Strava cleanup: more than 1.5M activities removed from Segment Leaderboards

The company found a way to detect forged results and remove them from official statistics.

One of the “features” Strava includes to keep its competitive users interested is its Segment Leaderboards.

Here, Strava users can compete in various sports, including running, cycling, and swimming, with the app tracking, matching, and ranking users' GPS data for those activities.

However, users began to notice that some athletes were cheating to reach the top of the leaderboard.

The discussion among Strava users revealed that the times and results shown in the leaderboard sometimes didn’t make sense, hinting that these results may have been forged.

For example, one netizen noted that while there are only 730 hours in a month, “yet some people have clocked thousands of hours,” sharing a screenshot of a profile that already has more than 3000 hours.

Together with the examples of such instances, users also questioned Strava's stance on the matter.

It seems the company finally took action on the issue, with its Strava engineer sharing the results of its efforts to solve it.

The company has focused on cycling activity by reviewing the top 100 activities on each Ride Segment Leaderboard.

Strava’s engineer shared that the company employed a new machine learning model (MLM) that can detect activities recorded as riding a regular bike, even when in fact they were achieved by riding an electric bike.

The company also “cleaned” its ride segment leaderboard to make sure that the activities that were achieved through cheating, by riding an e-bike or a vehicle, as well as entering an incorrect sport type, were removed.

Strava also launched another model that is supposed to catch when a wrong activity is uploaded to the leaderboard, for example, when a user uploads their cycling results as a run.

During the cleanup, Strava removed 2.3 million e-bike and 1.6 million vehicle activities from the leaderboard.

293k athletes were returned to their “rightful spot” in the top 10, reveals the Strava engineer on Reddit.

How Strava Detected and Removed 2.3 Million E-Bike Activities From Ride Segment Leaderboards
byu/strava-team inStrava

Users react to Strava measures

The news about Strava’s efforts to maintain segment leaderboards has been met with widespread positivity from the community.

One user expressed that they’re happy the company has finally taken some steps to address the problem, also noting that in “heavy tourist areas” like Sedona, many tourists like to rent e-bikes, so every segment here is done on an e-bike.

This prompted further discussion on similar “anomaly detection.”

“I assume that to get into the Strava top 100 with an e-bike, at the very least, you have to have a Strava account, which probably will have records of other rides. Unless someone got Strava just for the purpose of cheating in KOM with an e-bike, it should be relatively easy to detect anomalies,” noted one user.

One netizen noted that users sometimes simply mix the categories: “Some people genuinely don't know that there is an e-bike category. They think of their e-bike as a bike.”

When it comes to running, some people online joked about cheating users who won’t stop their watch on the drive home after their run.

“They should get a notification saying ‘congrats on breaking the world record for the mile!’” wrote one user.

“You ran faster than Usain Bolt !!!1 why are you not at the Olympics ??!” added another.

Unlock more exclusive Cybernews content on YouTube.