An epistemology of speed

In Karl Popper’s magisterial book The Poverty of Historicism, the philosopher warns about extrapolating the future from the present; given that we are unable to adequately describe the present, even the future will be elusive to us. “It is not possible for us to observe or to describe a whole piece of the world, or a whole piece of nature,” he wrote. “In fact, not even the smallest whole piece may be so described, since all description is necessarily selective.”

Popper was talking about the great sweep of human history and criticizing the idea that there are laws of development in history and that these laws are evolving towards some sort of ideal end point. But even on a smaller scale, Popper was wary about relying on certain ‘truths’. In his epistemology, we accumulate knowledge of the world through the process of deduction. We put forward hypotheses and test these with appropriate observations. The rules or laws that we discover are at all times subject to falsification through further testing. Some laws withstand a substantial amount of testing, and are therefore more robust than others, but are – in a philosophical sense at least – still falsifiable. Overall, we must always be critical. “For if we are uncritical, we shall always find what we want,” Popper said. “We shall look for, and find, confirmations, and we shall look away from, and not see, whatever might be dangerous to our pet theories.”

This post will look at the question of speed on a bicycle, and how we achieve it, with these two approaches in mind: the difficulty of accurately describing the present situation, and hence the difficulty of predicting the future; and a critical approach to established theories of going faster. As such, I argue for a much less obvious epistemology of speed.

Fast bikes and fast wheels

It is common, in advertising and in reporting, to see claims for fast bikes or fast wheels. Witness Cycling magazine recent testing of a suite of ‘super fast superbikes’. An ad for Ritchey components (and I don’t pick on Ritchey in particular, because the advertising style is common) reads: “Tom Ritchey swore he’d shave his moustache when Ritchey components stop winning races.” The accompanying picture shows Ritchey still with his moustache.

The implication in both these examples is that bikes and components have certain intrinsic properties and that they can be described as ‘fast’ or, by extension, ‘slow’. But these properties are not intrinsic to the bike or the components, they are all relative to something else and they are only made so by the addition of a rider. For example, Ritchey’s great looking wheelset in the ad described above is not going to make me win races – that I can guarantee. If that is what the ad was implying, he had better shave that mustache. We might say, therefore, that these bikes or these wheels have the potential to be fast (or faster than our current ride, presumably) with the addition of a rider.

The most well-known universal ‘law’ in cycling is that lighter is faster and that a lighter bike (and particularly lighter wheels) will enable one to go faster – and more speed is, after all, what we want the most. For example, in a recent test of a bike in Bicycling magazine, the tester noted that for the particular bike, “we noticed the difference on group rides, when we struggled to keep up with riders on lighter bikes.” This is powerful stuff – is was not that the other riders were 140-pound Spanish whippets doing interval training over rolling hills, or that the test rider was an out-of-shape Cat.4 racer with a hangover, but that the bike was too heavy. Given that the bike in question, at 18 pounds, was about half a pound (the weight of a half-full drink bottle) heavier than the bike that Lance Armstrong won the 1999 Tour de France on, this is a serious charge meriting further investigation.

In the spirit of Popper, then, let us take a deductive approach and put forward a hypotheses, and one that we can really test. Focusing on wheels is a good choice. Riders now have a huge array of custom built wheels to choose from, not just the old choice of 32 v 36 hole, single v double-butted spokes, and clincher v tubular. We now have varying rim materials, rim depths, spoke configurations, aero options, and a range of weight differences (usually in the 1 pound range, except for the lightest of the light). Let us test, then, the following hypothesis: If I have a lighter wheelset, then I will go faster.

Bad science

If you, dear reader, would prefer not to wade through Karl Popper’s dense tome on the philosophy of social science, your author can strongly recommend the book by Guardian columnist and medical doctor Ben Goldacre titled Bad Science. In the book, Goldacre provides numerous examples of science done badly, and how we come to rely on the conclusions of bad science, ultimately to our detriment.

Bad science in bike testing is not hard to find. In a test of wheels in the March 2011 issue of Bicycling, several wheelsets were subjected to “systematic testing” (note the imprimatur of rigour by saying that the testing was systematic, as opposed to random or ad hoc). In reviewing the performance of tubular versus clincher tyres, for example, the testers reached the following conclusion: “Empirical testing may reveal that there is little, if any, quantitative difference between the performance of similar tubular or clincher tires [sic], but anyone in tune with his or her bike will still feel the difference.”

