GhostRider62

Not the offset, the linearity. If both power meters were miscalibrated in the "factory" with the same incorrect linearity, how could Zwift know that.

If the actual power is 200, 250, 300 watts but both meters read 280, 350, 420 watts due to an extra 40% factor. If the torque reading coming off both strain gages is incorrect, how could Zwift know that. Whether it is possible to change the linearity of the strain gage isn't something I have investigated, but I suspect it can be done on the Powertap hubs.

PeteHski

One thing that puzzles me is how they account for variations in power measurement between different power meters, trainers and smart bikes. Some are measuring power directly at the cranks, others directly at the hub and others indirectly via the motor torque etc. There are going to be inherent differences of up to 5 or 6% depending on where power is measured in the system. At a high level of competition, those differences are going to be critical. I had a quick look in the guide you linked and it doesn't seem to cover this. It just states that you can source power from the trainer or any power meter directly on the bike. Do riders simply choose the most flattering method of power measurement allowed?

Jughed

If you think about it - instead of team scouts needing to scan all of the youth and local racing data spread out all over the world, they can simply pull data from zwift and follow riders. Riders with high power to weight ratios, high sprint power, high threshold power - that may never be discovered, or considered too old... it's easy for a team to find the talent, put them on the road and see if they can ride/handle a bike...

That being said - some of the local Strava KOM's are completely borked.

We have one short climb, .21 miles with 2 very tight hairpin corners that make you slow down to 10 mph+/- and 3 climb sections of 6-10% - the top three times are 40+mph. Top 10 times are all over 32 MPH. Virtually impossible - even with a full on crotch rocket.

livedarklions

So Billy Mitchell is KOM?

BTinNYC

My favorite way to ride Zwift (for the non-interval Z2 rides) is to turn off ALL the data on the screen (Keyboard H key or the Companion app) plus remove my avatar from sight. No constant drafting encouragements, sprints or KoMs.

noimagination

Didn't the description of the product, "Zwift: Indoor Cycling & Running VIRTUALTraining App" (literally the first result when googling "zwift") give you a clue that it is not real? Your outrage is completely unreasonable.

himespau 

Which is why I only look at "My Resuts" when checking out Strava segments, especially on Zwift. As a low B, I'm not at the level where I'd be competing for KoMs from the top people on Strava even if all segments were real.

himespau 

At the highest level of Zwift racing, they are all required to use their Smart Trainer as the power source that is in the race, which would imply that they're all subject to the same amount of drive train losses. Whether or not that's true (and whether smart bikes with a belt drive have a similar drive train loss to a geared bike with a chain) is up for debate. In reality, well-funded, high end teams as of a couple of years ago were referring to it as the "smart trainer lottery" and buying multiple versions of the same identical trainer that met the 1.5% or 1.0% manufacturer-published accuracy threshhold required by ZADA, and looking for the one that gave them the highest power readings and selling the others on the used market for a slight loss. Buy 20 trainers at full price, sell 19 of them at a loss of $100 each as "open box" on eBay, and get one that gives you a 2-3% benefit at an overall cost to you of ~$2k plus the cost of the trainer. For some people, that was worth it.

Bald Paul

I had an issue where Zwift was estimating my power (dumb trainer, no power meter) and it suddenly went bonkers. As soon as the pedals moved, it was showing 300+ watts. I rode along for a bit, got some badges, and then "the notification" came. It was fun while it lasted.
Oh, never was able to fix it. Reinstalled and recalibrated the trainer, etc. The only way my numbers came back to reality was after buying a new smart trainer.

himespau 

When I used to ride with a powertap wheel as a power source on rollers, if the hub had fallen asleep mid ride (got off for a bio break shortly before a race for example), when I made my first pedal stroke to wake it back up after getting back on, it often would have a 1s spike in power of 1500+W, so I got all those power badges without trying. Just sort of ignored those as a result.

