. This removed the need to limit people’s hardware choice for remote racing… Maximizing participation.
For the serious racing, one design generally rules, where each boat is as close to the same as possible… Just like the Canyon race.
Best of all, it will introduce cyclists to the wonderful world of PHRF committee type politicking. Nothing beats hearing the whining of a boat owner when his boat is rated 3 seconds faster at the beginning of a new season. Ahhh, I miss it.
Excellent ! :-)))
Seriously though, I like the idea of normalizing everybody to a specific weight/height for some specific races. Why? even though for sure I don’t have the best FTP in the World, I like the TT races where all drafting is disabled because it’s you against the clock – no BS about drafting and wheel sucking !
It works for sailing because (best effort) normalization of boat differences puts skill into the spotlight. There isn’t much skill in grinding a trainer.
In other news, have you seen the proposal to include breakdancing and skateboarding, amongst other retarded things, in the 2024 games?
Why not competitive eating and drinking while at it.
I did see it.
Though, I think it was for exhibition events. Either way, things move forward. Some sports will work, others won’t. I’d personally rather see skateboarding than millionaires trotting their horses in private jets around the countryside to pickup pedals for their show cases.
Just me though.
Less is more, you can cull the horses, no need to replace it with breakdancing.
If you want breakdancing, let’s include tango or samba as well. 🙂
Great blog! I have watched all of the Kiss Super League Races and enjoy them. It’s going to be a real challenge for Zwift to move from recreation to sport. I love Zwift and recognize that it’s really just me riding with and against, what appears to be, someone else. That perception if easily blurred by ideas of cheating but not to the point of frustration. I am 92kg, I am never going to get to 65kg and be able to race in the A races. Probably won’t win any C races. But it’s a helluva lot of fun trying. I suspect Zwift will have to adopt weight and height brackets and set their algorithms accordingly. Maybe 35-50kg, 51-75kg, 76-95kg sort of thing. Perhaps develop software for the trainers that allows Zwift to monitor the power developed by the trainer to recognize human impulses and untampered equipment. I am sure they have some smart people trying to figure all of this out.
1) The commentary…: It’s pretty off putting for people used to Phil Liggett. Actually I think the guy they use speaks well (if a lot), but it’s not clear that he has really world racing experience (perhaps he does?) or has raced the courses himself. Real world experience is important as most of the rider’s are real world racers.
As you say, they are finding their feet etc, but this is arguably near the top of the list to get right. Perhaps bring Matt Stephens into the commentary box.
2) in-game cameras & replays. For the round 5 men’s race we basically we’re all in the dark about how it was won. Even the replays didn’t show what was happening in the pack. I have no idea how the winner created his his move.
Lemmie know any thoughts.
I actually think Nathan does a great job. My only feedback would be:
A) Widen the camera angle a touch
B) Add a second person
As I touched on in the article, I think Nathan is great at covering the Zwift-specific elements of the race, and manages to keep track of all the numerous things going on in the game. I have zero complaints there. However, I think the broadcast needs banter. It needs humor, and it needs people to play off one another. As you noted, pulling in Matt would be ideal, because he could related to aspects of real-world racing and over time he’d probably learn more about Zwift racing and find a good balance.
Totally agree on in-game replays. It’s a mess.
They really need a broadcaster type mode. I’d hope though that they don’t underthink it and only make it available to KISS type folks. Sure, beta with them, but I think such a mode where you could:
A) Define the route
B) Place static cameras on the route that you could switch to in real-time (such as sprint banners)
C) Be able to set groupings to focus on (breakaway, lead rider, etc…)
D) Be able to overlay and customize stats (perhaps even templates)
E) Be able to do dual-camera views down the road
All of that would be flippin’ awesome for the community, and I think it’d actually cement what Zwift wants to do more and more. Tie in stuff like the Elgato stream deck for pulling up camera angles and man, that’d be epic.
Maybe they’re listening. Maybe.
For me I think one of the best features that they could add would be to allow a viewer to choose their specific camera selection/multiple cameras on screen.
