A thousand waves you cannot count, and three numbers you can.

01 / the surface cost: everything

Standing too close to the water

Stand on the bluff above the bay entrance and try to count the waves. You can't. The water won't hold still long enough.

From up there the sea looks like pure event — a surface that never repeats, every crest already collapsing into the next, every trough filling before you can name it. There is no obvious unit to count. A wave is not a thing so much as a place where energy is passing through, and the place keeps moving. Watch long enough and you begin to suspect the ocean is simply too much information to hold: a different number for every square meter, refreshed faster than thought. This is the honest first impression, and it is also the trap. The mind that insists on recording everything ends up recording nothing it can use.

Twenty nautical miles west of Newport, a three-meter foam buoy named Station 46050 rides the swell over Stonewall Bank and reports on a fixed schedule, every reading time-stamped and sent ashore¹. Closer in, off the mouth of the Alsea, sits Station 46098 — Waldport Offshore — one more disc of instrumented foam doing the same patient work². Each one watches the same restless surface you do. Neither keeps a picture of it. They are not cameras. They are something stranger and more useful: machines for forgetting almost everything on purpose.

A single recent reading off Newport came back as height 4.3 feet, period 5.0 seconds, out of the northwest³. That is the whole afternoon, folded. Thousands of individual rises and falls, the chaos you couldn't count from the bluff, arrive on shore as three quiet figures a harbormaster can read in a glance. Nothing essential was lost. Almost everything was thrown away.

02 / the numbers cost: three terms

What survives the throwing-away

None of the three is a wave. Each is a claim about the whole population of waves — a statistic, in the old sense of the word, a single figure that stands in for a multitude. Significant wave height is the average height of the highest one-third of the waves in the record⁴. The choice of the top third is not arbitrary, and it is the part most people find surprising: it was reverse-engineered to match human perception. When a trained observer is asked how big the sea is, the number they say out loud lands almost exactly on the average of the largest third — not the mean of all waves, which feels too small, and not the single biggest, which feels like a lie³.

Made precise, significant wave height is defined as four times the square root of the area under the wave spectrum — equivalently, four times the standard deviation of the sea surface elevation⁵. That definition is the bridge between the eye and the instrument. It says: the figure a sailor would have guessed is the same figure a buoy can compute from raw motion, with no human aboard. Two completely different observers, the salt-stung pilot and the anchored disc of foam, are made to agree by a single equation.

The second number, peak period, is the dominant period — the wave period carrying the most energy in the spectrum. It decides the entire character of the day. A long peak period means swell born in some distant storm, ordered and powerful, marching in from open ocean; a short one means local chop, raised by the wind blowing past the bridge that morning. One figure, and a small boat at the bar knows whether the crossing will be a gentle lift or a treacherous stack of steep water.

A statistic is a confession about what you have decided not to remember.

And the figures are honest about their own incompleteness, which is the part engineers love. The significant height is not a ceiling; it is a center of gravity. A mariner is expected to read it as a range, with individual waves running from roughly 60 percent of the value up to twice it — the rogue, the lull, all implied by that one number without being written down³. The three numbers do not pretend the sea is simple. They pretend only that its simplicity and its violence can both be predicted from a few honest terms. So far, on every coast that has tested them, they have been right enough to bet a hull on.

03 / the price cost: this much, in bits

A rule for how much to keep

A man working at IBM in the 1970s gave this instinct a rule with teeth. In 1978, Jorma Rissanen published a paper called Modeling by Shortest Data Description and named the principle inside it: Minimum Description Length⁷. The idea is austere and a little thrilling. When several models could explain the same data, prefer the one that lets you write the data down in the fewest total bits — counting both the model and whatever the model leaves unexplained⁸. It is Occam's razor with an accountant attached. Simplicity stops being a matter of taste and becomes a quantity you can measure to the bit.

The accounting has two columns, and the genius is in the tension between them. You pay once to describe the model — the equation, the parameters, the structure. Then you pay again to describe the error: everything in the data the model failed to capture, the residual you still have to spell out by hand⁹. A model that is too simple is cheap to state but leaves an expensive pile of error. A model that memorizes every wave has no error left but costs a fortune to write down. The truth, Rissanen argued, sits at the bottom of the sum of the two.

Drag the slider below. At the far left you are storing raw water — every ripple, every gust-driven flutter, the full unrepeatable record. As you pull it right, the noise falls away and a single spectral peak rises in its place. Watch the description length in the corner fall with it, from thousands of bits toward a few dozen. You are not deleting the sea. You are discovering how little of it you actually needed to keep.

Here is the claim that turns a coding trick into a theory of knowledge. Under this rule, compression is not a convenience laid over the world after the fact — it is the act of understanding. Any genuine regularity in the data can be used to shorten its description, and any description you can shorten reveals a regularity you had not yet named¹². The shortest description that still predicts is also the truest one you can hold. The three numbers survive not because they are pretty but because they keep paying for themselves — on tomorrow's record, and the one after that, on water no instrument has measured yet.

04 / the coast cost: a century of crossings

The same rule, written in concrete

You don't need an IBM lab to find this principle at work. It is carved into the Oregon shoreline, paid for in hulls. Roughly three thousand ships have been lost in Oregon waters over the recorded history of the coast, a great many of them at the river bars — those shallow, shifting mouths where ocean swell meets outflowing current and the sand bottom rearranges itself between one crossing and the next¹⁰. A bar is the cruelest kind of complexity: it looks different every hour and kills the same way every time.

The bar pilot who crosses the Alsea entrance does not attempt to model the whole sea. He runs Rissanen's accounting in his head without ever hearing the name. The few variables that decide whether the crossing holds — the height of the swell, the interval between crests, the stage of the tide — and nothing else worth the weight. Everything he ignores is a bit he refused to pay for. A pilot who tried to track every ripple would be dead before he reached the channel. Survival on the bar is a daily exercise in keeping the description short.

The bridge is an argument about which forces are worth remembering.

The great arch over the bay was the same distillation, set in concrete. Conde McCullough, the engineer who gave the Oregon Coast its chain of bridges, was a master of exactly this discipline — reducing the brute problem of spanning a tidal mouth to the few load paths that actually carry the weight, and letting everything else fall away into elegance¹¹. His original Alsea Bay Bridge, 3,011 feet of reinforced concrete, opened in 1936 after roughly three years of construction, one of a family of coastal spans completed across that decade⁶. A bridge is a hypothesis: these forces matter, the rest can be neglected. Stand it up over salt water and the sea spends decades trying to falsify you.

Salt air won the first round — the original bridge corroded and was replaced within about fifty years — but the lesson held. The span that took its place carried the same crossing on the same few honest assumptions, refined rather than abandoned¹³. The model was wrong about the lifespan of the material; it was right about the physics of the load. That is what a good short description does. It fails in the cheap ways and holds in the expensive ones.

05 / the wager cost: almost nothing

The bet you are already making

Complexity is mostly an artifact of resolution.

Look closely enough at anything — a wave, a coastline, a person, a market, a sentence — and it dissolves into endless detail, each part fractaling into smaller parts, until you are drowning in a record you can neither hold nor use. The bluff above the bay is always available, and it always tells the same lie: that the world is too much to know. But you are not standing on the bluff to admire the chaos. You are standing there because at some point you have to act — to cross, to build, to predict, to decide.

Step back to the resolution at which action happens, and the honest parameters are almost always few. The buoy proved it with three numbers⁵. Rissanen proved it with a sum of two code lengths¹². McCullough proved it with an arch that is still standing in spirit over the Alsea¹¹. The thousand waves were always three numbers. You were only standing too close to see the spectrum.