What is it?
A great option for the rider who wants to keep his or her training going strong when outdoor riding loses its appeal for the year, Zwift offers the one of the first fully integrated virtual cycling programs that actually makes indoor training an enjoyable process — any time of year. For $10 per month, you can make your own custom Zwift avatar and engage in racing with real opponents around the world, fully customize individual workouts, and ride in three different worlds, each with multiple courses to choose from.
The foundation of the Zwift platform is real-time power meter data, heart rate data, and/or speed and cadence. With one of these metrics, plus your weight (assuming you’re honest), Zwift gives athletes an incredibly accurate simulation of the riding experience. Plus, in six months of testing we’ve yet to see a car.
- Inexpensive subscription package
- Real-world, real-time interface with cyclists from around globe
- Makes training at home truly social
- High degree of hardware compatibility and accessibility
- Provides good to amazing simulated ride feel, depending on hardware and connectivity
- No cars
- Decent graphics
- Motivating to ride with others in real-time races
- Limited integrated user checks/hardware inconsistencies can lead to rampant imbalances in equipment readings (intentional or not)
- Inability to truly assign w/kg protocol in group ride setting can make group rides annoyingly competitive
- Software bugs, but always improving
- Full roads during peak periods
- Frequent lags in data transfer experienced on multiple networks with multiple hardware setups
- Battery intensive, must plug in before plugging in
Now in its third full season, Zwift’s virtual cycling platform is hitting its stride. On a computer monitor or compatible iOS mobile device, it enables you to circle the roads of three virtual worlds (the original Watopia island course, London, and the Richmond World Champs circuit).
Zwift has a number of bikes to choose from, with availability based on experience level attained (1-25), including a handful of real-world road bike replicas and TT rigs. With a simple heart rate connection and any of a number of common trainer modules, anyone can access this platform. Further connection to a power meter provides another level of real-world simulation and stretches the accessibility to almost any trainer on the market.
In what has been a particularly harsh fall/winter in the Pacific Northwest, I put in a ton of hours on the system, testing limits (both my own and the software’s), identifying weaknesses, and watching Zwift rapidly improve in terms of reliability and accessibility.
When paired with a basic magnetic resistance home trainer, Zwift is fairly consistent and offers a good influx of stimuli to make training both enjoyable and efficient. Unfortunately, such a setup offers a limited degree of simulation. Uphills and downhills don’t feel any different. At a constant 200-watt effort, a downhill corresponds with an observed rise in speed on the screen. But the resistance on the basic trainer lacks correlation in this scenario. The same is true of climbs.
But when paired with a smart trainer, you’ll feel every undulation almost to the point of feeling like a real ride minus the steering. You’re effectively on a track in the world of Zwift, able to choose a limited number of set turns with the click of a button. All other avatar steering is built into the program, automatically accounting for staying in your lane or avoiding other riders who fall in your way. An important note here is that the simulated ride feel described above is observed only when your smart trainer is set to sim mode (default when the system is in free play or in event mode). When your smart trainer is set to erg mode (default when the system is in workout mode), however, such simulation is absent.
Power protocol and resistance are coupled almost perfectly. This setting comes in handy for the most efficient training workouts. But it’s also not at all representative of riding a bicycle, which can have the unfortunate effect of turning you into a bit of a robot after a long winter.
Route Choice & Different Gameplay Modes
The three worlds available in Zwift are rotated through at Zwift’s discretion, thus your only choice as to which world you’d like to ride is rooted in which day of the week you choose to ride. Zwift’s original Watopia world, our favorite for its variation in terrain and broad range of scenery, is typically available 3-5 days per week, while Richmond and London are typically available 3 or fewer days per week. Within each of these “worlds”, there are a number of preset route choices that are available upon login. Some are hilly, some are flat, and all are adjustable in-ride if you get bored simply click an arrow key upon hitting a route junction.
There are three primary gameplay modes in Zwift. Freeride mode is where the rider can choose any route they wish within the predetermined world. Simulated riding is most effective in this mode on a properly connected smart trainer. In this mode, you can also select to ride with a friend in a list of all riders currently active on Zwift, joining them in real time as you start your ride.
In workout mode a rider can select a workout from a fairly substantial list, adjusted to their chosen functional threshold power number (FTP). Or you can create a custom workout, though the custom feature is currently unavailable in the mobile app. In this mode the rider sticks to a preselected course and focuses only on the heads up, easy-to-read feedback based on power protocol, cadence, and/or heart rate data. Workouts are easy to follow along with in Zwift largely thanks to its innovative heads-up display and power/cadence queues.
Finally, in event mode, riders line up at a set time with others who have chosen the given event. There are typically 1-5 events per hour, and groups can range from a few to several hundred riders, depending on the popularity of the event. Once the event begins, you’re launched into a preset course and either racing your competitors to the finish or riding at a predetermined w/kg protocol to simulate an organized group ride. While the races are a great way to motivate each other and push your limits, the group ride option is mostly a bad joke when applied to the “recovery ride” or low wattage protocol scenario. That’s because there are always a handful of riders determined to break the ride into pieces by riding above the agreed wattage.
Not a Perfect World
As you may be have already realized, this article should be taken with a grain of salt. We’ve done our homework and put in our time with the goal of examining the system’s general benefits and drawbacks for a variety of riders. But we’ve also determined that such wide system integration — precisely what makes Zwift so successful — is also a disadvantage from a certain perspective.
