Carbon plate super shoes explained: what they actually do

Stand at the elite pen of a big city marathon and look down. The runners vary. The shoes barely do. Every foot is in something tall, curved, brightly coloured and stiff, and has been for several years now.

Carbon plate super shoes explained honestly turn out to be three things working together rather than one clever part. There is a thick slab of compliant, high-rebound foam. There is a stiff curved plate buried inside it. And there is the shape the two of them force on the sole, which is the component almost nobody discusses and quite possibly the one that does the most.

The sentence you will hear in shop windows and in television commentary is that the plate acts like a spring. It does not. That explanation is wrong in a specific and instructive way, and once you see why, everything else about the shoe falls into place.

Carbon plate super shoes, explained in three parts

Take the shoe apart conceptually before taking it apart physically.

The first component is the midsole foam. It is thick, it is light, and it is made from a polymer that deforms a long way under load and gives most of that deformation back. That combination is the hard part. Foams that are soft are usually not resilient, and foams that are resilient are usually not soft. The chemistry that produces both at once, and at a weight that does not punish you for carrying it, is the genuine technical achievement here.

The second component is a stiff plate, usually carbon fibre, moulded with a curve and buried in the foam. In most designs it sits closer to the ground under the forefoot and rises through the middle of the stack. It is not a flat sheet and it is not centred.

The third component is geometry. A thick stack has to be shaped or you cannot run in it, so the sole is curved along its length, most sharply at the front. That curve is called a rocker, and it changes how the foot travels over the ground during the fraction of a second it spends there.

None of the three works alone. A foam that thick and that soft without a plate is unstable and mushy, and the foot sinks in without getting much back. A plate on its own, in a conventional thin shoe, has been tried repeatedly over decades and does very little. The rocker without the stack has nothing to rock over. The reason the whole class of shoe arrived at once, rather than gradually, is that all three had to be present before any of them paid.

That is also why the phrase "carbon plate shoe" is a bad name for the object. It names the least important of the three components. Manufacturers were happy enough with that, because a black woven plate photographs better than a slab of foam.

Why the plate is not a spring, and why that matters

A spring stores energy by deforming. Push on it, it bends, and the energy you put into bending it is held in the strain of the material until it is released. The amount stored depends on how stiff it is and, importantly, on how far it moves. A very stiff object that barely moves stores almost nothing, no matter how much force you apply.

Now consider the plate. Its job description in the marketing material is to bend under load and snap back at push-off, catapulting the runner forward. If that were true, it would need to bend a long way. A plate that bends a long way is a flexible plate, and a flexible plate cannot do the thing the plate is actually there for.

Measure the bending that happens inside a real shoe during stance and it is small. Much smaller than the compression of the foam above and below it. The foam is deforming by a large fraction of its own thickness. The plate is deflecting by a fraction of a millimetre in places. Whatever energy is being stored and returned in that sole, the overwhelming majority of it is in the foam.

There is a further problem with the spring story, and it is the one that finishes it off. A spring returns energy at the moment it unloads, on its own timetable, not on yours. For a stored-and-returned mechanism to help a runner, the return has to be timed to arrive during push-off, in the direction of travel, at the point in the stride where it can do useful work. A stiff plate embedded in foam has no mechanism for choosing when to release. It releases when the load comes off, which is when the foot is leaving the ground anyway.

So the plate is not a spring. It is a stiffener. It changes the shape of the shoe under load and it changes where the load goes. That sounds less exciting than a catapult, and it is worth about ten times as much.

What the foam actually does, and what it cannot do

Every foam is lossy. Compress it and release it and you do not get back what you put in. The difference comes out as heat, and the technical name for that gap is hysteresis. A tennis ball is a high-hysteresis object, which is why it does not bounce back to the height you dropped it from. A superball is low-hysteresis, which is why it nearly does.

Running shoe midsoles have always been lossy in the tennis-ball direction. Traditional foams absorbed impact well, which is what they were sold for, and gave back a modest fraction of the energy they took. That was accepted as the price of cushioning, and the alternative offered to serious racers was to remove the foam almost entirely: thin, hard racing flats that returned little because they stored little.

