Get rid of the bad data!

(Dan Wilko) #1

This has been mentioned in various others ways but I have to reiterate it - there needs to be a distinction between power meter and non-power meter users. Whereas the data provided by power meter users is relatively reliable and valid, the data from virtual power users is at times farcical. Today there was a someone pushing 14 watts per kilo for 2 minutes apparently?! This bad data is contaminating what is otherwise an excellent product. Don’t let it happen Zwift! 

(Seb Barrett (U19) AUS) #2

very demoralizing i agree - even if we could just make anyone of our choosing invisible would solve a lot of peoples angst…

(David Sack 50+) #3

I normally don’t get bothered by it but last night I would have had a chance at a green jersey which has only happened twice since I have been riding except for a flyer that was a good 5-6 seconds clear of everyone else on the island.  This person was smart enough not to trip the alarms and receive the Zwift dunce cap.


(Myles Willis RGR) #4

Reporting users might be the answer to this. 

(John Murphy) #5


And I’d like to address the counter-argument that power meter users can cheat too:

 It’s extremely difficult to alter your power meter to generate higher readings.  Calibrate it on the moon?

 Lack of incentive. You’ve invested in this rather expensive training tool primarily to make to make you faster and compare to your competition.  All this goes out the window if cheating.

(Mike Blaszczak) #6

It’s easy for power-meter users to cheat. For example, I entered an honest body weight when I set up my Zwift account. If I entered a lower weight, my watts-to-kilos ratio would go up and I’d climb a lot faster for the same power input.


I like the “Ignore” feature that John suggests. Zwift could be a cool online multi-player game, but since making friends is impossible, it’s really just a single-player game. The other live users might as well be ghost riders.  In adding friends, groups, or teams, an “ignore” feature will be completely natural.  If a user is ignored by many users, it can be inferred they’re misbehaving either socially or with bogus stats.


(Dan Wilko) #7

I didn’t make reference to cheating in my original post - I was talking about the validity of data. The problem with virtual power is that there is too much potential for ‘noise’ in the data making it unreliable - far more so than power meter data. In fact I would think that many of the virtual power users who are producing some dodgy times are not doing it intentionally: it is the result of an unreliable measuring system. I think one way of dealing with it would be for Zwift to query suspicious data and ask the users to double check their set-up/calibration perhaps even using data from Strava to validate it.

(Seb Barrett (U19) AUS) #8

I’d be interested to know the parameters for causing the flyer warning sign to pop-up.  to my uninformed eye it does seem to occur at the extreme end of absurdity - like 13 w/kg sustained for 30 sec or similar.

i would have thought some riders doing 7 w/kg for long periods ( > 2mins ) while not impossible is at the elite level.  i’m not saying they should be flagged.  however i’m guessing a lot of innocent /accidentally mis calibration is occurring amongst this group.

would it be feasible to say compare their performance with a professional rider and gently/diplomatically suggest unless they themselves ride at that top end of the sport, they are simply in a dream world? 

in the end the “Ignore Button” might be the only true way to keep everyone happy.

For example:

Filters to Power Only Riders

Further filter out specific riders.






(Mike Blaszczak) #9

Cheaters are a subset of bad data. I thought you wanted to get rid of bad data?

(Dan Wilko) #10

Yes Mike you’re quite right - cheaters are included in the ‘bad data’ set and this could of course include power meter users. However, I think Zwift’s ultimate aim should simply be to iron out the bad data and not get too bogged down with whether people are intending to provide bad data (i.e. cheat) as frankly that would be a fruitless task.

Admittedly if there was any leap of logic in my original post it was that ‘non-power meter = bad data’. I’m sorry, but as any regular user of Zwift will testify, the fliers - those whose times seem to be off the scale - are almost always exclusively non-power metered users. This is not a coincidence. As I say, intention - whether or not they are purposefully doing this is irrelevant - the data is just plain bad and needs dealing with.

Therefore to take your point, yes, if and when power meter users start producing equally suspicious data and times (which would probably have to include deliberate cheating as you suggest) then absolutely; it should be queried and dealt with. But for the time being at least, it would seem that the significant majority of bad data is coming from the non-power meter users.

(Marcus Durant) #11

To add to to that, I’m really not sure why someone would invest hard earned cash buying a power meter to simply cheat in a game, when they can do it for free with zpower. It simply doesn’t make sense.

Ergo, the majority of cheaters will be on Zpower

If the only advantage is to make yourself appear like you have a power meter, then this seems to be achievable via Trainerroad (don’t ask me how but apparently it is)

(Jeremy Brazeal AETNA R/T (A)) #12

Just to clarify entering a lower weight does not equal just faster climbing. It allows the rider to go faster on lower watts. speed = w/KG in ZWIFT. This does make the same riders a lot slower on the downhills which is one area where ZWIFT is not a good representation of the physical world. Believe me I am under 140 LBS and I can loose as much as 40 seconds on the reverse decent if I am not drafting off heavier riders. This never happens in a real race. I am usually one of the faster riders on decents. All ZWIFT needs to do is add the actual watts next to the w/KG in the feed. If a riders is doing 7 w/KG but only putting out 200 watts vs 400+ for an average size rider that would signify the weight is off. Then other riders can let them know in the game if they have not been flagged yet.