Popper would be dismayed at such a conclusion and any friend of science should be too. The concession is made that there is no difference that can be measured in performance, but that testers could ‘intuitively’ feel a difference – they could feel something that might not actually be there. Two comments arise from this. Firstly, for this to have been a more objective test, it would have needed to have been a ‘blind’ test; that is, the test riders should not have known which type of tyre they were riding (presumably this was not done). Such blindfolding is necessary for a test to be objective, otherwise we are influenced by our expectations (“tubulars feel different, everyone knows that”). Secondly, as numerous scientific studies have shown, we are very bad at making objective judgements. Many of the observations we make are highly subjective and are influenced by external factors (particularly our social environment and perceived wisdom) and internal factors (our own beliefs) – this is how our brains work, and there is not much we can do about it. This is why we perceive things that are not there, and how we fool ourselves (wishful thinking, for example). This is why scientific testing, proper science, requires all sorts of safeguards to prevent our so-called intuition from getting in the way.

Good science

If we are to test our proposition about lighter wheelsets, we can turn to some good science (Popper’s caveats on falsification aside, for now) to help us test our hypothesis. The main forces that a cyclist needs to overcome are wind resistance and gravity (and a small amount of rolling resistance). In general, the flatter the road and the higher the speed, the more importance that wind resistance plays, hence the focus on aerodynamics for time trialling; the greater the gradient and the lower the speed, the more importance that weight plays – particularly for climbing and for accelerating from slow speeds. The excellent website www.analyticcycling.com has some great tools for calculating the forces on riders, and the impact of certain changes, particularly for wheels. Without using examples just yet, it is clear that it is not just a case of lower weight equals greater speed because there are a number of variables in play, and that these variables are important, notably wheel weight, rotational inertia, and drag coefficient (aerodynamics).

When looking at wheels, the author of the site, Tom Compton, cites a scientific paper on the aerodynamics of bicycle wheels that reaches the following conclusions. The total drag of wheels is in the 10-15% range of total drag on a bike and rider. Different wheels choices can give improvements of up to 25%, so 2-3% of the total drag – which is not a very large figure. Also, the rotational drag of different wheels does not change with speed or with different wheels. We can conclude, therefore, that aerodynamics matters, and can be improved with different wheels, but that the magnitude of possible gains is relatively small.

A more interesting example with wheels is acceleration, where wheel weight and rotational inertia are other important considerations – although certainly not the only considerations when calculating acceleration rates. To make these calculations accurately we need a bewildering array of information: effective frontal area, drag coefficient, bike weight, rider weight, wheel weight and inertia and drag, power output of the rider, air resistance, gradient of the road, and a coefficient of rolling resistance depending on the road surface. According to the necessary calculations, in a 250 metre sprint for example, where two riders are accelerating all out, starting at 36 kph, a rider with wheels that are a pound lighter, and all other forces being equal, would be faster by less than a bike length. In fact, in a sprint that starts at a relatively high speed, the more important factor is not wheel weight but aerodynamics.

A more interesting case is what Analytical Cycling calls the ‘criterium corner’, the idea that accelerating out of a low-speed corner will be assisted by lighter wheels. This is true, according to the model, as a one-pound lighter wheelset (or bike, for that matter), all other factors being equal, will give an advantage of around a tyre width or two after 100 metres of acceleration (lower rotational inertia can also compound this effect; lower weight wheels typically have lower inertia as well, but this depends on their design; finding out and comparing inertia rates is a difficult proposition).

The example of the long climb (5 kilometres at 8%) gives more of an advantage to lower weight. All factors being equal, a lower weight of wheel can save real time over a heavier weight. Aerodynamics is still a factor, but it plays less of a role due to lower speeds. Indeed, on this long climb example, the weight of the bike, rider and wheels is relatively important – a 1 pound saving of either bike or rider weight equates to roughly 7 seconds or 30 metres in distance. (Throwing away that full drink bottle at the start of the climb could win you the race.)

So, using the good science and crunching all the numbers we can conclude that wheel weight does matter. A lighter wheelset will enable you to go faster. As we have seen, the gains range from small (winning a sprint, accelerating out of a corner) to relatively large (a long hill climb). But these need to be kept in perspective. In both the case of the criterium corner and the hill climb, producing just 1% more power is enough for the advantage to be offset (you can run all the numbers in the models on the website). There are cumulative effects, to be sure (multiple climbs, multiple corners) in a ride or race but we should be careful about ascribing a too generous of an impact to a factor that is a small part of a much larger and more complex calculation. And maybe we should leave the last word to Cervélo race engineer Damon Rinard, who is somewhat interestingly quoted in the same Bicycling magazine review test premised on the value of lighter wheels, who said: “Heavier wheels aren’t the big performance disadvantage most riders think they are. I’ve done the math, and although rotational inertia is real, it’s tiny.”

