Speedonomics

Posted October 30, 2026

Why the fastest way to lose is to floor it — and the dirty secret about the speed limit itself

A data-driven look at what speeding actually buys you, what it costs you, and why the number on the sign is probably wrong anyway.

Every driver runs the same silent calculation at the on-ramp: push it a little, get there sooner. It feels obviously true. The thesis of this piece is that it mostly isn't — and that the small, real savings are swamped by the risk of getting caught, or worse, getting hurt. Then I want to turn the argument on its head and go after the villain nobody suspects: the speed limit itself, which is often set by a method so crude it practically manufactures speeders.

Let's do the math first. Opinions are cheap; minutes and probabilities aren't.

Part 1 — The coast-to-coast fantasy: New York to San Francisco

Take the biggest canvas possible: ~2,900 miles of mostly interstate, New York to San Francisco. Assume a representative 70 mph limit and ask what shaving the clock actually gets you.

At first glance +15 looks great — seven hours! But notice the trap hiding in "pure driving time." Those numbers assume you never sleep, never refuel, never eat, and never slow for traffic or weather across ten states. The moment you reintroduce a human being who needs ~8 hours of sleep and a few fuel stops, a cross-country drive stretches to three or four days regardless of your cruising speed. Your heroic 7-hour "saving" gets diluted into a rounding error against 70+ hours of total trip time — and it's the first thing you'll surrender the instant you hit rain, a work zone, or a wall of brake lights.

Now count the places you can get caught

Ten states. ~2,900 miles. State troopers run radar and LIDAR continuously, and the exposure is cumulative — every patrol you pass is an independent roll of the dice.

Using a transparent, deliberately conservative assumption of ~1.5 active radar/patrol exposures per 100 miles, you pass roughly 44 distinct points where you could be clocked. If each one carries even a modest 3% chance of a ticket while you're sitting at +15 over:

P(at least one ticket over the trip) ≈ 73%.

That's the compounding nobody feels in the moment. Any single trooper is unlikely to get you. Forty-four of them, strung across a continent, probably will. And one stop — 15–20 minutes on the shoulder plus the ticket, plus the insurance surcharge that follows you for three years — erases the entire time "saving" and then some.

Part 2 — The short hop: 20 miles

Fine, you say — nobody speeds to save a cross-country week. People speed on the daily 20-mile run. So let's do that one too, at a 65 mph limit.

There it is. The thing you risk a ticket, your license, and your life for on the daily commute is three and a half minutes — and that's at a reckless +15. Push +5, the far more common cheat, and you've "saved" 79 seconds. You will lose that at a single long red light on the way home.

This isn't just my arithmetic. A peer-reviewed analysis ("Are the time savings worth the risk?", 2015) found the average driver saves about 26 seconds a day by speeding — and that the population pays for those seconds at a rate of one fatality for every 24,450 hours saved on 100 km/h roads in dry conditions. We are trading lives for seconds and feeling clever about it. [@a34756]

Psychologists even have a name for why it feels bigger than it is — the time-saving bias: when we're already moving fast, we systematically overestimate how much more time a speed increase will save. The faster you're going, the worse your intuition gets, in exactly the direction that encourages you to speed. [@e923e5 @7a64d3]

Part 3 — Surface streets: where speeding is pure theater

On surface streets the case against speeding stops being "weak" and becomes "self-defeating," because of one thing: the lights don't care how fast you drive between them.

Model a typical 2-mile urban arterial with 8 signals, roughly a quarter-mile apart, timed as a coordinated "green wave" for the posted 30 mph. Compare a driver who obeys the limit and rides the wave against a driver who floors it to 45 between every light.

• Law-abider (catches the green wave, ~1 stop): 4.5 minutes
• Speeder (outruns the wave, then catches ~5 reds): 5.2 minutes

The speeder is 40 seconds slower. Not tied — slower. Flooring it to 45 saves about 80 seconds of pure motion, and a single extra red light (≈30 seconds) more than eats it. By breaking the coordination the signals were designed around, the speeder arrives at each light before it turns green and simply waits there, engine running, having burned extra fuel and brake pads to go nowhere faster. The aggressive driver who roars past you between lights and whom you then pull up next to at the next red is not an annoyance — he's a data point.

On surface streets, speeding between lights is a performance of urgency, not a source of it.

Part 4 — The physics: what the speed actually buys you (in risk)

Here's the part that should end the argument. The time you save grows linearly and slowly. The danger you take on grows geometrically.