Like you said having better/wider camera angles would certainly be helpful, but if I could choose which rider’s POV I could watch that would be cool. Or if I wanted to watch the peloton chase down a breakaway while watching the breakaway to that specific riders POV.
I think that would add a lot to the viewer experience, since it is something that is lacking in world tour events to my knowledge. I’m not sure that they would be able to utilize youtube for those features, but a way to make future events better perhaps.
Yeah, once they provide a clean feed out of the game, you could route it via any of the streaming software systems (OBS/etc…) to any streaming platform (Twitch/Facebook/YouTube/etc…).
Of course, in a really cool ideal world they’d find a way to control multiple Apple TV boxes, thus allowing one to have 2-3 Apple TV units as a stack of cameras that one could record in real-time or leverage for playback of certain views. That’d be super cheap/ideal for semi-pro productions that wanted to capture multiple angles concurrently.
You nailed in your last comment – it depends on what Zwift want to do. Perhaps I naively thought that the real-world racer was a major target audience. The fitness group, the pure power beasts and those interested in the game aspect are also important to Zwift I’m sure.
I can really only speak as part of the first group.
What’s amazing about Zwift is that racing is surprisingly like the real thing (except for road placement). In most of my races I’m employing the same tactics as on the road: timing moves, faking fatigue, forcing the other guy to chase down a move, bolting when someone has an issue, even trying to convince the guy ahead to wait for me so we can help each other (yeah, I actually did this via the companion app – perhaps that was a bit creepy… and it didn’t work – link to youtu.be – 9mins). All of this translates amazingly well for a real-world biker – I’m serious, I think it’s pure gold dust – and if you don’t recognize that in doubledraft races then you:
1) Fail to fully connect with that first group;
2) Miss an entire narrative which promotes the sport (especially for those who just think the guy/girl who pedals hardest wins);
3) Doesn’t recognize the miracle that Zwift has managed to achieve in facilitating road tactics on the platform;
4) Reduces the experience to the stats: watts, w/kg, kg, and height (a major player in the speed algorithm we now find).
So yes I think Nathan is good for the gameification part, and he’s a good speaker, but the above is why I think it’s so critical to focus on commentary.
Reading your comments about the cameras I’m just aware how hard it is for the live editor. They also need to know if a move in the pack won’t amount to anything so we can keep focus on the break, for example. During real life events we count on them to know these things and it’s hard to calculate. You need to know as much as the racers. On top of that, there is more forgiveness over a 4 hour bike race with ample time to insert replays. Appreciate this one is really hard for Zwift, but if they miss the final move of a race where not much happened until the final sprint then that entire first group of watchers will wonder why they spent an hour on this.
Thought: could Zwift do picture-in-picture for breaks? Shouldn’t be too hard to apply an algorithm which 1) identifies a break (as opposed to a bunch strung out after a climb) and 2) puts them in their own picture when a small timegap threshold (2/3 seconds) is breached 3) when the break makes it stick and goes over an upper threshold gap (10 seconds), put the bunch in the picture and the break on the screen…. and variations….
I just did the Canadian Championships and there was quite a bit of chat about waiting. There was a group of 5 about 10 seconds ahead of our group of 10 that we asked to ease up to increase pack size. In general chat I saw a half dozen requests to join groups.
Interesting article. I think anything that gets more people interested in cycling, be it participant or spectator is a good thing. It’s got a long way to go before I’d take it seriously, but adding Phil Legget to those broadcasts would go a long way. LOL
My biggest question though is if they’re using tacx neo 2 trainers and Zwift, are they using separate cadence meters or broadcasting over ant +? Neo 2 and cadence over Bluetooth on Zwift is unreliable at best.
Keep up the great posts Ray!
I’m not sure a proprietary protocol is needed. Use BLE’s own authentication and encryption features I think could be used to prevent use of a proxy. Or at least make defeating it left open only to the hardcore hacker. I’m not sure that any of the trainer/app vendors actually use these features (yet), but they could.
I was hoping that Zwift would use some of the funding to make the many needed improvements in overall UX. For example:
– The most comprehensive user guide is a third party web site. The information on Zwift.com is barely thorough enough to get up and running. What do all the icons mean? How do you earn XP? What are the many configuration options? What do the various markings along the road mean? The only attempt to document those is from a third party, and it’s sparse.
