It begins - Virtual Power tomfoolery and Strava leaderboards.

(Mike Swart) #1

This guy’s powercurve broke the strava chart:

Here’s another one:

Both of the rides I’ve linked above have already been flagged on strava of obvious problems. In another thread here, Zwift employees have stated that they will have a system for flagging riders in game with wacky power data.

This activity will always be possible. For example, set up virtual power, disengage the resistance on the trainer and spin at 40 mph+ for new KOM’s. Maybe you change your weight in the profile to be 100 lbs instead of 150 lbs. I’m sure there are many other ways to trick the system. Greifing in MMO’s is always present.

Obviously the game is ‘for fun’ and we shouldn’t take it too seriously.My thoughts are around how this will be handled as Zwift becomes more ubiquitous and naturally more competitive.

When I sit down for an hour+ session on the trainer with a random selection of folks from around the world, I would like some assurances that everyone is being honest. Maybe that means having the ability to specify that I only want to ride with folks that have a ‘real’ power meter. Or maybe it means that we need the ability to create invitation only ‘races’. I’m not really sure what the answer is here.

(Peter Lin) #2

seems like the approach used by MMO of having private games is well understood, so that’s probably the solution that people are familiar with.

(Greg Gibson-Haymarket) #3

I don’t think the riders above to trying to game the system. I think these are just beta issues, particularly perhaps with the virtual power that needs to be refined.

There are definitely questions on how to maintain fairness in any real competition. There is a big thread on now about data inconsistency between Kickrs.

My own personal Kickr reports about 10-15w less power than my Quarq indicates. This seems reasonable to me in that the Kickr is measuring power at the hub versus the Quarq measuring power at the cranks. A 10-15w loss of power through the drivetrain seems appropriate to me.

On the other hand, I have a friend who’s Kickr reports 13-16 PERCENT higher than either SRM or Quarq data, even after proper calibration. A 383w 5 min effort using a Quarq was reported at 447w by the Kickr. I’d love to have ridden the 40k challenge on that Kickr!

Do you consider someone who has a “factory juiced” Kickr to be lucky? Perhaps in a Zwift competition maybe, but it’s not going to help you in the real world. It’s one thing to knowingly cheat the game, but what about someone who’s equipment gives them an advantage they are unaware of by simply reporting slightly inflated data?

We all want a level playing field and definitely now there are some questions about how that will be accomplished. I think in the short term we all need to keep in mind that this is just for fun and there is no real competition. I have full faith the Zwift guys will come up with solutions when the time comes…and if not, I’ll enjoy using Zwift for the social aspect as well as a great way to make indoor training more tolerable and efficient…which is the main reason I am here anyway.


(N. Jekov) #4

Thinking the same. I am not sure whether I can keep up my motivation for Swift long-term, if leaderboards of segments are all taken by people with virtual power trainers and superhuman efforts.
Perhaps can we have separate leaderboards for people with real powermeters, if only for the reason of comparing pure wattage output.
Also liked the whole system a bit more when everyone was 75 kg, since a lot of folks will skew weight/results just for bragging rights.
As an idea combine the riders names on the screen with a symbol/number that indicates what power measurement is being used. Also to see the riders weight would be a nice addition.

(Greg Gibson-Haymarket) #5

My motivation for using Zwift has nothing to do with placement on leaderboards…that’s just icing on the cake as they say.

(Chuck Kozlowski ZSUNR) #6

I really like the idea of knowing who has a real power meter vs. virtual power (I guess this is derived from speed). I use a Stages power meter on my current bike and I really like it. As far as people gaming the game - unfortunately - I think that is bound to happen. They need to be banned from Zwiftland and forced to ride outdoors during the winter.

(Peter Lin) #7

I’m not super competitive, so cheating isn’t something I’m concerned about. I do trainer so that I’m hot hurting like a dog in the spring. I don’t waste energy preventing cheaters, I just focus on having fun and getting a good workout. Having said that, discouraging cheating isn’t a bad thing and can only improve the experience for users.

(Casey Schumm [X] ) #8

I can confirm that zwift has at least attempted to identify bad data as it happens. Just to test the virtual power yesterday I set up virtual power as opposed to my real power meter. I spun the bike up to 20mph by hand (tire not touching the trainer). Apparently the acceleration was too fast and zwift told me to stop pedaling and tighten my wheel on the trainer. The message said to not pedal until the message went away and that your power was not being registered. I had to wait until the back wheel almost stopped before zwift would remove the message. Then if I spun up slowly it would work fine. I was just testing for a minute or 2 and didn’t upload to strava.