Crash severity is governed by kinetic energy, which rises with the square of speed — the Federal Highway Administration states the relationship plainly: higher operating speeds mean more severe crash outcomes because energy scales with speed squared. [@782440] But the real workhorse is Nilsson's Power Model, the standard in traffic-safety research, validated repeatedly by Elvik and others:

• Injury crashes scale with speed squared
• Serious-injury crashes scale with speed cubed
• Fatal crashes scale with speed to the fourth power [@505596 @18daad]

Run our highway numbers through it, starting from a 70 mph baseline:

Read that bottom row against Part 1. On the daily 20-mile hop, +15 over bought you 3.5 minutes — and more than doubled your chance of dying if anything goes wrong. That is the whole thesis of Speedonomics in two numbers: a trivial, linear reward against an explosive, fourth-power cost. The trade is not close.

Part 5 — The plot twist: the speed limit is probably wrong anyway

So far this reads like a lecture to slow down. It isn't — or not only. Because the number on the sign is itself frequently mis-calibrated, and in both directions. My standing theory (which the evidence largely supports): speed limits are set by static rules when the road is a dynamic thing.

How the number actually gets chosen

The dominant method in the U.S. is the 85th percentile rule: engineers measure free-flowing traffic and set the limit near the speed at or below which 85% of drivers are already going. The logic is that the reasonable majority self-selects a safe speed. [@3f99e3 @cc0a57] It's a crowd-sourced number dressed up as engineering — and by the 1950s–60s it had hardened into regulatory gospel, applied far beyond where it was ever validated. [@ab5e02]

This produces the two failures everyone feels but can't name:

Too slow for the road. When a limit is posted below the 85th percentile — often for political or optical reasons — it doesn't make anyone safer; it just turns almost everyone into a technical "speeder." The FHWA's own Parker research found no safety benefit, and often disbenefits, from limits set below the 85th percentile. [@f91e94] Even a city FAQ puts it bluntly: arbitrarily lowering a residential limit below the 85th percentile isn't effective and breeds broad disregard for the limit. [@93cbce] A limit that the road itself contradicts doesn't get obeyed — it gets ignored, and it erodes respect for every other sign too.

Too fast for the conditions. The same static method cuts the other way. A number set on a dry, clear, empty afternoon stays posted at 2 a.m. in freezing rain, in fog, past a school at dismissal, through a work zone, next to a hospital. The sign is frozen; the road is not. A limit that felt right in June can be lethal in a December ice storm — and the sign has no idea.

The fix is already proven: dynamic speed limits

If the problem is a static rule on a dynamic road, the answer is a limit that moves with the road — adjusting for weather, time of day, congestion, incidents, and zone activity. This isn't speculative; it's deployed and measured:

• Before/after studies of freeway variable speed limits in Georgia and Wyoming estimated crash reductions of roughly 34% and 29%, respectively. [@4e4f17]
• FHWA developed formal Crash Modification Factors for VSLs using data from Virginia, Wyoming, and Georgia, treating them as a genuine safety countermeasure. [@4b5037]
• Dynamic speed-monitoring displays alone cut average speeds by about 7 mph, with the reduction holding for a full year. [@54775c]

A dynamic limit fixes both failures at once: it can rise when a wide, empty, well-engineered road is being throttled by a too-low static number (restoring credibility and compliance), and it can fall when rain, dark, or congestion make the posted number dangerous. The sign stops lying in both directions.

Conclusion — and a modest joke

Here's where the data leaves us. The time you gain from speeding is small, linear, and routinely erased by a single red light or a single traffic stop. The risk you take on is large, geometric, and — per Nilsson — rises with the fourth power of your speed. Economically, Speedonomics is a terrible trade: you are buying seconds with a currency of fatalities.

But the sign you're disobeying is often wrong too — set by a half-century-old crowd-sourcing rule that's too slow where the road is good and too fast where the conditions are bad. The honest answer isn't "always slow down" or "the limit is bogus, floor it." It's that the number should be dynamic, because the road always is.

And so, the joke to end on: in a world of mis-calibrated signs and 85th-percentile traffic, the single most dangerous thing you can do is rigidly obey the speed limit. Plant yourself at exactly 55 in the left lane because the sign said so, and you won't win a safety award — you'll get tailgated, high-beamed, cussed at, dangerously overtaken on the right, and quite possibly rear-ended. The law-abider becomes the obstacle. Which is the final, delicious irony of Speedonomics: the system is calibrated so badly that both speeding and strictly following the limit can get you killed.

The only safe speed, it turns out, is the one the road is actually asking for — and we should build signs smart enough to tell us what that is.

Methodology note: Travel-time and traffic-light figures are computed from stated assumptions (distances, baseline limits, signal spacing, patrol density) and are illustrative, not survey data — adjust the inputs and the conclusions hold directionally. Risk multipliers use Nilsson's Power Model (fatal ∝ v⁴, serious ∝ v³, injury ∝ v²). Enforcement-exposure probability is a transparent independent-trials model, not a measured ticketing rate.