– The idea that the only comprehensive racing results are from a (different) third party is ridiculous — as you note. If Zwift wants to be the sanctioning body, then step up and be the sanctioning body.
– The in-game UX is flaky and confusing. There needs to be much more customization available for the heads up displays. The map is tiny, the scrolling sidebars are hard to use, the choice of what’s large vs small should be flexible. The companion app is fiddly and very hard to use while riding. And if you’re going to make Ride-Ons some kind of coin of the realm, make them easier to give. I just found out about the “click my circle to spray” option yesterday, after six months of using the app.
– Invest in more courses. Or at least allow any course at any time. Boredom is their enemy. It’s the sworn enemy of everyone using an indoor trainer. They need to fight it.
– The Training plans are half-assed at best. As others have noted. This could be a real asset, they should seriously eat some crow and realize this is one place where third parties can help. Re-visit the relationships with TrainerRoad, or TrainingPeaks, or Sufferfest, or …
Finally, they need to remember that not everyone is there to race, or for “a game”. I’m there to sweat, to get better/faster. I don’t need a social experience or a race to do that. Just a great UX and some interesting courses.
That’s where I’d spend $120m.
Agree with your final comment. Some of my best Zwifting is done with the internet connection switched off after the course is loaded. Then there are no other riders, and you can just ride away all by yourself!
Smart software, with access to appropriate Zwift data, could identify patterns of rider activity such that rider legitimacy can be ascertained. For instance, a heavier rider would have the greatest difficulty in simulating light rider bike behaviour in finest detail. The same with sandbaggers’ power delivery. So such as weight, age, gender, and mechanical economies-of-truth can be unmasked. We already do this in the health sector, through blood tests, biopsies and other measurements, so huge amounts can be assessed about the patient without even observing them in person. Were appropriate developers allowed necessary data access, plus inducements to design and engineer such pattern identification algorithms, the challenge might become how to help those who were knowingly cheating so they can come to terms with what they have been doing and go straight. We’d not want to be brutal with such people, making witch-hunts or staining their character forever. And no, claiming an excuse that “It’s only a game” won’t wash. Future employers might want to know if a person have been deliberately cheating. This for a world where there is virtual bicycle racing in one area, and another area is for training, workouts, group riding, games, recreation, weight loss, injure recovery, social interaction, … everything that is not virtual bike racing. This, right across the board so for professionals, serious racers, older racers, weaker and beginner racers whose performance is far from high caliber or entertaining. If aspects of these analyses are broadcast during live events, the general public could participate in blowing whistles where devious activities become apparent. This approach might even be rolled out in other sports where performance can be appropriately instrumented and is measurable.
> Smart software, with access to appropriate Zwift data, could identify patterns of rider activity such that rider legitimacy can be ascertained. For instance, a heavier rider would have the greatest difficulty in simulating light rider bike behaviour in finest detail.
The primary data stream is only watts, cadence, heart rate. And all of them with rather murky fidelity. How would those three parameters differ between a heavy rider and a lighter, but equally strong rider? All the additional values (climb, speed, acceleration) that would allow one to guesstimate rider build from e.g. a Strava dataset are just derived from a subset of the primary data in Zwift (watts only). There is absolutely no way to extrapolate weight or height from watts. The only way for Zwift to avoid defeat from cheating is to focus on venue events, cheatproofing home remote racing is a battle that cannot be won. The closest thing they could attempt and still not suffer defeat is multi-venue events.
I may be an outlier but I view “spectator sports” as an oxymoron. I’d much rather be doing sports than watching it. Now some sports are more interesting to watch than others but watching cycling is barely a notch better than watching golf.
Want to see perineal KISS Super League podium placer Alex West cheat in his win round 5 on Wednesday? Go to Zwift official Youtube channel and see my comment (George Smiley) and click on the time hyperlinks in the comment to see the cheating. link to youtube.com
I don’t see the comment there. Maybe they deleted it?
