Skip to main content

First off, I am happy to see the big improvement in the automatic detection of impossible efforts across more activities.  Specifically, I am seeing it a lot more for running where people who are posting activities with sub 4 min/mile paces are frequently being marked as having a problem.  This is good and a great step forward from where things were. 

However, there is still an issue.  These activities that have been marked as auto-detected are still being listed in the challenge leaderboards.  The top 20 of the 5k running challenge for example is almost all activities that have been marked with “We detected an issue with this activity”.  Some of them are a week old and still showing on the leaderboard.  Why aren’t those activities removed from the leaderboard immediately when they are detected?  If the person ends up clearing up the issue, then the activity can be re-indexed onto the leaderboard, but in the meantime, they should be removed as soon as the system sees there is an obvious problem.  

@Jane ​@Scout - Has anyone taken a look or can explain why tons of activities that have been auto-flagged are still being allowed to remain in the challenge leaderboards?  The first two pages of the 5k running challenge are almost entirely made up of activities that have been auto-flagged, yet they are still there taking up the top leaderboard spots.  This is really disrespectful to those who legitimately post a time that would put them high on the leaderboard yet because all of these obviously false activities that have been identified but left in place are taking the spots, they are listed many pages back.


Hi ​@anchskier, we appreciate the feedback! One of our product team's next steps in iterating on the upgraded auto-flagging system (which you can learn more about here) is to reprocess the top 100 activities on every global ride and run segment leaderboard with this updated system to help ensure they are as free from vehicles, incorrect sport types, and eBikes as possible.


@Marya - It isn’t that the rides/runs aren’t being detected.  They are, which is good.  The problem is that they aren’t being removed from the challenge leaderboards.  They sit there taking up pages of the top challenge leaderboard spots even days after they have been auto-flagged.  It isn’t the segment leaderboards where this is happening, it is the challenge leaderboards (like the 5k running challenge for example).


How do you reclaim wrongfully flagged activity? Not very confident of the algorithm yet. A quite steady pace club race (1 hour long) with roughly 41 kph has been flagged.. If it starts flagging those kind of activitie oh boy..


@Joltin_Joe On https://support.strava.com/hc/en-us/articles/360040099312-How-to-Resolve-an-Activity-Flag is explained how to remove a wrong flag.


@Jan_Mantau but there’s no “my activity is fine” option for automatically flagged activities.. 

 


@Joltin_Joe - Can you post a link to the activity in question?  There might be something else in the data that is tripping the filters.

 


Sure, here’s the link: https://www.strava.com/activities/82892741


I’m astounded the Strava algorithms analyze 16 year old activities. As there is no heart rate or other sensor data in it (pretty normal for that time) in conjunction with the high average speed and no road or triathlon gear attached it seems to find that ride suspicious. I fear you have to ask the support to remove the flags.


Ok, thanks. I believe at a time this has been tracked with Sportstracker mobile app etc and exported to Strava. Didn’t know that 40 kph gets you flagged;)


I am wondering if it has to do with the elevation profile compared to the speeds.  Was it really that hilly?  There are a bunch of spots along the course that it shows you going 25-30 mph or even faster at locations with a grade of over 10%.  It looks like they are just very short bits of the course, but could look off to a computer analysis.  I’m wondering if there was some bad elevation data collected in the file and it is causing the analysis to think the speeds are unrealistic.  


Reply