Many of the leaderboards for segments in urban areas (especially those next to highways) are riddled with blatant vehicle travel "rides". In nearly every case, a rider neglects to stop recording their ride when they get in the car to drive home. The result, in the case of most segments following major roadways, is the top 30 leaderboard spots boasting outrageously inhuman times--i.e. 72 mph on a flat road for 10 minutes, etc.
I suggest that a flagging algorithm should be setup in the Strava software that automatically removes certain rides from the leaderboard under strict parameters. For example, on a flat/uphill road, it is physically impossible to ride a bicycle, electric or otherwise, faster than 70 mph, for any length of time, barring outrageously extreme circumstances. To prevent vehicle travel from clogging up an upsetting proportion of urban segment leaderboards, Strava could auto-flag those rides which exceed human-effort plausibility, such as the parameter described above. If auto-flagged, the rider should have the opportunity to appeal if his/her ride was legitimate.
If a feature like this already exists in the Strava software, then I argue it is far from sufficiently strict. I have witnessed leaderboard clogging in nearly every popular segment in the Dallas area, and it is impractical to manually flag each one. Seriously--I've tried. After flagging ten or so blatant vehicle travel rides, my ability to flag is temporarily removed. I emphasize that I use the flagging feature judiciously, yet Strava eventually deems its use inappropriate no matter how ridiculous the ride in question is.
In summary, please consider the following suggestions:
1) instate an auto-flagging algorithm that removes rides from the leaderboard which obviously exceed the realm of human-on-bicycle ability
2) if an auto-flagging feature already exists, strengthen its parameters significantly
3) if the above are impossible, then allow human users to flag more than ten rides before the feature is temporarily revoked (yes, I will painstakingly flag every blatant vehicle travel ride in Dallas if given the opportunity)
I am a runner but we are facing the same problems like riders. You would think there is a simple algorithm to tell the difference between running and car driving / GPS error but there are segments full with people finishing 300-400 meter runs in 2-3 seconds... This week I flagged a "running" activity which was actually skiing in the Alps in January and also included riding the ski lift. The same guy also recorded a 30km run which was actually cycling.
In previous comments there were suggestions to compare speed and heart rate data, but Strava obviously lacks even the basics, e.g. flag everything which is 20% faster than WR.
One strange thing I found today: someone has been notified via comment that the activity included 10 km running + 20 km car driving, he cropped the activity but still holds 7-8 segment CRs.
Look at this amazing segment. This is not even done by a car. It´s physically impossible. Unless... It´s on the highway that is above the tunnel and even that is completely insane 😄 https://www.strava.com/segments/18964978 I don´t even know what the actual time should be. It´s not 1:08 either, because I stopped during that ´try´. I also really love this one: https://www.strava.com/segments/9234842 Which is a small street with bumps on the road. It´s impossible to drive 75km/h . Ah well. I myself had the app on accidentally while I was on a train. No segments KOM obviously yet, my average speed was absurd high and the same counts for the max speed. 140km/h Yeah. No thanks. If I do that in real life my wife would probably kill me
@VincentKleijn its all in the maths and the variables used, the speed is calculated on the Segment distance rather than the actual distance covered along the segment. Strava matches your GPS data to the Segment nearest gps points to the start & end of segment and a couple other points (adhering to route).Lets look at pujadeta segment 83.2m at 4 secs(Maria KOM) the speed formula is (83.2/4)*3.6 = 74.9 kmh, Maria used an android app to record her activity, lets say she was moving at 39 kmh, its not recording every second(secs is smallest time measure), lets say her nearest point along the segment was 50m past the start and 10m past the end, the overall distance is 43.2m (83.2=-50-10) covered along segment path, the speed formula is (43.2/4)*3.6 = 38.88 kmh which sounds more realistic, when I looked at the activity its bad gps data from the map of the activity.
below image actual segment points, blue stars start & end points matched.
These aren't useful examples of segments with vehicle rides because the times mostly are that short due to gps inaccuracies. Such short segments aren't allowed to be created anymore for exactly this reason. But the segments show exactly why Strava just can't go ahead and flag every ride automatically if some speeds are exceeded.
Alright thanks. Makes sense indeed. Kind of a pity. Ah well. I live quite close to the small street with the bumps so I can create a better segment there. Yeah, it´s now at least 500m. Anyway, that 172kmh is just too funny. The person who had that probably also had a good laugh about it and a strong story to tell in the pub 😄
How is it that Strava continues to allow PBs that are ridiculously faster than the current validated WR?
People are deriding Strava for not doing simple data filtering:
It's annoying, not because of the people (who will sometimes do it by accident), but because it's permitted by the platform.
Strava, don't insult us by asking us report what you could filter out in a hundred lines of code and an interpolated WR lookup table.
Or can Strava please block all activities from a user who is just recording car trips. I have no idea why they are doing it - learner maybe as on average they are actually slower than a bike - then they hit the open road... Autoflag simply does not work.
I don't have any KOMs, but I pay premium and I just want the leaderboards to be clean. My wife got a trophy this morning and I was checking out the leaderboard - completely full of cars - and this one user recording all their driving. You start cleaning a segment up and you realise just how poor Strava data is...
Yes...thank you Davos. This issue has literally rendered this app useless unless fixed. Your suggested fixes are well within range of being implemented. I have also flagged blatant 50-70mph error rides and with no effect which is a waste of my time.
I also suggest the following:
1. Create separate leaderboards when riders contain heart rate and watt data as well. I personally only count a KOM if I see either of these in the metrics. Without either then I just assume its someone on an E-Bike/car.
2. Side suggestion/request... I also suggest an unlock via signed waver to allow live segments with -% grade to be implemented for those who train on segments with longer durations and require live feedback despite the grade.