Hey Ray, great article, and I think you have covered most things. However, on the topic of trainers, and what the riders who are remote are using, I think it is worth mentioning that a friend of mine is on team Ribble. For the Kiss series, Zwift sent him (and his teammates) a Kickr18 so that everyone is using products of a similar standard. Watching Cameron Jeffer’s YouTube Vlog (riding on a different team in the same series), it also appears that he has the Kickr18, so that would follow that they are being sent them. I was also told that the riders had been required to do a weigh in, which needed to be videoed to show it was actually them, and uploaded to a drop box. All of this aside, it is still relatively simple to cheat the system, but at least there are rudimentary steps being taken to try to ensure that in the very least, Zwift is watching.
> Zwift sent him (and his teammates) a Kickr18
Glad to see my 15/mo put to good use. Not.
Thanks. Indeed, Zwift came back with some clarification on this point as well. I updated the post with as follows:
(Minor update: Zwift has clarified that for the KISS Super League series specifically they’ve and Wahoo have sent all teams Wahoo KICKR 2018 trainers. However, some teams that already have existing sponsors like Canyon with Tacx and Wiggins with Elite. In those cases those teams will ride the sponsored trainers, whereas others should be riding the KICKR’s. Zwift argues this makes it more event, but I’d actually argue in this case they chose to send out the one trainer that’s easiest to tweak out of virtually every trainer on the market. It takes to tools or special equipment/software to do so.)
In other words, while it’s good there’s at least some standardization in theory, I’d also argue they selected the one trainer that’s frankly the worst when it comes to easily dorking with wattage number.
Note: Wahoo is the sponsor here, so it’s highly doubtful that Zwift paid anything.
I’m interested in doing races on Zwift but I don’t have a power meter or a trainer that has a power meter in it so my power data isn’t accurate. What’s the cheapest way to get real power data without building my own power meter? ?
There are lots of options out there, here are a couple:
– Second hand smart trainer: Tacx Vortex smart trainer on Gumtree 150GBP ish
– Left crank only powermeter:
4iiii (350GBP and up),
Stages (335GBP and up)
Avio PowerSense (199GBP and up)
are cheap and reliable AND work on your bike on the roads outside!!
Sorry to nickpick, but there is a reference to “Cycling Austria” in the “Going Forward” section. Australia and Austria are a very long way apart geographically :). I think you mean “Australia”.
Doh – good catch!
1. Trainers. Data is packeged in each zwift ride log, Zwift recognise which trainer is being used and ‘applies’ an automatic offset for a hidden calibration test that is part of the race. Similar to the offset applied to trainer difficulty of 1.0 if set to 100%. This ‘original’ log file is applied to the results.
2. Doctors certificate for height and weight.
If we’ve learned anything from the opioid problem, it’s that doctor’s prescriptions and/or certifications can be faked.
Brilliant article Ray. So many questions and so many hurdles and opinions.
Unfortunately I think this is going to be the death of Zwift and they can’t see it. A focus on racing / esports (and all its inherent complexities) is likely to appeal to investors who see big $ down the line from sponsorship etc.
Fair enough to be about entertainment but entertainment for the ultimate pursuit of profit/ROI is not appealing to me in the least. What’s Zwifts purpose that’s going to make a real and sustainable contribution to society? That can change our lives, our kids lives and contribute to making or world healthier and happier. How can they use their market share to do good and have fun along the way. I think their motto is something like ‘training is fun’ perhaps they should change it to ‘racing is profitable’. Love to be proven wrong and hope they don’t just focus on the small amount of people who want to race and think about their wider role.
I am very new to blogging and really enjoyed your blog post. Almost certainly I’m likely to bookmark your blog post. You surely come with fantastic posts. Thanks for sharing with us your webpage.
Is it just me or is one of the biggest challenges facing Zwift is that the racing is nowhere near what a real race is like? Last I checked, there was no accordion effect going around a corner, no need to brake or freewheel or exhibit any skills whatsoever. Every race I’ve seen or done is just a full-on effort from start to finish by everyone in the race… there’s no attacking, no lulls in the bunch, no teamwork to block or chase, no crashes. If you get gapped and have to chase back on, that’s probably the only realistic thing. But even then it’s not realistic because the bunch never slows down? So ultimately you can predict the results based on their w/kg or sprint power before the race even starts (unless people show up on a bad day). That’s not road racing, that’s mass TTing with a few sprints thrown in. Which is fine, if that’s what people want, but it’s not really representative of the real sport on the road at all.