I know people will manipulate it but at least zwift is attempting to flag obvious bad data. Some other software would just allow the super human efforts to continue.

I would like to see a way some way to know what tool each rider is using - Powermeter, smart trainer, Speed/cadence. On another racing software I use you can see what tool other riders use in the results of a race. When you see people having odd bursts and such in the game you usually find out after the fact they are running speed/cadence. I have learned to accept it and not get too bothered but obvious bad calibrations or blatant manipulation of the tools. If that is what it takes to beat me I hope I see you on the road some day.

(Mike Swart) #9

You can already see who has Virtual power and who does not in game. It is indicated by a lighting bolt next to their name in the list.

(Clint Westhoff) #10

I have seen in another program (Tour De Giro) the impact of bad virtual power. But I would say the really bad problems are not hard to identify and there will be ways to work on those problems.

But keep in mind that Virtual Power, used and set up correctly, can be just as accurate and just as reliable as a power meter. What you see in post race comments on Tour De Giro that is more problematic is the accusations when the VP is high, but not impossible. Strong riders using VP seem to be accused of cheating just because from the perspective of some other rider their power is “superhuman” when it is really just “real good.” What I have seen in the Tour De Giro cheating allegations, post race, is often a third party will be able to verify that the alleged “cheater” was not cheating.

So for those cases I think the best solution is to get over it. Who is to say the person isn’t really that strong? There are some really strong riders out there so as long as the power profile isn’t actually superhuman, aren’t you just riding with a stronger rider which actually happens in the real world? I have a power meter, but if I didn’t and my VP was 10 watts to high isn’t that the same as riding against me when my FTP was 10 watts higher a year ago? And keep in mind that just because someone’s power and speed looks superhuman to you, that does not mean that it is. Look at Tour De Giro race results - there are plenty on there that look superhuman to me, but their power is from a power meter.

(Vincent Beggs) #11

As was mentioned in an earlier post, the Zwift developers will be adding a way to flag riders in game and of course there is already a way to flag rides in Strava. It will be interesting to know how the in-game flagging will be handled. Will it only take one rider to flag another rider (which could cause it’s own set of problems) or multiple? How many? I would suggest more than one and perhaps just two would do. Will they be removed from just that session’s leader boards or eliminated from all future boards until they appeal? I would start with affecting just that current session but perhaps Zwift would be able to keep track of flags over sessions and if a rider is flagged more than X number of rides, they are kept off the boards automatically until someone can evaluate the rides/rider.

That should handle the egregious cheats but as far as riders that adjust their weight or mis-calibrate their power meters for an unfair advantage, it’s going to happen but isn’t actually going to have much more impact to my enjoyment of the ride than the fact that even with everyone doing their best to play fair, the inherent inaccuracy of the various v-power from dumb trainers, smart trainers and power meters will mean some folks will just have some level of advantage/disadvantage versus others. It’s admirable to do as much as possible to make that discrepancy as small as possible and it appears the Zwift folks are and see it as important to make the experience as enjoyable and engaging as possible.

As long as I have a reasonably fair shot at competing with Greg and am summarily left in the dust by Jens Voigt, its all good with me!

(W ish I was Outside) #12

I’ve been logging the power on a Garmin+Powertap simultaneously with my Zwift/Computrainer combination and when calibrated the two are very close overall on average power, with the Zwift/Computrainer reporting very slightly lower. The one consistent big difference is that the Powertap reads 50-100W lower on efforts above 750W (generally more in the 50-70W range but I have seen discrepancies as high as 100W). Once power is below 500W they are very close to each other, within a percent or two with the Zwift/Computrainer exhibiting some lag that affects the game but in the end the power and energy numbers look pretty much the same. So the numbers can be fairly close to reality when everything is configured properly. My actual average watts today were the same within 0.3% i.e. 0.003 on the two systems, with the Strava weighted average being 5W lower on the Zwift version vs Powertap because the peak efforts were being recorded lower.

Anyway, to me the point is not about a level playing field for competition, it’s about motivating me to put in a good workout. Zwift is pretty effective for that but the glaring superhuman efforts are the ones that are somewhat discouraging, like the 2000km/h AI sprints. Anything within reasonable human range is fair game as long as it doesn’t fundamentally alter the ability of the simulation to motivate me. Because there are pro riders hitting Zwift that have FTPs in the 450 range vs my 300, when I encounter them on the circuit it will be immediately apparent that they are in another league and I won’t be able to keep up even in their draft. Whether it’s a real pro or someone that is cheating is immaterial, I’m going to encounter riders in the virtual world that are that much stronger. It might be interesting, eventually to support virtual world events that would be restricted to registered event participants, where the event organizer could declare it to be a handicapped or non handicapped event based on a user’s declared or past recorded power levels, this would put the pros and us rank amateurs on a level playing field. (And I am particularly rank.)