The newer polymers changed the trade. They deform a long way, which is the precondition for storing a lot, and they come back a long way, which is the precondition for returning a lot. Neither property alone would be enough.

Thickness then multiplies the effect. A thicker slab of a given foam compresses further under the same load, because there is more material to compress, so it stores more energy per step. This is the plain reason stack heights climbed as soon as anyone worked out how to make a tall shoe runnable. It is also the reason the rules ended up being written in millimetres of sole thickness rather than in some measure of springiness: thickness is the variable a referee can check, and it is a decent proxy for how much energy the sole can hold.

Two things the foam does not do, both of which get claimed for it.

It does not return more energy than was put in. Nothing does. A shoe that gave back more than it received would be a perpetual motion machine, and the reason a runner goes faster in these shoes is not that energy is created but that less of it is wasted.

It does not make the running free. The runner still has to generate every joule of forward motion. What changes is the fraction of that effort which ends up as heat in a shoe, heat in a tendon, or motion in a joint that goes nowhere useful. Reducing waste is a real advantage and it is not the same advantage as being pushed along.

The toes are where the energy quietly disappears

Here is the part of the mechanism that almost never gets explained, and it is the reason the plate exists.

Run through a stride slowly in your head. As the heel comes up and the body passes over the foot, the load moves onto the ball of the foot. The joints where the toes meet the rest of the foot, the metatarsophalangeal joints, are now carrying most of your body weight through a small area, and they bend backwards under it. That bending is called dorsiflexion, and in a normal shoe there is a lot of it.

Bending those joints costs energy. The muscles and tendons crossing them resist the bend, absorbing work as the joint gives way. And here is the problem: they give almost none of it back. Unlike the ankle, where the Achilles tendon behaves like a decent elastic store, the toe joint is close to a one-way street. Energy goes in during the second half of stance and very little comes out.

It is a small leak per step. Multiply by the number of steps in a marathon and it stops being small.

The plate attacks that leak directly. A stiff plate running under the forefoot resists the joint bending, because the joint cannot bend much unless the shoe under it bends too. Less bending means less energy absorbed and lost at the toes. The foot behaves, mechanically, more like a rigid lever and less like a hinge that eats work.

Stiffening the forefoot does something else at the same time, and this is where the picture gets more interesting. Increasing the stiffness of a shoe's forefoot moves the point about which the foot effectively pivots at the end of stance further forward, out towards the tip of the shoe. A longer lever arm at the ground changes the demand on the calf muscles and the Achilles: they have to produce force over a longer moment arm, which alters how fast the muscle fibres have to shorten. Muscle is more efficient at some shortening velocities than others, and shifting the joint into a better part of that curve can reduce the metabolic cost of the same mechanical output.

That is the honest version of the mechanism, and note that it has nothing to do with springs. Stiffen the forefoot, stop the energy sink at the toes, and change the leverage under which the calf works.

Note also that it is not free. Push the lever arm too far forward and the demand on the calf goes up rather than down. Different runners, with different foot lengths and different ankle stiffness, sit at different points on that curve. That is your first clue as to why the same shoe helps one athlete and does nothing for another.

1. 1The foot arrivesThe leg lands carrying kinetic and gravitational energy. Some of it is lost immediately to the ground, to soft tissue wobbling and to the collision itself. No shoe recovers that portion.
2. 2The midsole compressesFoam deforms under the load and holds energy as strain in the material. A thicker, more compliant slab deforms further at the same force, so it holds more.
3. 3Hysteresis takes its cutNo foam gives back everything. The gap between energy in and energy out leaves as heat. High-rebound polymers narrow that gap. Nothing closes it.
4. 4The plate spreads the loadThe plate bends very little. What it does through mid-stance is distribute pressure along the length of the sole instead of letting it concentrate under one point of the foot.
5. 5The heel lifts and the load moves forwardBody weight transfers onto the forefoot. In an ordinary shoe the toe joints now bend backwards under that load.
6. 6The toe joint is stopped from bendingA stiff forefoot resists dorsiflexion. Energy that would have been absorbed at the toes and never returned stays in the system instead. This is the plate's real job.
7. 7The lever arm lengthensWith the forefoot rigid, the foot pivots further forward at the end of stance, changing the moment arm the calf works against and the speed at which its fibres shorten.
8. 8The foam unloadsStored strain energy returns as the shoe decompresses. It helps if it arrives while the foot is still pushing. It is wasted if it arrives after the foot has left.
9. 9The rocker carries the roll throughThe curved sole moves the contact point forward so the foot leaves the ground without needing the joint range of motion it would otherwise have to supply.
10. 10The shoe becomes a costThrough the swing phase the shoe is dead weight on the end of a long lever, which is why every gram in the sole has to earn its place.

