Strava announced new features and updates to their app and platform today, including changes meant to make Leaderboards more fair and accurate. The company says that earlier this year, 4.45 million activities were removed from leaderboards, and that the service has improved detection of e-bike rides uploaded either by mistake or on purpose.
According to a press release, Strava has introduced new machine learning models to identify “irregular, improbable or impossible efforts,” and has already begun using the model to remove offending activities from segment leaderboards in the app. Google explains that “machine learning is a subset of artificial intelligence that automatically enables a machine or system to learn and improve from experience. Instead of explicit programming, machine learning uses algorithms to analyze large amounts of data, learn from the insights, and then make informed decisions.”
Leaderboards, in the context of Strava, are used to track the fastest times across designated cycling and running segments, and many athletes use them to see how they compare to others. If a posted time seems suspicious, an athlete can manually flag an activity for investigation.
With the latest changes, Strava reports that flagged activities related to e-bike use are down nearly 85% year-over-year, along with overall flagged activities. The company says e-bike rides are now automatically detected and removed at the time of upload and are moved to e-bike leaderboards, which are separate from the leaderboards for unassisted efforts.
Strava did not share how the model works, presumably in part because sharing this information could make it easier for riders to find new ways to cheat the system. Athletes can still manually flag any suspicious efforts that the model may have missed.
In many cases, rides are unintentionally miscategorized by Strava users, and this update promises to address those rides as well. Before posting a ride to the platform, athletes declare whether an activity was completed on an e-bike or not, and it can be easy to forget to update this setting if you’re not a regular eMTB rider. Leaderboards can also be affected when riders forget to stop their GPS after the ride, and the drive home is recorded as a (very fast) ride.
Matt Salazar, Chief Product Officer at Strava, notes the improved models should make for “a fairer platform where competition can take place.”
Additional product updates
In addition to the leaderboard integrity improvements, Strava also announced a big update to their mobile route builder tool. The update is said to include additional route details along with integration with the Strava Global Heatmap so riders can quickly identify popular routes.
New Power Skills and Training Zones features are coming to the platform, as well, and should be available beginning in September.
0 Comments