(D THANG) #13

Good topic, I’ve been curious about how this will be handled. Even on setups with Quarq or SRM there are going to be instances where the offset is different, even on identical hardware configs. Quarq has omnical and auto zero which is great. I use the factory slope, and zero the trainer before a ride. In real world I end up zeroing at stop lights while slowing or track standing. Still, offset can accumulate after a hard effort and show different watts for the same real efforts. And proper slope can change depending on chain rings used, etc.

I think this only really matters where there is real competition with prizes at stake. In those instances it might help to have hardware requirements, or groups - “kicker only race”, etc. It would be really cool if Zwift could send some sort of instruction to a smart trainer to calibrate and zero it prior to an event. I suppose that still leaves body weight as an unknown. When I train I take photos of my scale about once a month and it helps me keep track of watt/kg instead of just an ftp with out considering body weight changes. I suppose even scale pics could be faked though. One final thought is having competitions at trade shows, bike shops, etc, where one representative in control could make sure everyone is playing fair, just like at a real race event.

(Andrew Williams) #14

I’m not sure what the answer is either. I saw a virtual power rider take all three jerseys today and then claim the top spot on the strava “polka dot jersey kom”. I think it’s still up.

I looked at his other rides and although he normally rides outdoors with a power meter and has predictable HR responses to predictable high-level-amateur power numbers, this indoor ride was off the charts.

How do I feel about it? I don’t know. I don’t like the idea of flagging the ride because ultimately I am using Zwift for my own training, but I am torn because I felt like the jersey competition just wasn’t a motivator any more.

On Tour de Giro, I can pretty much tell when someone is riding way beyond normal capabilities using a speed/cadence sensor, and I remember who those riders are and pretend like they aren’t there. This is different because those jersey awards are there every lap, again and again. There is no actual “race” to the finish. Perhaps this won’t be so much of an issue once Zwift moves on to long road races.

(Eric C. (Zwift HQ)) #15

Hey everyone,

This is definitely something we’re aware of and we’re working on. Right now, making sure ride data is accurate for our zPower trainers is paramount. This doesn’t not mean we’re not actively looking into ways to police inaccurate data, especially as this delays our efforts in working on Zwift.

There is a whole tangled nest of issues and every time we think of a simple one, tons of varying scenarios show that it might cause more harm than good.

We just thank everyone for their patience as we work towards the best possible solution.

(Onno Pierik) #16

Hi Andrew, I saw you ride last week. I have a Kinetic Rock n Roll, so my rides are base on zPower. I did some power sprints last week. I got a jersey for Col d’Zwift. Did you talk about my effort or maybe another rider?

In my case, I user a Powercal when riding outdoors (which is a useless power meter, it is not a power meter)

(Andrew Williams) #17

Onno: No I don’t remember your name from that day.

(Thomas Fabian) #18

i’m not gonna lie…i disengaged my trainer resistance the first go around the island because i was working out issues with my ant dongle being wonky…and it just made it easier to hand pedal without the resistance being engaged. I feel like heart rate data could add a level of authenticity without the costs of power meters or fancy trainers. The problem is to be totally accurate you’d need everyone on the same trainer and same meters all CALIBRATED and that would just be silly. In the world of video games, response rates in mice and keyboards, refresh rates on monitors, video cards and even headphone quality all have an effect in creating that minute advantage, so I don’t see it as something that is overly concerning. Obvious cheats will always be obvious, HR data would provide the evidence to challenge authenticity. Someone putting out 1,000 watts and only pumping out 100 bpm would obviously be cheating.

(D THANG) #19

There were some good ones over the weekend. Some guys were able to hold 1200W for over a minute up the Col to get kom status. Not sure Tony Martin could even do that. I see these same guys on strava (it’s easy to see when you click the seg - view full leaderboard) accepting plenty of kudos and congratualtions with no flinching. I would never flag a ride, but it’s just odd people try to pass it off as real.

(Andrew Williams) #20

It’s gotten to the point where none of the Strava segments really mean anything now. To go in and individually flag each ride that is clearly superhuman would just be too much effort.

7+ w/kg laps, 36 (!) mph KOM runs, ordinary amateurs crushing world class pros. etc.