As for the cheating, people will always find a way. But, to make it more difficult on the tech side they could only offer prizes/qualifications for those at an in-person event, using trainers that have been calibrated then “locked” with a special firmware (maybe using an in-house race system), and all rider metrics can be measured on site. Then it would mostly be an issue with drugs (or genetic doping, this is the future after all), and they would need to enforce strict rules like a death penalty to keep those out.
Good post. Problem with putting in firmware is that the ant+ or bt protocols are easy to spoof. So the protocols have to be secure and that means reduction in interoperability and massive increase in cost (and end user frustration).
But, agree that on-site is the way forward for now.
Yeah, need to forget about wireless and go wired. Wattbike and Concept2 performance monitors have ethernet built in for group training/racing scenarios. It probably wouldn’t be hard to make a race device to attach to a smart trainer and add an ethernet port, but it would be better to get the manufacturers to build it in.
Wired vs wireless doesn’t change the ability to hack the protocols. In fact, it arguably makes it easier.
CLWill, what I was suggesting is that you don’t use wireless at all. Each trainer wired to a central race system. That way it doesn’t matter if someone uses a wireless proxy because it isn’t being used at all for the race. That would be far more difficult to hack. Now, if they were using wireless to change settings or recalibrate the trainer that is a different story. Maybe trainers used for races would just need wireless disabled. This is similar to how Concept2 uses the Venue Race software with USB or RS485, or how Wattbike uses Power Cycling for group training with USB connections (not the disabling wireless part but in the wiring to a central system).
7 minute effort on a Kickr I managed 30-40 watts average MORE than my Neo…
Makes racing non Neo riders a little difficult with that much disadvantage!
That is a broad statement that is simply not true. My v3 kicker reads 5% below both of my quarqs. A neo 2 read 1-2% above my quarq. An older kickr v2 read 2% low. It all depends. A friends kickr reads 10% above his quarq. Lots of variability in the trainer market. It can depend on the software, update status, spindown, etc.
It is true if the trainer has not been calibrated properly… I can do the same with my old computrainer.. set the press on force to 2.5, turn the tension back 1/4 turn and ride with what is potentially a 2.0 press on force… according to the racermate manual every .01 pound off presson force the trainer is out equals 1/2 watt..doesnt sound like much, but in the miscalibrated trainer example I just gained 25 free watts… You can’t cheat the calibration on a Neo because there is none! I found both my SRMs and my P2Max were bang on the Neo, there is a very slight variance from the SRM to the computrainer, but I havent used my CT in 3 years since the Neo came out.. its now just a wall ornament in case the Neo fails until I can get a new one…
My 1st gen Kickr (calibrated) reads approx 50watts higher than my Neo , and the Neo is probably 10 watts higher than my drivo.
In races, that is a HUGE difference.
Still, I love Zwift racing. (I’ve done over 200 races) It motivates me to go hard, but I do wish just wish people would be honest.
I do racing as well from time-to-time on Zwift too, but it seems that regardless of what is being done, the whining and bitching never stops. People simply don’t like to lose, and if they do, it’s “obviously” because others are cheating, not because they have trained more and are stronger.
It’s the same type of whiners that are complaining about rich people, who obviously have all cheated and stole money from other people.
We live in a sick world, where apparently nobody has agency, and everybody else are cheaters and crooks.
As IRL, if people who raced against each other were not anonymous, and actually interacted personally, person-to-person, a lot of this would go away.
I think zwift should not allow anonymous profiles for racing.
I don’t mind losing, I’m 57 and tons of people are much stronger than me.
very, very good and deep article. I am absolutely no fan of Zwift whatsoever, but the whole topic made me curious.
But after reading the text and watching the race video I am asking myself… wtf?