PeteHski

Regardless of potentially different physical drivetrain losses across different trainers, there's also the differing methods of power measurement and calibration used. For example I have an Elite trainer that measures power via an optical torque sensor and a Kickr Smart Bike that derives power from motor torque and voltage. I've also heard rumour that the Wahoo calibration may include some generic factor for drivetrain losses, but I couldn't find any confirmation of that. Not that it really matters for my use. But if I was racing competitively on Zwift I would probably choose something like a Stages SB20 with crank based power measurement over any of the trainers measuring power downstream of the drivetrain. I notice my peak sprint power measures 100W+ higher using 4iiii crank power vs my Elite OTS, even though average power is much closer. Basically the OTS acts like a smoothing filter on the power data.

livedarklions

It would be if the virtual product wasn't awarding titles based on what are supposed to be legitimate objective criteria, and advertising how well it simulates actual meat space riding.

You know that airline pilots train on VIRTUAL flight simulators--do you really want them drawing lessons from inaccurate models? I'm not claiming that the accuracy of the Zwift modeling is as critically important as that of flight simulators and other safety training programs, but just pointing out that the word "virtual" is not the gotcha point you obviously thought it was.

himespau 

Oh yeah, I'd forgotten that the SBP20 measures at the crank. I have several friends who are sprinters and who talk about the SB20 as a PEB (performance enhancing bike) because they get higher sprint power on it compared to their other bikes on smart trainers. At roughly threshold power, I don't really notice a significant difference (only done it a few times) between my assioma duo pedals and my elite direto xrt trainer. I haven't done a full ride dual recording, but if I just pedal and go into the pairing screen on Zwift, I have seen the pedals running maybe 5 W higher in the 340-350 W range (I'm ~102 kg, so that's about 3.4-3.5 W/kg), which is in the 1-2% range, but, again, that's not really scientifically measured.

spelger

funny, you are turning off all the stuff that most riders praise zwift for. you might want to look at Fulgaz or rouvy since teh scenery is real. been riding spain lately, rode teh most beautiful ride on one of the canary islands. amazing lighting from the sun.

noimagination

Perhaps you're right, but the purposes, uses and operation of virtual flight simulators differ quite significantly from Zwift, I would argue. Expecting things like KOMs, stats of other users (watts, I assume, don't know what other parameters might be tracked), "speed" (if this is like a multi-player video game where people can set up virtual "races"), and the like to be "accurate" when there is no way to detect "cheating" is unreasonable and shouldn't be expected.

Zwift should be viewed as a training tool, or as entertainment. It should absolutely not be used to compare oneself to others for any purpose other than purely for giggles. You can't take any "races", "KOMs", etc. seriously. That's just not reasonable.

PeteHski

If you have a well calibrated power meter, then Zwift is realistic enough. Imagine perfect roads with no headwind, the fastest bike money can buy and countless drafting opportunities. Zwift speed on climbs is very close to real life, but quite flattering on the flats.

The ridiculous speeds the OP complained of only happen with poor equipment calibration, software bugs and blatant cheaters/hackers. It's not really a big problem in a Zwift event. You might see the odd guy flash past cruising along at a sustained 10+W/kg and think yeah right, sure. But you naturally end up riding or competing with people very close to your own personal level - whether those people are genuinely at your level or not. It really doesn't matter if you don't personally know them. "Rider X" who is pushing you to your limit up a KOM could well be cheating, but who cares if it helps your own performance?

himespau 

I have a series of mass start races I like to do. Generally 40-80 people show up. With a few really fast folks, I'm never going to win, but I can push myself to try to stay in the front group as long as I can and then usually find a group of 3-8 other people to push myself against all the way to the finish. After a while, you get used to seeing the same names over and over and know who you can push yourself against, who you should beat, and who usually beats you. Doesn't matter to me so much if they're accurate performances, days when I can find a way to get past those guys that "should" beat me (because they usually do) are good days. Days when I'm dragging, I've got the "I can't let rider ____ beat me because I know I'm better than him" in the back of my head pushing me to latch back on to the group or chase him/her down. So much more motivating to me than words on a screen saying "spin faster" or "more power".

PeteHski

My Direto X reads about 5% lower average power than my 4iiii single-sided crank meter (in line with typical drivetrain losses). Initially I did some constant power comparisons at various power levels and found that 5% difference was quite consistent from 100-300W. Sprint peak power difference was however much higher. Like I said earlier, it was like the OTS was effectively smoothing the power data, which is what you might expect measuring downstream of the drivetrain vs crank measurement.