Jens overcoming rotational inertia

Back to Popper

Returning to our hypothesis, we have a tentative answer: yes, a lighter set of wheel will enable me to go faster. Although perhaps we should say that lighter wheels have the potential to allow me to go faster, all other factors being equal. We have learned, though, in many riding situations, that aerodynamics matters more than we think. Which is why Bicycling magazine is correct in its conclusion that aero wheels are “the fastest overall option for most riders”. But before one rushes out to purchase their favourite choice, the ZIPP 404 at US$2,700 a set, we should pause to reflect not just on the small relative gains to be achieved but also on Popper’s concerns.

To know what will happen in the future, we have to know the present. The key factor in the conclusion to our hypothesis is that all the other factors in play from physics in our speed equation must be equal – and there are a lot of other factors. We have dealt with a hypothetical example, but in the real world it would be extremely difficult to calculate all those factors for any given speed. Our ability to ride at, say, 36 kph is determined by all these factors having a number that we can punch into the equation. We may be able to calculate our wattage (using a power meter) but the drag coefficient of our wheels, their inertia, and the drag coefficient of ourselves as the rider would be highly difficult.

The problem is this: if we do not have a value for each of the factors required to calculate all the forces causing us to go at the speed we are going, or inhibiting us from going faster, we cannot change just one of those variables and be able to know the impact that it is having in a future scenario. We also cannot know by how much these variables have changed in a future scenario. If I ride faster tomorrow with a lighter set of wheels, I cannot know if this is the factor that has allowed me to ride faster. In a Rumsfeldian epistemology, it is a known unknown – we know that we cannot know this. As Popper warned, if we cannot describe the present, we cannot describe the future either.

As riders, we (mostly) know this to be true. We know that we ourselves are the biggest variable. From day-to-day we produce varying amounts of power, weigh more or less, and ride in a different aerodynamic position on the bike depending on circumstance or how we feel (and carrying an extra full drink bottle may offset a one-pound saving in bike or wheel weight). Because these factors are changing all the time, we cannot measure them accurately like we can with bike and wheel weights (and possibly drag coefficients). This is why our hill climb times on the same hill and on the same bike vary considerably, why sometimes we sprint past they guy on the $3,000 carbon aero wheels at the local crit and sometimes we get spat out the back of the pack as soon as the pace kicks up. It is also why we should at least keep our critical faculties engaged when reading product review articles that say a wheel is too heavy for racing or a bike to heavy for keeping up in the group ride. In terms of epistemology, we are unable to know enough to make that conclusion, and we should challenge these pet theories.

Perceptions and the placebo effect

But this is not the end of the story. Also in Ben Goldacre’s book is an excellent discussion of the placebo effect. In medicine, a placebo is when you are given a treatment (such as a pill) that has no medicinal ingredients but which you are told will help you and that indeed you find that there is a benefit (typically subjective) to your condition. Much of the effect of the placebo is due to social conditioning and to expectations, such as belief in medicine and authority.

The placebo effect is particularly good in subjective cases, such as pain, depression, or minor ailments (that heal themselves anyway, which is why homeopathy ‘works’). Thus, we can say that a lighter bike might have a placebo effect. We pick it up, feel the relative lightness of our new wheelset, and when we ride it we expect it to go faster. This is how we have been conditioned. And because the feeling of speed is subjective, we may ‘feel’ that we are going faster.

The controversial aspect of the placebo effect is whether there is actually any measurable or objective effect from it, an actual therapeutic process. In one study, referenced on www.badscience.com, Goldacre’s website, the placebo effect produced actual and real exercise health benefits in a group of workers who were educated about the effect of the exercise they were doing. They actually got measurably fitter, despite doing the same amount of exercise as the control group. Thus, there was a measurable effect. One study does not make a universal law (neither will multiple studies, if you follow Popper to his logical conclusion) but the suggestion is, therefore, that if you feel faster on the bike, and expect to go faster, you may end up actually going faster than the potential gains from the weight change – your brain makes you work harder to achieve that speed increase.

In a recent interview with Bicycles Network Australia, Jens Voigt was asked about bike technology. He said: “I swear, in the first year on the Cervélo, with the wheels and everything, I felt like sitting on a sailing boat. You don’t pedal and this bike moves. You are just looking at the other ones around you and are going ‘Oh, you poor thing, you’ve got no chance, you’re already beaten’.” Perception is important. Feeling faster is a good thing – and it can be fun (although very light wheels can feel sketchy, at least for your author). Feeling like you are faster on the hills because of the bike, or quicker to accelerate in a sprint, will improve the enjoyment of your ride. Maybe it can even make you go objectively faster.

