The Verification Venue · one hump, three ways to stop
The Advice That Rusted Onto the Pedal
Pump the brakes in a panic and, on dry pavement, you can roughly double your stopping distance. The old advice was right for the car it was born in and is wrong for yours. Braking grip rides a single hump, the tire-road mu-slip curve, which peaks near 15 percent wheel slip and falls off toward a locked wheel. ABS parks each wheel near that peak 10 to 15 times a second while it keeps rolling and steerable; a person who pumps fully releases the pedal, braking at nothing for half the stop. Below, the Grip Bench lets you race all three and watch the pump lose, then flip the surface and watch the ranking invert.
Here is the shape the whole page hangs on. The friction a tire can make while braking is not fixed: it depends on slip, how fast the contact patch is sliding relative to the road. Plot the grip coefficient μ against slip from 0 (free-rolling) to 100 percent (fully locked) and you get one hump: grip climbs, peaks at roughly 10 to 20 percent slip, then falls off toward a lower sliding value as the wheel locks. A locked wheel is on the far, lower shoulder of that curve, and it cannot steer. Everything about how to stop is a fight to live near the top of this hump. Drive the bench and watch where each strategy sits.
The Grip Bench
one mu-slip curve, wired to a three-car braking raceWhere each strategy lives on the μ-slip curve
The race: released together from the hazard line
| Strategy | Avg decel | Distance | Time | Grip captured |
|---|
The third car's stop, brake phase (colour) vs coasting (grey), across the whole stop time
–
Stopping distance grows with the square of speed, so this slider moves every bar a lot.
Sets the peak and sliding grip. Loose gravel adds a plow term to locked wheels: the special case where the fossil advice still holds.
How many full pedal cycles per second. A person manages maybe 1 to 3.
The fraction of each cycle the pedal is actually down. This is the number that decides the pump, and it is an assumption, so drag it.
Naive pumping locks the wheel to the sliding shoulder, then fully releases. Threshold braking never releases: it holds just shy of lockup, near the peak.
ABS hunts around the top of the hump, not exactly on it. This is the honest fraction of peak grip it holds.
Start on Dry and press Hazard. The pump car sails past the other two while its timeline flickers grey. Then drag the duty cycle, flip the third car to expert threshold, and finally switch the surface to loose gravel and watch which car wins change.
The order you did not expect
On dry asphalt from 60 mph the three strategies do not finish in the order the folklore implies. The naive pump is worst, not best, because a pedal that fully releases twice a second brakes at zero for about half the stop. A fully locked skid is actually the middle: it brakes continuously, just on the lower, sliding shoulder of the curve, so it stops only about 8 percent longer than ABS, but it has thrown away all your steering. And ABS is shortest, because it lives near the top of the hump the whole time and keeps the wheel rolling. So on pavement the true ranking is ABS < locked < naive pump, and the last of those is roughly double the first. That is the whole reason the modern instruction is stomp, hold, and steer.
The advice was not stupid. It outlived its car.
First: pumping was right for the car it was born in
On a pre-ABS car a panic stomp locks all four wheels. A locked wheel loses steering and sits on the lower shoulder of the mu-slip curve, below the peak. Pumping was the layperson's crude way to keep intermittently unlocking the wheels, recovering both their grip peak and their steering. It was a real technique for a real problem. The bench shows the crude version losing badly, because a full release means zero braking for half the stop. But flip the third car to expert threshold and watch it nearly catch ABS: the skilled pre-ABS technique was never to pump, it was to hold the brake just shy of lockup, right at the peak, and never let go. The folk word pump is a garbled memory of threshold braking.
Second: ABS is a steering device, not a shorter-distance device
The headline that ABS stops you shorter is a bonus, not the point. What ABS actually buys is a rolling, steerable wheel. A locked skid on dry pavement stops only about 8 percent longer, but you cannot steer it around the thing you are braking for. ABS trades those few percent of ideal distance for the ability to swerve. That is why the honest verdict is stomp and steer, not stomp to stop shorter. On dry pavement it happens to also be shortest. That is luck, not the mission.
Third: the surface flip, where the fossil still holds
Switch the surface to loose gravel or deep snow and the ranking inverts. Now the locked car stops shortest. A locked wheel plows a wedge of gravel or unpacked snow ahead of itself, and that heap of material is an extra anchor a rolling wheel never builds. NHTSA measured this on the test track: on loose gravel, ABS lengthened straight-line stopping distance by an average of 27.2 percent, precisely because it keeps the wheels rolling and refuses to plow. So the fossilized advice is not universally wrong. It is right car, wrong car: correct for a non-ABS vehicle, and correct on loose surfaces, where even a modern car with ABS trades a little distance for the steering you would rather keep. Check your owner's manual for a non-ABS classic, and remember that the thing ABS is really handing you is the steering wheel.
