cancel
Showing results for 
Search instead for 
Did you mean: 

Leaderboards - automatic re-categorisation of performances to 'E-bike'

Likesahill
Mt. Kenya

I am an athlete who has always used leaderboards as a motivational tool.
As I get older, I have focused more on the age-related boards. Having crossed the age boundary into 55+, all the hills that I target are heavily populated by mis-categorised ebike rides. 
I appreciate that the AI tool is looking for obvious cheats/mistakes, perhaps a KOM in a car.
For the E-bike problem I see, I suspect it is largely about more casual riders and ignorance of the rules. I see consistent patterns in the data: Low athlete activity levels, recorded on a phone, no HR/power data, cups on hills (but no other segments) slow times on all key metrics 5k, 10k, etc. 
My suggestion is that you set a performance threshold and automatically recategorise all rides that meet it. For example, given the E-bike issue (in Europe maybe not the US) is largely around hills, it would be simple to set a transparent rule that if you acheive, say, a VAM of over 1,000 on a segement and you don't meet certain criteria, then your ride will automatically be converted to an ebike ride. You could send a notification that this was happening and the athlete would then have the option of converting back to a normal ride. This would leave flagging for the hardcore 'cheats' and sort the problem with recreational riders who aren't aware of the issues they are causing on leaderboards.

1 REPLY 1

Jane
Moderator Moderator
Moderator

Hello @Likesahill 

Thanks for your post. Happy to hear you're enjoying segments on Strava. We have an existing automated system that detects invalid segment efforts (including E-bikes categorized as ride activities)

Although we do not catch all invalid efforts, we do detect and prevent many from ever reaching leaderboards.  We continue to iterate on our detection mechanisms and algorithms. 

Last month at Camp Strava we announced that we are working on AI-enabled Leaderboard Integrity, will harness machine learning to automatically flag irregular, improbable, or impossible activities recorded to the platform. Trained by millions of activities, this feature allows all users on Strava to play fair and have more fun.

Check out the Camp Strava announcements here

In the meantime, if you do see what appears to be invalid efforts on segment leaderboards, you can use our activity flag tool, or for repeated violations, report cheating.  Details here

 


Jane (she/her)
STRAVA | Community Hub Team