All this huge amount of money, time, energy and what else to make this run… maybe itˋs the way we will compete against each other in the future – but today it seems to me still like some kind of nerdy computer game…
So many open questions, like you pointed out. I still think they (Zwift) should keep it at a reals entertainment level, just for the fun of it. Not some kind of serious racing on a “professional “ – level.
So many open questions and tasks still to resolve and answer…
The biggest question to me is: do the world really needs this? If people want to race – there are plenty races every week outside in the real world…
again – thanks for your great job!
I attended a Kiss event at the Madison Cycles HQ this week, loved it, a few beers, frites and waffles courtesy of Lazer Helmets and a Saki ceremony courtesy of Shimano, thoroughly enjoyed the whole event! This morning joined a Zwift ride for the first time with Amersham RC, loved the banter and would have been interested to make it to the last lap ‘race’ but dropped my phone and got dropped after I stopped to pick it up 🙂 Next time!
It would be trivial to turn something like a Raspberry Pi into an e-Bike booster. Was thinking of doing it for a good reason so that my spouse could ride with me on Zwift.
I have two suggestions
1. Since pedal and crank based power meters use strain gauges, I would think that power meters could be developed where before starting the ride, the cyclist stands with the cranks in the 9 and 3 o’clock position (putting stress on the pedals/crank arms). The power meter would then weigh the cyclist by measuring the strain. This doesn’t exist today but would seem to be in the realm of possibility for future power meters.
2. Fans want to be able to put a face to a name. Instead of just the basic computer avatar faces, for the KISS and similar series, the riders should be able to upload a photo of themselves which Zwift would then use to create a CGI style image of the face on the avatar. This would allow fans to get to know and support a cyclist from race to race.
Too easy to shift a significant fraction of your weight on the handlebars while doing the 3/9 weight measurement. But I like that kind of creative thinking. Yeah, many people *do* already have a weight scale built into their bikes, it’s just lacking software (and cheating controls). Might still be a very cool way to do the weighing in on-site events (provide some form of grip for stabilization where the gripping surface travels freely along the vertical axis, problem solved if there are witnesses).
Just don’t “hand out fame” in online events, make only venue events count. Online is a battle that cannot be won.
People would still cheat to hell and back in those online races to qualify for venue events (where they would only embarrass themselves if they cheated harder than their peers) which would hurt the quality of venue events. The solution, I’m afraid would be to make those online races not actual qualifiers but merely signals for a black box invitation process, where a combination of ML analyzation and human common sense would be used for vetting invitation candidates. Lots of drama and crying foul would be guaranteed (some actual cronyism would probably be unavoidable in the invitation process, human nature, and much more would be claimed by those who feel unfairly uninvited), but in the end that’s just free attention for the organization. Look at how football is prospering, in significant part because so much time is spent discussion referee discussions.
Remember the never-off-the-ground “CompuTrainer Racing League”? Yeah – they realized how easy it was to cheat, screw up calibration, alter body weight, etc.
But the REAL issue is FRONTAL SURFACE AREA. No one has adequately addressed displacement through the air, based on position, size, speed, etc. We all spend thousands on aero equipment, and while Zwift Eracing is “supposed” to be a great equalizer, it isn’t, it can’t be, and it never will achieve this as long as people of different sizes, weights, positions, posture, and with different equipment (albeit slight), cannot be accounted for.
The CompuTrainer chips were designed with the Frontal Surface Area of a 160lb, 5’11” male, on the hoods, wearing a 1985 or 86 Bell helmet like the ones you’d see in “American Flyers”. So a 5’2″ female has a drag handicap, and a 6’3″ male would be at a distinct advantage.
The only way to negate this is to keep speeds below maybe 15mph (closer to 12’d be better), and make them uphill. Then, you’ve got RRC issues, since lighter riders have to deal with higher RRC’s than they would EVER have to deal with when outdoors. This is on either a wheel-on or wheel-off trainer.
I’ve got 28,000 rider hours of data on 4 different types of trainers to back up this claim.
E-Cycling Sports isn’t equal, it isn’t viable, and it’s going to continue this arc towards popularity-then-disillusionment that the rest of the sport, and other E-sports, continues to suffer.
