Training social platform Strava is having a troublesome time trying to seperate the frauds from the phenomenal on its segment leaderboards. This is causing outcry amongst its users and is adding to backlash the company has faced in recent times. So what can be done to tackle the problem of cheating on the app’s keynote feature?
The concept of a Strava segment is simple. After uploading a sporting activity to Strava, a cyclist can use their GPS file to create a segment of their route to offer a friendly competition as to who is the fastest person to ride that segment. These can be sprints, TT courses or climbs. The only factors for a segment are that it must be a minimum of 500m in length (up from 100m when the platform first started, and 300m from a few years ago), must be part of a real road, trail or paved network, and is not in a place where cycling is prohibited. Also, segments can be flagged as dangerous if segments follow shared footpaths or other obstacles that could cause harm to a Strava user or a member of the public.
Segments are meant to be a fun and friendly way of getting competitive without any reward or incentive. You don’t gain anything by obtaining a coveted crown on your local segment. The sole purpose is to challenge your mates on the local climb/course, to push each other and have fun whilst doing so. So surely, there’s no incentive to cheat, right?
Not so, it seems. The number of activities that have been flagged as potentially cheating of late has seen a huge rise, and there are beliefs that many more are being missed. It is frustrating as a platform user, as it sucks the enjoyment out of what is one of Strava’s key features that draws in so many users globally. So why would they do it, how do they do it, and what measures can be taken to stop them getting away with it?
Even though it’s been nearly two years (yet feels like only yesterday) that segment analysis and detailed KQOM leaderboards became exclusive to subscribers. However, you can still see a Top 10 leaderboard and obtain a K or QOM position on said leaderboard if you are a non-subscriber. Subscribers are by far less likely to be flagged as questionable, as quite often they provide additional data from their rides, such as cadence, power, heart rate and other performance metrics that can see in detail that they achieved goals legitimately. However that does not mean that mistakes happen (forgetting to finish a ride before packing your bike into the car, leaving your device running) or they won’t try to cheat at all.
Strava uses data analysis such as acceleration and speed data recorded by devices to automatically check for potential fraudulent activities. If the algorithm deems an activity’s speed or acceleration data to be above a certain threshold, an activity will automatically be flagged. In instances which fall under that criteria threshold, users can flag an activity if they suspect it falls foul of cheating. It will then be analysed before being declared plausible or fraudulent. Sounds easy, but it is actually quite complicated.
So what methods do people use to cheat? In the case of cycling, one way to cheat is to use Electronicly Assisted bikes; these have their own activity category and KQOMs in Strava, and are not supposed to be eligible for cycling segments. They must be uploaded as E-bike activities. This is a fair system, as E-bikes do give a significant power advantage on technical terrain and going uphill, until their limiter cuts out. However, even accidentally, people can upload E-bike rides as a non-assisted ride activity. This is probably one of the harder ones to spot, but there is more often than not a much lower speed variability between uphill and on flatter or downhill terrain. But the more blatant cheats will use fully motorised vehicles. Specifically, motorcycles, as they can dodge traffic much easier and dirt bikes/motocross bikes can easily be manouvred onto natural trails that do not have many defence mechanisms against motorbikes. If a rider times their speed and acceleration carefully, they can beat the algorithm meant to try and detect them.
So in order to register a segment attempt on Strava, all you need is a GPS activity file recorded direct from your phone or smart training device (smart watch or GPS computer) from when you complete the segment. So providing your cheat method is not spotted, and you keep your speed within the tolerances of the algorithm, your attempt will register as genuine. But how is cheating kept at bay on other platforms? For instance, online virtual riding platform Zwift hold ameteur open races with no prizes, yet has a fairly robust system which deters and detects cheating very effectively. In order for race results to be registered, riders must have a (free)ZwiftPower account connected to their Zwift profile, and must also submit data from] a Heart Rate Monitor (mandatory, especially if using a “dumb” trainer) and preferably the power data from their smart trainer. This is to ensure riders are putting in the efforts required to perform on the platform, and any anomalies lead to disqualification from the official race results, and could lead to bans from races for repeat offences. If you do not have this data then your result will not be verified on ZwiftPower, even if you come last. So can we do something like this on Strava?
In theory, yes. One move Strava could make is to make the leaderboards subscriber only. This, however would be a highly unpolular move and would likely see a drastic drop in users. This is also a highly unlikely move as the company CEO has previously stated that the course segments are an integral part of what attracts people to Strava and will not be put behind the paywall of its subscription service. So what else could it do?
Strava could make it a requirement that all segment attempts are void unless certain data forms are present in the upload. This could be as simple as using a bike speed and cadence sensor, or, like Zwift, use power meter and heart rate data. The latter, however, is likely to be less popular, particularly amongst legitimate casual users of the platform. Power meters are expensive, only dedicated cycling enthusiasts can justify such an expenditure, if their pockets can stretch that far. It will exclude many of the platforms users despite being the most effective solution out of them all.
A HRM is much cheaper and would work, but again, casual platform users are less likely to own and use a HRM. (It is worth pointing out, some newer smart watches do have the capability of measuring wrist pulse built in, but they are also an expensive item).
The way in which it would negatively effect the least amount of legitimate users would be to use data from speed and cadence sensors. These are a similar price to entry level HRMs and are usually the first sensor purchases by most cyclists at any level, however, they are fiddly to install, and will need to be calibrated to your wheel size (either manually, or through using GPS data in use). Furthermore, Strava removed sensor connectivity from its phone app, as most people who used sensors paired them directly with a third party GPS computer. The use of Speed and cadence sensor data to deter cheating can be bypassed in the case of e-bikes, providing the bike’s own systems aren’t connected to the recording device.
The cheating cannot be left to continue unchecked as it is. It is not viable, and will see people leave the platform, and damage the integrity of Strava’s business. So action is required, and tying in supplementary data seems like the most obvious way to implement such measures. I, personally, wouldn’t mind either the speed/cadence or heart rate data (but perhaps not all of it) being used to verify my KOM attempts, even though I own 5 different bikes for different riding disciplines (on and off-road). Most of my bikes have speed and cadence sensors already attached but I would need to either make sure the remaining bikes have their own sensors, or invest in a HRM which I can use with any bike. Like many people, I cannot afford a power meter and even the cheapest options on the market are worth more than any of my bikes (except maybe my best road bike, which is probably worth about the same). People can choose whatever option will suit them best if they want to continue using the platform, and is unlikely to see a mass exodus of disgruntled users. There may be some who leave, but by excluding as few as possible, the app should maintain its integrity amongst the fitness and cycling communities.