A qualitative account of the sequence, not a measurement. Each step describes a direction of effect rather than a quantity, because the published measurements of these quantities disagree with one another.

The rocker is doing more of the work than the plate gets credit for

A thick, stiff sole cannot be flat. If it were, the foot would have to pitch over a rigid edge at the front, and the shoe would be unrunnable. So the sole is curved along its length, rising towards the toe, sometimes towards the heel as well.

Watch a foot roll through stance in a conventional flat shoe and the shoe deforms with the foot, flexing at the forefoot as the heel rises. In a rockered shoe the shoe does not flex much. Instead the whole assembly rolls, like a rocking chair leg, with the contact patch travelling forwards along the curve.

Three consequences follow.

The first is that less joint range of motion is required. The rolling that would have come from the ankle and the toes now comes partly from the shape of the shoe. The joints do less work through the last part of stance because the geometry has taken some of it over.

The second is that the shoe brings the runner forward. A rockered sole is inherently a little unstable in the direction of travel. Once your weight passes a certain point on the curve, you are going over the front of it. Runners describe this as the shoe wanting to tip them forward, and the description is accurate. It is not propulsion, it is a controlled fall that the shoe makes slightly easier to initiate.

The third is that the plate and the rocker are the same design decision seen twice. The plate holds the rocker shape under load. Without something stiff inside it, a tall soft sole would simply squash flat where the foot pressed hardest and the curve would disappear at the exact moment it was needed. A large part of what the plate contributes is preserving the geometry, not storing energy or even stiffening the toe joint. It keeps the shape the shape.

The exact curve is where manufacturers differ most and say least. Where the apex of the rocker sits along the length of the shoe, how tight the radius is, how the curve interacts with the heel bevel: those choices decide whether a given shoe suits a given runner far more than the grade of carbon does. They are also almost impossible to evaluate from a photograph, which is convenient for everybody selling one.

Why two runners in the same shoe get different results

Published testing keeps producing the same awkward finding. Put a group of runners in the same shoe and measure their energy cost, and the group mean improves, but the spread around that mean is wide. Some runners improve a lot. Some improve slightly. Some are no better off. A minority are measurably worse.

That is not noise, or at least not all of it. Retest the same runners and the individual responses tend to hold. Some people are non-responders to a specific shoe, and some are non-responders to the whole category.

Several plausible reasons, none of which anyone has fully separated from the others.

Foot length and lever geometry. The benefit from stiffening the forefoot depends on how the resulting lever arm compares with your own foot. Two runners with different foot lengths in the same shoe are not receiving the same intervention.

Where you already were on the efficiency curve. If your calf and Achilles are already operating in a favourable part of their force-velocity relationship, a change in leverage moves you off it rather than onto it.

Speed. A shoe optimised around the loading pattern of a fast marathon runner is being asked to do something different under a runner covering the same ground much more slowly, with lower forces, a different foot strike and much longer contact times. Foam that compresses ideally under one set of loads may barely compress under another.

Rollover pattern. A rocker helps if its curve roughly matches the path your foot was going to take anyway. If it does not, you are fighting it, and fighting a shoe costs energy.

Stability tolerance. A tall, soft platform is a wobbly platform. Runners who need to spend muscular effort controlling side-to-side movement on that platform are spending energy the shoe was supposed to save. This is the mechanism most often proposed for the runners who get measurably worse, and it fits the observation that some athletes find the shoes fine in a straight line and hate them on a twisting course.