So the single number, pumping roughly doubles your distance, is true on dry asphalt with a naive full-release pump. The honest completion is that it depends on the surface you are on and on whether you release the pedal or merely modulate it, and that what ABS actually gives you is not a shorter stop but a steerable one.
The four ways to stop, side by side
| ABS | Locked skid | Naive pump | Threshold | |
|---|---|---|---|---|
| Where on the μ-slip curve | Near the peak, hunting | Sliding shoulder (100% slip) | Swings 0 to 100% slip | Held near the peak |
| Wheel state | Rolling | Locked | Locks and frees, cycling | Rolling, on the edge |
| Can you steer? | Yes | No | Only in the off-phase | Yes |
| Dry, 60 mph distance | shortest (about 45 m) | about 49 m (+8%) | about 94 m (about 2.1x) | about 46 m (nearly ABS) |
| Loose gravel ranking | about 27% longer | shortest (plows a wedge) | still worst | longer than locked |
| Who should use it | Any modern driver: stomp, hold, steer | Nobody by choice | Nobody: the misremembered technique | Skilled non-ABS driver |
The check: every distance recomputed in front of you
Nothing here is stored. For your current settings the page recomputes each stop from the same model the offline verifier uses, live:
The offline gate recomputes all of it and exits 0 only if every check passes: node research/should-you-pump-the-brakes/verify-should-you-pump-the-brakes.mjs.
Every free choice and its uncertainty. The mu-slip curve is one representative single-hump model (rising parabola to the peak, quadratic fall to the sliding value) with its peak fixed at 15 percent slip; real curves vary in shape and peak between 10 and 20 percent slip. The peak and sliding grip for each surface are the cited hpwizard SAE-table anchors (dry 0.90 / 0.75, wet 0.60 / 0.50, packed snow 0.20 / 0.15, ice 0.10 / 0.07); these are ranges in the real world, not constants, and tire, temperature and wear move them. The loose-gravel peak (0.55) and slide (0.45) are a representative loose surface, and the extra plow deceleration on locked wheels is calibrated so ABS runs 27.2 percent longer than locked, matching the NHTSA test-track figure; it is a lumped stand-in for a messy real effect, not a measured force. The pump frequency (default 2 Hz) and duty cycle (default 50 percent) are modeling assumptions, not measured human constants, which is why they are sliders: a person pumps maybe 1 to 3 Hz. The pump distance integrates the on/off cycle starting on a stomp, so the exact metres shift a few percent with where in the cycle the panic begins; the direction, that a full-release pump is far worse than ABS on pavement, does not. ABS efficiency (default 90 percent of peak) is honest about ABS hunting around, not exactly on, the peak. The ABS cycle rate of 10 to 15 pulses per second is the canonical Bosch figure; the occasional "100 times a second" claim is not supported for typical passenger systems and is not used here. Idealizations: no reaction time, no brake-bias or weight-transfer detail, no aerodynamic drag, one representative vehicle, and mass cancels out of every distance because it appears on both sides of the equation of motion.
What is exactly true here, and what is a model
Exactly true (the sourced physics). Braking grip follows a single-hump mu-slip curve that peaks at low slip (about 10 to 20 percent) and falls toward a lower sliding value at a fully locked wheel. ABS modulates brake pressure about 10 to 15 times a second to hold each wheel near that peak while it keeps rolling and steerable (Bosch; NHTSA). NHTSA consumer guidance is to step on the brake hard once, hold it, and steer, and to avoid pumping even if the pedal pulsates. On very soft surfaces such as loose gravel or unpacked snow, ABS can lengthen the stop, and NHTSA's test-track evaluation measured an average 27.2 percent increase on loose gravel. Stopping distance for a constant deceleration is d = v² / (2a) from the work-energy theorem, with a = μ g.
A model, not a measurement (the numbers). The exact metres depend on the curve shape, the surface grip values, the pump timing and the ABS efficiency, all of which are ranges. The page picks one honest, cited setting for each and lets you move it. The orderings and directions are the robust result: on paved surfaces ABS is shortest and a full-release pump is worst; on loose surfaces a locked wheel wins because it plows. Those flips do not depend on the specific numbers.
Named simplifications. No driver reaction distance (this is braking only, not total stopping distance); one lumped vehicle with no front/rear brake bias or load transfer; no aerodynamic drag or engine braking; the loose-surface plow is a single calibrated deceleration term rather than a modeled wedge of soil; and the pump is an idealized square on/off cycle. None of these change the two reversals at the heart of the page.