Inconsistencies inherent in various models of power meters (even within the same model group), differences in trainer technologies, and even disparities in connections speeds, can all contribute to some fairly sizeable differences in virtual power readout, which in turn changes riding speeds on Zwift per given exertion level, the basis of the program. Not all watts, in spite of what your Zwift screen may have you believe, are created equal.
The pairing screen that every ride begins with is fairly straightforward, assuming you’re not in a room full of other riders with their own ANT+ or Bluetooth devices. If the latter scenario describes you, you’ll need to isolate your device codes either through trial and error or memorization to properly pair with your computer. All you need to make this pairing work is the ANT+ dongle or compatible Bluetooth enabled mobile device. In testing, this pairing almost always worked as intended.
Workout mode was almost always impacted to some extent by lag time in communication between power meters and the computer (several different varieties on several different connections), sometimes even forcing us to bail on the entire workout out of utter frustration. A sustained 300-watt effort (as confirmed by redundant readouts on a Garmin Edge head unit) often varied between 400 and 200 watts on the Zwift monitor, prompting the workout screen to alert us of either “LESS POWER” or “MORE POWER”, and a subsequent flunking of the workout as prescribed. This seemed to be more common with multiple riders on the same network, and more common on smart trainers, especially when in erg mode.
Logging in occasionally provoked its fair share of frustration, as well, especially when a desired group ride’s departure time was nearing. This was experienced in multiple system configurations on multiple networks. Bugs in which my avatar drifted off the track were also experienced multiple times, never truly presenting an issue, but often provoking some uncertainty as the avatar drifted from the center of the lane into the middle of the ocean or off the bridge. These minor drifts always self-corrected, however, and seem to have finally been corrected indefinitely.
Another bug observed toward the latter part of 2016, but not yet in 2017, involves the avatar simply disappearing from the screen, temporarily making group rides and races extremely difficult. We found it mildly to extremely annoying to have to separate rides in Zwift when switching between gameplay modes.
When jumping to event mode, it’s convenient that Zwift both prompts you of a scheduled event time and then jumps your avatar to the staging area upon the click of a button without booting you out of the system entirely. But even in this scenario the ride files are annoyingly separated. The more substantially inconvenient effect of this procedure is found when jumping from event mode or freeride into workout mode. To make this leap, you’re required to quit the system entirely and reboot, which not only separates files but also requires another pairing which can also be a lengthy process.
The last major concern is the inherent system battery usage. This program is extremely energy intensive, requiring power connection for anything over about 30-45 minutes to avoid complete battery drain. Don’t forget to plug your computer or mobile device into the wall before hopping into a Zwift ride.
Inherent System Inequalities and Intentionally Exploited Inequalities
Beyond the honest technological limitations observed in our 6-month testing period, the Zwift community is rife with gross disparities that seem to exceed the realms of the aforementioned normal technological differences. One of the biggest of these disparities is the easy manipulation of rider weight. All speed simulation in Zwift is based on the basic function of watts per kilogram. This is the most straightforward metric upon which speed can be based, correcting for differences in rider weight.
For example, a 120-pound female can be produce a lot of power relative to her body weight, yet be producing significantly less gross wattage than a male rider at 225 pounds. Yet, that female rider’s watts/kg is likely far higher than her male counterpart and this system is designed to recognize that intricacy with regards to the rider’s simulated speed, especially while climbing. This is a metric borrowed from the real world, and it has wonderful application here in Zwift. Unfortunately, a 225-pound male can easily turn his 2.5 w/kg threshold into 4.5 w/kg or more simply by dropping his rider weight in the system by a few dozen pounds with the click of a button.
To be fair, Zwift does correct for this in some forms. One example is that King of the Mountain titles only honor riders who have completed these respective segments at the same weight they started the ride with. So I cannot start a climb with my friend, get dropped, lower my rider weight in settings, then catch back up and beat her up the hill without the system automatically disqualifying me from that KoM competition. Races actually have an even more significant safeguard for this kind of cheating, requiring riders with an average power output greater than 5.0 watts per kilogram (Tour de France-level numbers) to provide ZWADA (Zwift’s international governing body) with real-world substantiation of their ability to perform such an athletic feat.
Beyond identification, the prevalence of these “intentionally exploited inequalities” speaks volumes about the psyche that dominates the sport. You can see these trends on the road, too. But removed from direct observation in the real world, certain liberties are easily and often exploited in Zwift with the direct, sizeable, and absolutely intentional effect of displaying incongruous statistics that make the rider look far faster/stronger than he or she truly is.
For one, it begs the question, “To what end are these steps pursued?” Arguably, training does not benefit from artificially enhanced statistics — quite the opposite is likely true. But then the next, and more ominous takeaway is simply, what does this say about the character of our sport? The fact that domestic-elite riders, even seasoned domestic pros cannot win — or sometimes even finish — races on the Zwift platform that are comprised primarily of amateur, recreational riders from around the world, is at least mildly unsettling.
Despite e-doping, and presumably even biological doping, Zwift is nothing short of impressive. Increased course options and constant improvements in development have addressed most of our concerns. As long as you can deal with a few cheaters and the cumbersome pairing processes, there is a lot to be gained from honest participation in the program. Zwift is engaging in a way that riding a trainer along will never be. It’s social, allowing riders to communicate with, compete with, and train with friends and strangers from around the globe. And it provides sustainable motivation in a space that has long been the bane of riders across the entire continuum of skill and experience.
Furthermore, the software is impressively accessible and connects with just about any combination of equipment on the market. For all these reasons, along with its low-priced subscription-based packaging, we’re big fans.