The last word

In conclusion, therefore, we should be very cautious in ascribing effects to certain variables that are a) difficult to measure and b) may be less important to other variables that are also difficult, if not impossible, to measure. Ultimately, our own capabilities and characteristics are the biggest variable: we weigh around ten times as much as our bike, we are the biggest factor for wind resistance, and very small variations in power have disproportionate effects on performance.

We might reflect, too, on the conclusions of Heinrich Haussler, after his disappointment at this year’s Milan-San Remo when he was gapped on the Poggio from the winning group, despite being in a relatively good position. “I felt good the whole day and on the Cipressa even played with the idea of attacking,” he said, “but 200 metres before the top of the Poggio it was over. I just could not follow the decisive attack.” When we have trouble keeping up on the group ride, or in the race, if we are really honest we know that the problem was not really with our wheel choice but with our own performance. On the other side of the coin, when we blitz past the guy at the local crit who has a bike that cost three times as much as our own, we don’t look down at our own bike and wonder how it was possible, we know that we were the better rider on the day.

A lighter bike feels good to ride, of that there can be little doubt, and we can enjoy other (possibly highly subjective) traits such as quick handling. As such, this is not an argument against light bikes or expensive wheels, just a plea of critical thinking and for realistic expectations. For beyond the subjective we have the objective; according to the epistemology of speed outlined above, the realm of the objective, now and in the future, cannot be known. Ultimately, while we may want to go faster, it is not all about the bike, it is mostly up to us.

Haussler winning in Qatar

11 thoughts on “An epistemology of speed

  1. Great stuff, well written. While it is easy to ‘blame the bike (and parts of..) it is probably just denial that most amateur cyclists (99% of us) should just loose some body weight and train more.

    I am also sceptical of reviews. While I read them to get general advise it’s often hard to overlook the fact that on the next page is an ad for same produce. Saying that wheels are the No1 place to start when looking for parts to perhaps gain some advantage…. but ar $2k I am not sure I need it that much!

  2. This is why, after 36 years in the sport of competitive cycling, I still race on hand built 32 hole, 3x wheels I make myself. With a good pair of tubulars glued on well, I am as fast (generally) as those riders who coughed up close to $2,000 for their wheels. The only real difference between now and then for me (other than age!) is that I now ride a carbon Cervelo R3, and I love the feel of the bike. Subjective? To be sure.

    This was a great post and really did the job on debunking bicycle adds and “reviews”.

  3. The science is little more complicated than the bare physics of the analytical cycling website, in that particularly at low speeds and low cadences like those found on a very steep hill, the bike does not roll at even speed, which is a key assumption in their calculations – it slows as the cranks go through 12-6 on the clock, and accelerates again at 3-9, so those lighter wheels are constantly helping, especially for those of us who are fat, square-pedalling mashers. Also, those aero wheels definitely catch crosswinds and there is a sail effect which reduce power input for a given speed. The teams know of this, they wind tunnel test, and they keep

  4. . . .the results to themselves.

    But yes, the overwhelming effect is always going to be the rider – aerodynamically and in terns if power output.

  5. I ride steel with standard hand built 32 spoke wheels and a brooks saddle. I have often crushed people on bikes much lighter and more aerodynamic and they have often crushed me. My bike never loses a single race, but I certainly do.

  6. Magnificent.

    Because cycling has an important technical (equipment) aspect, temptation to over-analyze that aspect is great, and only augmented by the need for additional profits and hence a media machine fed on ever-”better” things to buy.

    Odd that with all of this new stuff, we’re within a few percent of EddyM’s speeds and records in quite a few areas even today. Adding up all the theoretical gains should mean that we’re twice as fast as Eddy was then. . but no.

    At least we can take comfort in that the paucity of analytical thought, and especially in sound logic, is not unique to cycling: it is a world-wide affliction.

  7. Good article. I actually read the Bicycling article you quoted about the bike being too heavy to keep up on a quick group ride and remember scratching my head a bit. I wonder, “all things being equal”, how many pounds of mass would be negated the aero cost of a a 1cm increase in head tube length, or perhaps a looser fitting jersey for that matter.

    The bike geek in me loves the technology in making a sub-1000 gram wheelset, but as a scientist, there is always a little voice in the back of my head saying “wouldn’t it be cheaper to just drop a few hundred grams by taking a piss.”

  8. You omitted to mention the effect that clothing has on speed. I’m sure that any cyclist wearing the latest Pro Team lycra purchased for a few hundred bucks will definitely out perform the poor hack wearing the $20 discount jersey and knicks. Ditto for helmets, gloves, shoes etc.

  9. “In a Rumsfeldian epistemology, it is a known unknown” – ftr the origin of this language is the Catholic philosopher Bernard Lonergan, not Rumsfeld.
    I never think of either of them while racing.

Comments are closed.