The size of the average effect is contested too, and it is worth being blunt about this rather than repeating a number. Different laboratories, different protocols, different comparison shoes and different subject groups have produced different answers, and the figure that gets quoted in the press typically comes from a small study of well-trained runners at fast speeds tested on a treadmill. Whether that transfers to a slower runner on a rolling road course over three or four hours is a genuinely open question that the existing evidence does not settle.

What the evidence does support, robustly, is the direction. On average, in these shoes, running the same speed costs less. How much less, for you specifically, nobody can tell you from a study.

Why carbon plate super shoes matter more in a marathon than on a track

Part of the answer is arithmetic, and it is the uninteresting part.

1 per cent1.8

2 per cent3.6

3 per cent5.4

4 per cent7.2

Constructed illustration. The percentages are inputs chosen to show arithmetic, not measured values, and a metabolic saving does not convert one for one into time saved. Minutes, from an invented baseline of 180 minutes.

Show the numbers

Invented worked example: minutes saved from a 3:00:00 baseline

Item	Value
1 per cent	1.8
2 per cent	3.6
3 per cent	5.4
4 per cent	7.2

Apply a fractional saving to a longer race and you get more absolute seconds out of it. Nothing surprising there, and it explains why the marathon was where the effect became visible to the public before it was visible anywhere else. Spectators do not notice fractions. They notice minutes.

5,000m18

10km37

Half marathon108

Marathon216

Constructed illustration. Baselines of 15:00, 31:00, 1:30:00 and 3:00:00 are invented round numbers and the saving applied is a hypothetical two per cent, not a measurement. Seconds, rounded.

Show the numbers

The same invented saving across four invented race times

Item	Value
5,000m	18
10km	37
Half marathon	108
Marathon	216

The interesting part of the answer is physiological, and it runs the other way.

A marathon is limited by energy. Not by oxygen delivery, which is the binding constraint over shorter distances, but by the cost of covering the ground and by the finite store of carbohydrate available to pay for it. Anything that reduces the cost per kilometre extends how far you can hold a given pace before the store runs down. That is the exact quantity a more efficient shoe changes, which is why the effect on marathon performance is larger than the raw arithmetic suggests. A shoe that reduces cost does not merely make each kilometre cheaper, it moves the point at which the wheels come off. Anyone who has planned a race around that point will recognise the problem from our piece on how marathon pacing actually works.

A marathon is also limited by damage. Two and a half hours of impact degrades muscle function, and the pace an athlete can hold at thirty kilometres is not the pace their aerobic system could theoretically support, it is what is left after the legs have taken a beating. A thick compliant midsole reduces the peak loads reaching the muscle. Less damage at thirty kilometres is worth more than any efficiency gain in the first ten.

A track race is a different problem. It is run closer to the ceiling of oxygen delivery and mechanical power output, over a duration too short for accumulated damage to decide anything, on a surface that already returns energy by design. The metabolic efficiency lever the shoe pulls is simply a smaller part of what determines the result.

And on the track the costs bite harder. Mass on the foot is expensive at any speed and more expensive at high speed, because you are swinging that mass through a longer arc more often per minute. A sole tall enough to hold a useful volume of foam is heavier than a spike plate and a scrap of upper. Track shoes are also asked to handle cornering, and a tall soft platform leaned into a bend is exactly the situation in which the stability cost is worst.

This is why the technical rules treat the two cases separately. The permitted sole thickness for road events is higher than for track events, and lower again for shoes fitted with spikes. That is not arbitrary. It reflects a judgement that the thing being regulated has different value and different risk in different events.

What World Athletics did instead of banning them

The obvious response to a shoe that makes people faster is to ban it. Swimming had done exactly that with polyurethane bodysuits a decade earlier, and there was a loud constituency arguing for the same treatment here.

World Athletics regulated instead, and the reasoning is worth reconstructing because it applies well beyond running shoes.

