Seeking purity

I’ve been toying with publishing this post for over a year, as I still don’t feel qualified enough to write about data analysis. Nevertheless, one can always learn more, so I decided to post this anyway.

Over the last few decades, technological developments have enabled what many cyclists currently take for granted. As a child in the early 90s, I was very proud to own a speedometer for my bike. It was a big red plastic box with a dial and a big needle in the middle, and worked by a being connected to a little wheel touching my front wheel.

Since then things have moved on a lot. With my current bike computer, I now have speed, distance, time, temperature, cadence, heart rate, power, a map with routing (including adaptive rerouting), elevation, and climb profile – all literally at my fingertips. It’s really is a computer in its own right, with a touchscreen, coloured graphics and internet access (either through Wifi or a sim card).

Some metrics can be displayed in real time, such as average speed, power over different timeframes, heart rate and power zones. This (in theory) allows me to train (keeping in various power zones depending on my individual goals, displaying live training exercises) and race more effectively (managing my power output whilst keeping an eye on my speed and cadence), and even properly navigate (designing or downloading a route, or simply letting my computer point the way).

After a ride, services such as Strava, TrainingPeaks and intervals.icu take this data, process it and present it in a variety of charts and graphs. I can view this for an individual training session or a race, allowing me to see how I performed and where I could improve, compare this to my historical data to derive patterns and trends, and even compare with competitors who share this data.

Its mind bending what has been achieved in such a short time, and how quickly it has been made available to the amateur scene at a reasonable price point.

So what’s the problem?

Purity vs data-driven cycling

From various online forums, groups and face-to-face chats, it is clear that this incredible development is not always welcome.

One argument is that cycling has become too data driven, and much like the days of old, cycling should be done by feel rather than analysing data. After all, the greats such as Eddy Merckx didn’t have access to bike computers or sensors, but still managed to dominate the professional scene, with their records standing for decades.

Developments in my bike computers – first Sigma 1606L, a simple device which worked using a magnet on the wheel, a Wahoo Elemnt Bolt, compact and designed for racing but had a relatively small map, and the Hammerhead Karoo 2, which is highly adaptable and receives regular software updates and new featured.

The opposite argument is that having access to this data allows for the optimisation of performance. After all, what can be measured can be optimised. The logical consequence is that a cyclist who can analyse this wealth of data and act on it should have an advantage over a competitor who is – for lack of a better word – riding blind.

The minimalist approach is simple enough – either ride with no bike computer or sensors, or use a computer which only offers basic data such as speed, distance and time. There is a certain romanticism about this purist approach, and even riders who comprehensively collect data are often told to ultimately listen to their bodies. Physiologically this makes sense – your body is chock full of sensors, and listening to it is arguably a healthy and sustainable approach.

Hopefully it goes without saying, but this post focuses mainly on competitive cyclists and not those who simply want to ride from A to B. Power, cadence and TSS might not be of much help to nipping out to the shops, or in fact to most people casually riding their bike.

Successfully training on a bike (as with other sports) involves pushing the body to its limit. Indeed, cycling coach Joe Friel notes that effective training involves flirting between overtraining and staying within healthy limits. Whilst fatigue is a valuable protective barrier, simply listening to the body whilst training risks erring on the side of caution and therefore not resulting in developing one’s athletic potential. Data collected by sensors and analysed allows for an objective review of one’s performance.

An interesting recent example of data in sports is the astounding performance of Anna Kiesenhofer at the 2021 Tokyo Summer Olympic Games women’s road cycling event. Almost from the start line, amateur time triallist and physicist / mathematician Anna broke away from the peloton, and in a second step broke away from the break away still with 40 km to go, and effectively time trialled her way to a win, clearly beating world class cyclists from the pro peloton. Watch it here, it’s amazing.

At that time, Anna was not a professional cyclist, as many female athletes cannot afford to cycle full time with current wages and prize money. In her day job, Anna’s specialisation was studying fluid dynamics as applied to cycling (at speed, air displays the characteristics of a liquid), which is directly applicable to her hobby of time trialling. Anna is also a self confessed data nerd.