We may as well just host R/C Car races or go buy AFX slot car sets and get fat, old, and grumpy.
Ray, I hope you find this on-topic. What are your thoughts on e-sports and their encroachment on studio cycling? Peloton will surely impact the likes of Soul Cycle, but more interested in Zwift’s impact on the independent power-based cycle studios. Do people truly want to ride alone in a virtual world? Will indoor group training be a thing of the past?
I think it’s kinda like cars and SUV’s. People set on buying a car will buy a car, and people set on an SUV will get that. But people that are kinda undecided may go either way.
As such, while the two don’t directly always compete – they definitely do. For one of our friends that was exactly the case, they were trying to decide whether to get a KICKR+CLIMB combo or a Peloton bike (husband/wife). They ended up doing the Wahoo combo, but then a few months later also got a Peloton bike. Now from their Strava workouts they use both quite actively. The guy is a very strong cyclist/racer, and the gal a recovering competitive triathlete. 🙂
It’s not terribly unlike us I suppose. But, one can see my tidbits on that in my post from today.
Great post Ray, kudos for really hitting the nail on several of those subjects & issues. Great dialogue opened and much of the feedback here is perfect – people openly trying to find solutions to very specific issues. BUT. I think much clarity is needed, it’s only recently Zwift have shown much interest in Racing. The whole ZADA & Zwiftpower scenarios and recent history could become whole posts in their own right – it would certainly aid peoples understanding on how far along Zwift are (i.e. not as far along as you’d all expect and hope).
As a long time Zwift racer I don’t forget the time it took Zwift to open up their API (GDPR related) for ZP to finally get results posted again. We where racing for MONTHS without ‘official’ results. Lets not forget that Zwift until recently have done nothing to help group rides and leaders (for fear of paying subscribers feeling like they where ‘excluded’ and cancelling subscriptions, regardless to the obvious cheating they where exhibiting). Only recently have ride admins been given ‘powers’ to actually admin their rides – we’re talking the last 6 months maximum. Previously race administration had to all be done after the fact on Zwiftpower, to produce the ‘official’ results on there as nobody trusted the ingame stats of riders, or leaderboard displayed on Zwift.
Many of these issues raised have been thrashed out time and again by the various ride/race admin teams for several years, with very little input from Zwift. Zwift supplying them with zero support and seemingly zero interest. Now there’s money and Olympic funding mentioned/hinted at they’ve done the classic U-turn and are all over it. Reinventing wheels left, right and centre.
The Zwift community is huge, but until now Zwift has done little to cultivate that (to be fair they didn’t need to), but equally they’re not using it to their advantage. Get the long term race organisations involved. Zwift should call on their wide experience of all of these issues. At the end of the day Zwift are totally responsible for the huge cheating problems they are currently facing. As they’ve turned such a blind eye to it previously. They can’t expect to fix things overnight and to be taken so seriously when it’s been such a blight for such a long time.
With all due respect, Zwift encourages cheating and has made no attempt to solve the problem whatsoever, because they systematically want you to cheat.
Cheating solves a big problem for Zwift and they need you to continue cheating, at least for now.
The problem it solves, is what Zwift offers you for your $15 a month that makes it a value, and what is well worth $15 a month, is the belief that you are physically stronger than you really are, and part of your self worth and how you view yourself to be dependent upon this lie.
When it comes to a large corporation with a $120 million dollar budget, a lot of hand wringing and pretend research and we don’t know what to do can be dismissed as absurdity.
You understand UI, you know it is a design pattern. If weight is important in a game based on Watts/KG, and it is, then why doesn’t the UI even ask you for your weight? Because weights are static and never change?
It is a fact that weights are not static, it is a fact that they do change. It is a fact that it will impact your race, and also that Zwift does not ask you for your weight.
The focus on the hard core cheating is an absurd misdirection, when Zwift is still encouraging systematic cheating. Let’s deal with the people that want to be honest first, and how Zwift encourages you to cheat as step number one.