A ban has to define what it bans. Every shoe assists the runner. A bare foot on a road is slower than a foot in a plimsoll, and nobody has ever argued that shoes are cheating. Once you accept that footwear may assist, you are drawing a line through a continuum, and "too much assistance" is not a natural category. Any line will be arbitrary, so you may as well draw an arbitrary line that is easy to measure rather than pretend to a principle.

A ban would have to be retroactive to be coherent. The technology was already on feet, already in shops, already under performances that had been ratified. Banning forward while keeping the record book intact produces a stratum of unreachable marks, which is precisely the outcome swimming created and has been living with ever since.

A ban would be unenforceable below the elite level. Millions of pairs were already sold. A rule that governs a championship start line but nothing else creates a sport where some of the fastest performances in the world happen at events the governing body does not control.

And there is a funding argument that nobody in athletics likes to state out loud. The shoe companies pay for the sport, from athlete contracts to meeting sponsorship. A governing body that outlaws its principal funders' flagship product is picking a fight it has few resources to survive. This is not a noble reason. It is a real one, and the same dynamic shows up whenever a sport tries to regulate a technical arms race that its own commercial partners are running, which is the tension underneath the way Formula 1 chose to cap spending rather than freeze development.

So the rules do three things instead. They cap sole thickness by event category, which is measurable at a start line with a gauge. They limit the shoe to no more than one rigid embedded plate, and where that plate is made in more than one piece the pieces have to sit in a single plane rather than being stacked to make a de facto multi-plate assembly. And they require the shoe to have been genuinely available to buy.

That third one is the clever part.

1. 1A manufacturer develops a shoeDevelopment happens in private and always has. Nothing in the rules stops a brand building whatever it likes. The rules govern what may be worn in competition, not what may be made.
2. 2The shoe is measured against the sole thickness limitThickness is measured by a defined method at a defined point, with different ceilings for road events, for track events and for shoes fitted with spikes. It is a physical check, not a performance test.
3. 3The plate provision is checkedNo more than one rigid embedded plate or blade is permitted. Where it is made in several parts, those parts must sit sequentially in one plane rather than being stacked or run in parallel.
4. 4The manufacturer submits it for assessmentBrands send shoes to World Athletics, which assesses them against the rules and publishes the outcome. Assessment is the manufacturer's responsibility, not the athlete's.
5. 5It goes on the published listWorld Athletics maintains a public list of shoes cleared for competition. Officials check footwear against that list rather than making a judgement at the track.
6. 6The availability clock runsThe shoe has to have been on open sale to any buyer, through normal retail, for a period defined in the rules before the competition. Announced is not available. Sold is available.
7. 7A development shoe takes the other routeThe rules provide for prototypes not yet on sale, subject to submission, approval and a commitment that the shoe will reach the market. A development shoe that never goes on sale loses its standing.
8. 8Kit is checked at the competitionFootwear is inspected in the call room alongside the rest of an athlete's kit, and the referee has the authority to prevent a non-compliant shoe being worn.
9. 9A record application asks what was wornFootwear compliance forms part of the ratification file for a world record. A performance in a non-compliant shoe can be a real performance and still not become a record.

The sequence set out in the World Athletics Technical Rules. The specific measurements and time periods are published in the rules themselves and are amended from time to time, so they are described here rather than quoted.

The sole availability rule, and the problem it was really solving

Of all the shoe rules, the one requiring a shoe to be on open sale is the least discussed and the most consequential.

Here is the problem it addresses. In the period when the first generation of these shoes appeared, one manufacturer had them and the others did not. Their contracted athletes raced in them. Everyone else raced in whatever their own sponsor made, which was not the same thing. For a stretch, the outcome of major races was partly a function of which logo an athlete had signed with several years earlier.

That is an unusual kind of unfairness. It is not doping, it is not cheating, and no rule had been broken. It is simply that a performance advantage was distributed by contract rather than by preparation.

The availability rule attacks one half of it. If a shoe has to be purchasable by anybody for a defined period before it can be raced in, then no brand can field a secret weapon at a championship. Whatever the athletes are wearing, you could have bought it. The rule converts a private advantage into a public product, and a public product gets copied.