Put these two factors together, and Anna was well placed to properly train herself for the Olympics, and to manage her energy on the day – not burning out too early, but ensuring that she could optimally manage her energy right up to the finish line. It’s worth noting that Anna trained herself whilst the Dutch team, with Annemiek van Vleuten coming in second, had much more structured support. Data collection and analysis, targeted training, strategic planning – and a handful of luck – clearly worked.

What’s the right way?

What about my opinion? Let’s caveat this by saying that by no stretch of the imagination could I be considered a serious / professional athlete nor an expert in the field of sports science. I’m a man with a time consuming job, who spends his free time riding his bike furiously at speed and improving through amateur racing.

As a scientist however, I am interested in data and its analysis, and through racing I have developed an understanding and respect of the capabilities of my body and mind. During a race, there is clearly only a certain amount of data I can and want to see, and I’ve set up my computer accordingly. Even so, I generally quickly look at the data rather than fixating on it. After training or a race, I would like to review it in depth, however to be honest I’m still trying out how to manage it all and use it properly!

I really do understand the wish of many to keep cycling pure, especially with the astounding and rapid technological developments such as aerodynamic carbon frames, disc brakes, integrated cabling, tubeless tyres, hookless rims, self inflating tyres (hell yeah) – the list goes on. The changes we are witnessing can be disorienting. One can easily feel nostalgic about the good old days where a cyclist was more defined by his or her strength, fitness and endurance rather than the bike and data, and seek to replicate that.

A different perspective – seeking romanticism?

Seen from a historical perspective, this nostalgia is arguably a logical fallacy one commonly encounters. After all, even in Merck’s prime, despite the limited technology when compared to today, each rider was looking to get the most out of him- or herself and the bike to win. Often a change in thinking or technology unlocked a competitive advantage (ignoring doping!), such as Greg LeMond’s use of aero bars to beat legend Laurent Fignon at the 1989 Tour de France by 8 seconds.

Plenty more examples of technological advantages exist, but one only needs to look at the original hobby horses originating in the late 19th century, compared to the first safety bikes introduced at the turn of the 20th century, compared the first racing bikes of the Tour de France which had one gear (what hipsters…!), compared to the racing machines in the 1970s with derailleurs. We can then start to appreciate how our modern high tech machines (which include the use of sensors and data) fit into the developmental spectrum.

Despite the merits of the purist and the data driven approach (as opposite ends of the spectrum), the criticism of technology in sport does seem rather unusual. There aren’t too many people who yearn for the good old days of double clutches in cars, unfolding and using road maps whilst driving, using paper address books and finding change for the phone box, or even flying in early propellor powered aircraft.

Right or wrong, there is a human tendency to yearn for the positive and romantic characteristics of things gone by, but we almost conveniently overlook the negative characteristics which don’t fit into our idealised image (known as the bias of declinism). After all, an LP is wonderful until it gets scratched, a VHS is homey until the tape gets stuck in the machine, and rim brakes are stylish and chic until you’re riding downhill in the rain. If we want purity in cycling, shouldn’t we also want the full package – the woollen jerseys, pedals with straps, five large gears, a paper map, downtube shifters, riding without helmets, … ?

In other words, seeking purity in cycling is highly subjective, and risks deliberately ignoring the benefits of the natural technological progression in our sport. That being said, for us non-professionals, the sport is what we make of it – each rider should find the right balance which gives him or her the most joy in cycling.

For those who like pure nostalgia, today there are races where only classic bikes are allowed (and you have to wear wool and use pedals with straps – this includes the well known “Eroica” series – aptly meaning “hero”.

I am a data nerd, but sadly lack the time and experience to properly analyse it. This is something which I can however learn and practice.

There are plenty of places on the purity-telemetry scale where people can find the balance that is right for them. Let them find out what is right for them. Either way, it comes down to the same thing – having fun on a bicycle.

---

PS. A quick shout out to two great websites which promote rationalism in arguments, some of which I’ve used here: on logical fallacies and cognitive bias. Don’t forget to download their free posters!

What do you think? Is a spectrum too simple an analogy? What type of cyclist are you when it comes to data?