It doesn’t take 120 million in research or some AI of the future to tell if you have updated your weight in the last 24 hours or haven’t. If you haven’t your weight is wrong. And the way it is wrong, is more likely to be that you are heavier than too light. People tend to gain weight over time – that is known. When you know that someone hasn’t updated their weight, and weight is required for accuracy, you create a UI that simply prompts someone for their weight before the race. 5 second timer, default value their last weight. With the prompt most people will enter a recent weight. With it buried, it is forgotten.
Same goes for the second most common cheating – trainer calibration. The easiest way to boost your watts is to calibrate your trainer then dial back the tension – instant watt boost. This cheating is encouraged all the time by Zwift. Because again, they keep no record of calibration and make no mention of it. And yet, it is fully well known, that your tires (speaking of wheel on trainers), it is fully well known, expected and proven, that tires tend to lose air over time, rather than air themselves up over time. The tension decreases, the non-calibration works toward exaggerating watt output.
Zwift doesn’t even prompt the would-be honest racer to confirm they calibrated their trainer that day. Rouvy can calibrate your trainer in-app…Zwift’s cycleops calibration is forever broken, its some pretend mystery for them.
Lets dismiss any idea that Zwift really doesn’t understand how to program even basic functions in the smart trainer app world, ,and say they are purposely not fixing it. I cannot believe anything else.
Not to mention Zpower is a complete fiction. We know that, but the results still show up in the Zwift app with a placing. Only dismisse don some Zwiftpower results, but not in the main Zwift results….why? Because they want you to believe it counts, thats why.
hello people outside the world my name is Carolyn I want to say that my lover left me in April. During one long month I have contacted many casters and bought almost ten spells without results. My ex wouldn’t contact me, wouldn’t answer my calls and emails. But I never lost hope because I knew we were soul mates. Dr.ancient cast the most wonderful spell for me 2 weeks ago and everything has changed since then. We are close to each other again and he is calling me many times everyday. Thank you Dr.ancient of firstname.lastname@example.org your help is very appreciated! I will keep in
Thank you for giving an insight with respect to the challenges within e-cycling when the goal is to race virtually and win. Where I live (Denmark) they are trying to get as close to a “fair” competition when they are going to compete for the National title by letting all the different riders compete in the same room, using the same type of trainers and where everybody needs to step on a weight scale just before the event. No individual powermeters are allowed and everybody has to qualify before with completed rides where the W/kg for a given rider is examined with respect to being within a “reasonable” range. Hence the basis for the competition should be as close to “fair” as possible. But having used different types of both trainers and apps for quite a while it is my experience that 5% on one trainer may feel very different than 5% on another trainer, Also 5% using the same trainer on different apps can feel much different. Even 5% using one trainer and one app can feel very different if I chose course A in the app instead of course B. So my question is: Have anybody tried to examine how much variation in power we can expect to climb for instance 5% at the same pace for two riders with the same weight where the only difference is the exact model of the trainer used?
Here’s a moral question from my bike club:
“…look for someone in Watopia who’s been riding for ca. 12km in more than 80 minutes – and join them! Chances are you’ll join them at the top of the Alpe and can bank almost 250XP on the descent before putting in any real effort. You could always double check their location as spectator before joining.
The ethics of joining at the summit may be questionable, but ethics aside it would be easy XP.”
They compared the above to Cam Jeffers emulator case which you noted wasn’t a specific British Cycling offence more a general misconduct issue to strip him of his title:
“For the purposes of these Disciplinary Rules, ‘Misconduct’ means any conduct that is unsporting and/or has the potential to bring the sport of cycling, other Participants, Race Officials or British Cycling into disrepute.”
It’s an invalid comparison for one simple reason:
A) What Cam did was against the Zwift terms of service (in multiple ways)
B) What you describe above isn’t against the terms of service (at all).
Simple as that. One can debate the ‘morality’ of earning a handful of XP’s while descending, but it’s clearly not in the same ballpark as having someone else use the ANT+ simulator to *ascend* automatically on your behalf enough times to earn the Tron bike, and then use said bike to win at the time the most important Zwift race/event ever in front of a live national audience on real TV channels across an entire country.
Thanks I agree – was just interesting to hear people’s ideas. Thank you