Drafting in the peloton

Why does a group of riders who all refuse to lead travel slower than any one of them alone?

A rider alone in the wind is doing the hardest job in the sport. Gather twenty riders and each of them can do that same job for a fraction of the price, which is why a working bunch runs down almost everything that tries to escape it.

And yet there are moments in every race when twenty fresh riders, every one stronger than the single rider up the road, visibly lose ground to him. Nobody is tired; nobody has crashed. The group has simply become slower than any of its members alone. What makes a collection of strong riders worse at riding fast than the weakest of them?

Start with the air, because at racing speed almost nothing else matters. Rolling resistance is small on the flat and gravity is neutral; what a cyclist mainly fights is the atmosphere, and the power that costs rises with the cube of speed — go ten per cent faster and you are paying about a third more. The rider at the front pays that bill undiluted.

Behind him the air has already been disturbed and partly dragged along, so a following rider meets far less of it. Second wheel is worth a substantial cut; a computational study of a full bunch of well over a hundred riders found that those buried in the middle met something like a tenth of the drag of an isolated rider — a much larger effect than older wind-tunnel work on short lines had suggested. Take the figure loosely, since it depends enormously on packing, crosswind and speed, but the direction is not in doubt.

Now notice the shape of what the front rider produces. Shelter is made by one person, consumed by everyone behind, and cannot be consumed by its maker. That is the ordinary free-rider problem, and if that were all, we would predict a group that under-supplies leadership and rides a little below its potential. It would not predict a group slower than one person alone. Something stronger is needed.

So ask a sharper question: what is a racer maximising? Not arrival time — nobody is paid for the clock. The payoff is ordinal: where you finish relative to the others. Re-examine a turn at the front under that accounting. It costs me energy, yes. But it also saves my rivals energy, at the same moment, in the only currency that decides the result. Leading is not an unrewarded contribution to a common good; it is a direct transfer to the people I am trying to beat.

That changes the standoff entirely. Refusing to lead is not laziness to be scolded out of riders; it is the dominant move. And if going forward is always wrong, what does whoever is nominally at the front do? Slow down, until someone comes past. Everyone reasons identically, so the pace does not settle at "slightly under-provided" — it collapses. This is a war of attrition over a position nobody wants, and its resting point is a crawl. The purest form is on the track, where two match sprinters come to a literal standstill, balancing motionless, each refusing to lead the other out.

And the rider up the road? He has changed games. Alone, there is nobody to shelter and nobody to hand an advantage to, so his payoff is his clock against theirs — a plain, cardinal quantity, and he rides as hard as he can. That is the whole inversion: he is solving an arithmetic problem while they are solving a strategic one against each other.

What ends it? Only something that converts the payoff back. A team does it structurally: a domestique is paid on his leader's result, so energy spent at the front costs him nothing he was ever going to win. A shared threat does it tactically — once the chasers agree the escape is the real danger, they form a rotating paceline, an institution with turn-taking and sanctions, since a rider who skips turns is shouted at, boxed in, refused shelter later, and remembered next season.

First, the peloton is not slow. A committed bunch is faster than any solo rider by a wide margin, which is exactly why breakaways are usually caught. The aerodynamics never changed between the fast case and the stalled one — only the payoff did. Physics sets the size of the prize for cooperating; the scoring system decides whether anyone wants it.

Second, the collapse is rarer and briefer than the vivid image suggests. The clean examples are small chase groups and track sprints, not a full field. And the front rider is not in pure loss: a rider close behind slightly reduces the leader's own drag too, by a few per cent.

Third, this is a cousin of the commons problem that empties a fishery, but not the same animal. A fisherman who restrains himself gains nothing, yet loses only what he did not take; his payoff is cardinal. A rider who leads actively improves his rivals' position, so cooperation here is worse than merely unrewarded — which is why the peloton's failure is sharper than a commons, and why it is fixed by rewriting whose result counts rather than by rationing anything.

• Why team time trials, on identical aerodynamics, produce flawless cooperation.
• How echelons form in crosswinds, making the road itself a rationed good.

Every term the collection defines is gathered in the glossary.