It does not attack the other half. An athlete contracted to one brand may not wear a rival's shoes, however available they are, because their contract says so. The rule guarantees that the best shoe is on sale. It does not guarantee that any given athlete is allowed to put it on.

What followed was a competitive scramble that mostly resolved the problem in a different way. Every serious manufacturer built a comparable shoe, some by licensing foam chemistry, some by developing their own. Athletes renegotiated. A few raced in blacked-out shoes while a deal was worked out. The gap between the best shoe available and the fifth best narrowed to something much smaller than the gap between the best shoe and no shoe at all. The rule helped, and so did ordinary competition among suppliers.

There is a real cost to the availability requirement that gets less attention. It slows innovation by design, because a brand cannot debut something at a championship. It also means the shoe an elite races in is, at least nominally, the shoe in the shop, which has pushed prices in the shop up sharply. A regulation aimed at fairness among elites has made the entry ticket for everyone else more expensive. Rules have side effects, and this one lands on club runners who were never part of the argument.

The fairness argument, taken seriously in both directions

The case against is not stupid and deserves to be stated properly rather than dismissed.

Athletics is unusual among sports in that its primary product is a number. A football result is a story about a match. A marathon time is a measurement, and measurements only mean anything if the measuring conditions are stable. If the equipment changes, the meaning of the number changes, and a record book that mixes eras with different equipment is comparing quantities that were never the same quantity.

The case also points at where the advantage came from. An athlete's training is theirs. A shoe's design belongs to a corporation, and access to it is decided by a commercial negotiation. When performance depends on things athletes cannot control and did not earn, something has shifted about what the sport is measuring. The ranking systems that decide who gets into championships are built on those performances, which is worth remembering when reading how athletes accumulate ranking points.

The case for is equally serious.

Every sport uses equipment and every sport's equipment has improved. Running surfaces went from cinders to synthetic tracks and every distance record improved as a result, with no asterisks awarded. Pole vaulters moved from bamboo to steel to fibreglass, which transformed the event, and nobody proposes separate record books. The javelin was redesigned specifically because throws had got too long for stadiums, and the marks were reset without anyone claiming the previous generation had cheated. Timing moved from hand to electronic and rewrote the sprint lists.

Golf has spent decades running the same argument in public, with governing bodies that write detailed specifications for the ball and the club and revise them when the numbers get away from them. That process, and the way it constrains manufacturers, is the closest parallel available anywhere in sport, and the reasoning behind the way golf regulates its equipment is worth reading alongside the athletics version. Motorsport regulates a technical arms race as its core activity. The precedent for governing bodies writing specifications rather than banning progress is overwhelming.

There is also a fairness argument that runs the other way, and it is rarely made. Before the availability rule, footwear advantages existed but were invisible, distributed through custom-made shoes for a handful of athletes that nobody could buy. The current rules require the equipment to be on public sale. Whatever else you think, the modern situation is more transparent than the one it replaced.

Where the case against is strongest is comparability of individual careers. An athlete whose best years fell just before the technology arrived has a personal best that means something different from an identical time run afterwards. There is no remedy for that. It is not the first time athletics has done it to a generation and it will not be the last, but the people it happened to are entitled to be annoyed.

Where the case against is weakest is the claim that this is somehow like doping. It is not, and the difference is structural rather than moral. Doping is concealed, individually administered, prohibited outright, and detected through a testing system built for exactly that purpose, as our piece on how anti-doping testing is actually run sets out. A shoe is visible, purchasable, checked in the call room and listed on a public register. Whatever problem the shoes present, it is not a problem of enforcement.

What it did to training, and where the evidence runs out

The race-day effect is the one that gets discussed. The training effect may be larger and is much harder to measure.

Start with what changed in practice. Serious runners now rotate shoes by purpose rather than by wear. A durable trainer for easy volume, where the softest foams would be destroyed and are not needed. A lighter plated trainer for workouts. The racing shoe kept for races and a small number of sessions, because the foams that return the most energy tend to be the least durable, and a race shoe has a competitive life measured in a few hundred kilometres rather than in visible wear on the outsole. That last point is a genuine expense that nobody advertises.

