Reliability of KOMs and Leaderboards? Should we have KOM repeats or median derivation?

I'm not a Strava guru but am wondering if/why Strava don't do more with flagging suspicious segments that produce KOM and top 10 leaderboard positions.  I'm certainly not making allegations here, but when there are significant GPS discontinuities in a activities or you (for example) see a 2km segment pace EXCEEDING WORLD RECORD 1500m PACE, you have to wonder a bit.  Why aren't KOMS/leaderboards a little more protected by Strava AI.  Maybe a KOM should only be aggregate or median of 3 attempts or something given all timing and data gathering is uncontrolled.

https://www.strava.com/segments/3813459?filter=overall

Why should users even need to flag this stuff?  It's immediately knowable.

9

Comentarios

12 comentarios
  • You'd think.

    Heck, on Strava, you can log a 2000km bicycle ride at an average pace of 490 kilometers/hour, and all's good. (https://www.strava.com/activities/858295114).

    Where, exactly, to draw the line is certainly a matter of discussion, but with cars, planes, and trains you often get sustained speeds over 100kph, which seems like a pretty low bar to handle.  I'd guess that most folks logging these kinds of "activities" don't realize that they are pissing in the global well, so one can imagine multiple benefits in auto-removing from leaderboards, along with some kind of "geez, are you sure that's really a run/ride?" notice.

    And in the end, if someone's valid activity is automatically flagged this way, the worst is that they unflag it, which itself is probably a heck of an ego boost.

     

    6
    Acciones de comentarios Permalink
  • Recently, I completed a ride and forgot to turn my Garmin 520 off. I'd driven a couple of miles before I realized what I had done, stopped, and turned off the device. So, for a couple of miles I recorded 100 kph speeds. Fortunately, I did not travel through any segments so it did not matter in that regard. I suspect this is what happens on those high-speed segment results. 

    On the other hand, there are some "legal" ways to attack KOMs. On sprints, people get lead outs, and on longer segments, group or team riding can lower times. And then there are windy days. 

     

    2
    Acciones de comentarios Permalink
  • Rick, you can trim the car part from the ride (via the web interface, not the phone).  That'll avoid getting your whole ride flagged if someone ends up making a segment in the future that covers the car part. It also keeps your stats "pure", in case you care about that kind of thing...

    2
    Acciones de comentarios Permalink
  • @ Jeffrey Friedl, thank you very much. Trim completed. Conscience clear.

     

    1
    Acciones de comentarios Permalink
  • It's problems like this that make it hard to justify paying for Premium. Heck, that one run linked by Jeffrey took place at 30,000 feet. Anyone care to find me a physical location in North America that's at 30k elevation? On Earth? Maybe the AI could start with "There's no place on Earth this high."

    3
    Acciones de comentarios Permalink
  • Wow... I had a chuckle at someone running @ 36Kph for 2Km and only having a HR of 133... No... no wheels involved there surely! then saw the KOM in 6seconds that Jeffrey linked to. Wondering where I get wings on my bike like that...

    I agree that there should be some kind of automated BS detection in place. Results that are a long way outside your regular performance for example. There should well and truly be enough data around for Strava to do analytics on this and determine that it wasn't a real attempt.

     

    3
    Acciones de comentarios Permalink
  • Screening these obvious impossible feats from the segment leaderboards seems doable and I think everyone would agree it would be an improvement.  Just think what you could accomplish, Strava team, if you spent all that time trying to predict what order you think we want our activities to appear on an actual problem.

    Or maybe you could randomly shuffle the segment leaderboards or try to predict what order we would want the names to appear ;)

    1
    Acciones de comentarios Permalink
  • Same for challenges. Always wondering how it is possible to ride more than 4K elevation day by day during the whole month.

    1
    Acciones de comentarios Permalink
  • I've had to flag a few rides for obvious car segments, both for KOMs of mine that were surpassed by guys driving home from a ride with the gps on, and for other segments where I was looking at my own performance and just noticed that some dude with an obvious mountain bike ride at a place I've ridden my MTB many times didn't really start going 50 mph as soon as he hit the pavement on his way back home.

    Yeah, you'd think Strava could implement some fairly simple BS detection. I mean, just start with world record performances and flag anything higher. I mean, if some guy really did set a world record and had to contest a flag, then OK, there's that, but how often is that going to happen compared to what we see now, with some many legit KOMs being blown away by doofi (plural of doofus) who strap their bike to the back of their car and drive home before stopping their ride?

    2
    Acciones de comentarios Permalink
  • It's the biggest drawback of the platform. It is incredibly frustrating for athletes using the prospect of a CR or KOM to keep them motivated just to have it taken away by some doofus (thanks for that word Seth) who records a ride as a run,...etc. The issue has been around so long Strava obviously don't regatd it as a problem. They've never made any comment even acknowledging the problem as far as I'm aware.

    2
    Acciones de comentarios Permalink
  • Upgrading to Summit is a total no-go for me unless the segment leader board accuracy is sorted out.The entire integrity of the platform is in question when a key motivational aspect is broken.

    Implementing basic rules such as that the average speed cannot be greater than the maxiimum speed would be a great help and simple to program I expect.

    It is frustrating when you work hard and achieve a good time only for someone to come along and claim the KOM with bogus data that is obviously wrong (avg > max) and yet Strava does nothing to prevent this.

    So no Summit upgrade from me as long as this problem stands.

     

    1
    Acciones de comentarios Permalink
  • I'm not upgrading to premium either unless this auto-flag system is put to work. There are so many "rides" that would require continuous power over 3000w that it is just ridiculous that nothing is done.

    If someone is riding a full fairing recumbent streamliner, fine, at least require HR and power data to validate it.
    If someone drafts behind a truck at 70km/h, fine as well, have the system do a data check and also check if the person was cycling before and after the effort, because very often people do a ride, forget their device recording, then go drive. At least someone doing the rare thing of drafting behind high speed trucks will have normal cyling data just before and right after the event... So, if they are actually on their bikes, "no problem". Plus in those cases the system should also require some data such as HR at least, or even power, to validate results, since no one rides really fast with 80bpm, even if drafting behind a huge truck.

    0
    Acciones de comentarios Permalink

Iniciar sesión para dejar un comentario.

¿No encontró lo que buscaba?

Nueva publicación Agent Only: Button marked