I think my Wahoo Kickr Bike reads slightly lower than my Elite trainer, but that's subjective because I can't compare them directly. I just find it that little bit harder to hold my usual cruising power on the Wahoo, although I've had a few PRs recently on it.

RChung

Depends on how sophisticated the cheater is, and how much effort he or she wants to put in. It would be very, very, difficult to detect this if you could alter both the on-bike and smart trainer torque sensors to match in exactly the same way. For most of us they don't care. At the upper levels of Zwift racing, they require dual-recording from two different power meter sources, and there's a list of approved devices. Plus, at the highest level, you have to show up in person, where they weigh you and have you race on their hardware. But this is too much effort for them to do when the stakes are low to medium. Most cheaters just look for a way to increase power, not torque. I caught a guy who tried to inflate the readings from an on-bike power meter. That was a weird and ultimately sad story. Some of the ways I caught him are public, some aren't.

That said, the PT hub had 8 strain gages, Depending on the model, modern crank spider PMs have between four and 12. That's a lot of strain gages to modify so that they're consistent, and then you have to do the same with the trainer. Possible, but a lot of effort.

livedarklions

Yes, but they put this prominently on their web page:

"COMPETE:
With races around the clock for all ability levels and lengths, there’s always an event to help you test your fitness while getting in an awesome workout. And you can compete solo or with a team!"

If it's supposed to be a motivating tool, I think letting people know cheating may be rampant is a good counter to it perhaps becoming discouraging when you can't actually measure up to cooked up results.
I don't think OP would disagree with your conclusion, but OP sparked the discussion that made yours and others' explanations relevant. This data doping might be old news to you and others, but this is good info that ought to be mentioned out loud from time to time.

rm -rf

Yes. I set it to around 25-35% on long climbs, and can sit and spin all the way up. With low watts, I'm still spinning instead of mashing slow cadences -- then my road speed can be down to 2 mph quite easily. Riders never fall over on Zwift.

Kickr trainer resistance is quite different from climbing in real life. It's continuous resistance all around the pedaling circle. Out on the road, there's way less resistance in parts of the pedal rotation, with momentum of the bike+rider carrying through that brief amount of time. On the trainer, these higher cadences are much easier on my knees.

The Kickr has improved my pedal stroke out on real roads, I think.

PeteHski

What I find in reality is that the vast majority of riders on Zwift actually do ride with realistic, believable power numbers, so it makes virtual racing meaningful and fun. Most riders on Zwift are now using smart trainers with reasonably accurate calibration and you can see from their avatar if they are riding on a dumb trainer (no lightning bolt icon). Whether or not some guy has subtracted 20 kg from his real weight to boost his W/kg doesn't matter to me. They are only fooling themselves really. I tend to be broadly as competitive in Zwift as I am in real life, so it can't be all that bad. Zwift Power is also quite a good resource for weeding out obvious cheaters from race results.

terrymorse 

Virtual is not real, I think everyone understands that part.

But with smart trainers controlled by software, there's no reason why virtual has to mean ridiculous. A minimal amount of coding could flag the ludicrous performances.

Yes, this is what I'm saying. They could at least make a minimum effort to flag the ridiculous speeds. Just look at the Strava leader board for the Hilly KOM Forward segment. Page after page of impossible 28+ mph speeds, all with sensible power numbers. A simple filter would flag those with "nobody can go that fast with that little power".

GhostRider62

WRT 8 or 12 strain gages, you would only have to modify 2 or 3 since they are in pairs of 4 to form a wheatstone bridge. Adding the appropriate resistance in series to each bridge should get the cheater's job done. I've never disassembled, so, this could be all rubbish but I do know there are wires from the PT electronics into the shell where the strain gages are mounted. Many modern smart trainers do not use strain gages. In any case, few have the inclination or knowledge to monkey around at that level. I'm probably wrong and just not willing to mess around with one of my PT hubs, they are too valuable.

spelger

the argument i have heard more than once is that if zwift were to flag those riders then those riders would stop using zwift and that means less $ for them. i don't know if this is what would pan out but i agree with you, it would not take much to week these folks out.