The plausible training mechanism is recovery. If a shoe reduces the mechanical damage of a hard session, an athlete can repeat that session sooner. Over a training block, more high-quality sessions completed means more fitness, and that accumulated fitness arrives on race day alongside whatever the shoe does on the day itself. If this is real, the shoes have raised performance through two separate routes, and the studies measuring energy cost on a treadmill capture only one of them.

It is a good hypothesis. It is not established. Nobody has run the study that would establish it, because it would require years, large groups and a degree of control over training loads that no coach would surrender.

The injury question is where the evidence is genuinely thin, and it should be stated as thin rather than dressed up.

What exists is a set of plausible mechanisms and a quantity of clinical impression. The mechanisms: a stiff forefoot moves load away from the ankle and calf towards the midfoot and the bones of the forefoot, and load that leaves one structure arrives at another. A tall soft platform demands more stabilising work and offers less information about the ground underfoot. Athletes now run their hard sessions faster, and faster running is more damaging regardless of what is on the foot. Each of those would predict a shift in the pattern of injuries rather than a simple increase in the number of them.

The clinical impression, reported by physiotherapists and team doctors working with elite distance runners, is that certain bone stress injuries in the midfoot and forefoot appear more often than they used to. That impression is worth taking seriously and it is not evidence. There is no adequate longitudinal data. The comparison group is a period when training was different, volumes were different and reporting was different. Elite runners are not a random sample and their injury records are not systematically published. Any claim that these shoes cause a specific injury, or that they prevent injury by reducing impact, is currently running ahead of what anyone can demonstrate.

Two things can be said with confidence. Athletes and coaches report needing a transition period, and the complaints during that transition cluster around the calf, the Achilles and the foot, which is exactly where the altered mechanics would predict them. And a shoe that lets you train harder increases your training load, which is the single most reliable predictor of injury in any endurance sport, whatever caused the increase.

How to think about the record book now

The temptation is to sort the record book into before and after, and to treat one side as legitimate and the other as inflated. Resist it, for a reason that has nothing to do with shoes.

There has never been a stable baseline. Records were set on cinders and on synthetic tracks, timed by hand and by beam, at sea level and at altitude, with pacing arrangements ranging from illegal to institutional, on courses measured to standards that tightened over decades. Every generation ran under conditions a later generation would consider either quaint or unfair. The record book has always been a list of the best performances under the rules in force at the time, and it has never been anything more coherent than that.

What is different here is only that the change was fast and visible. A track resurfacing programme improves times gradually and nobody photographs it. A shoe arrives in one season, in bright colours, on television.

So read a performance with four questions attached, and they work for any era.

What were the rules at the time? Not just the footwear rules. Pacing, timing, course measurement and altitude allowances all changed and all matter.

What was the field? A time run in a race designed for time-trialling, with pacemakers and a flat course selected for speed, is a different object from a time run in a championship where nobody cares about the clock and the medal decides everything. That distinction was worth making long before the shoes arrived, and it is more useful than any argument about foam.

What was universally available? A performance advantage everybody in the race had is part of the conditions of the race, like the weather. One that only some had is part of the result. That distinction is the actual reason the availability rule exists, and applying it backwards tells you which performances from the transition period deserve an asterisk in your own head, if not in the record book.

What is the number for? A world record is an administrative object with a ratification file and a set of conditions attached. A performance is a thing an athlete did on a day. Conflating the two causes most of the arguments. You can hold the view that a mark is a legitimate record and also that it does not compare cleanly with one set twenty years earlier, because both statements are true and they are not in tension.

The shoes did not break the record book. They exposed something about it that was always true and rarely stated: that it is a record of performances under conditions, and the conditions were never fixed. The rest of our athletics coverage proceeds from the same principle.

The most useful habit is to stop asking whether a time was assisted, because every time was, and start asking what it was assisted by and who else had access to the same assistance. That question has an answer. The other one never did.