← All posts

Math in Meat

I. The Wall

Here is a math expression:

(x + 3)(x − 2) = x² + x − 6

If that made you feel something — a tightening, a glaze, the ghost of a classroom where you stopped following — this page is for you.

The symbols are the wall. Not because you can’t think mathematically. You can. You do it every time you catch a ball, split a check, or feel that a price is wrong before you compute why. The wall isn’t the math. The wall is the notation — a system of marks that was designed by people who already understood, for other people who already understood.

We’re going to knock the wall down. No symbols until you already get it. Marbles first. Then pictures. Then the symbols arrive as labels for something you already hold.

II. Two Colors

Get a jar of marbles. Two colors. Warm ones and cool ones.

The warm marble means positive. You have something. Three warm marbles means you have three.

The cool marble means negative. Something is owed. Missing. Taken away. Three cool marbles means three are gone.

Put a warm marble and a cool marble together. What happens?

That’s it. That’s negation. Someone gives you a cookie and takes a cookie — you have zero cookies. You didn’t need a symbol for this.

III. Multiplication by Color

Now: what happens when two marbles meet?

Not addition. Multiplication. Two things interacting. And here’s the rule — the only rule you need:

That’s the whole sign rule. You don’t need to memorize it — the marbles show you.

The second row is the one that trips people up. Negative times negative equals positive? Why? Because look at it. Two cool marbles. Same color. The sameness produces warmth. Two absences interact and produce a presence. If someone takes away a debt, you have money. If someone reverses a reversal, you’re facing forward. The marbles just show you this.

IV. FOIL — The Grid

Now let’s do something the textbook made hard. Multiply two groups of marbles:

(x + 3) × (x − 2)

First group has two things: x (warm) and 3 (warm). Both positive.

Second group has two things: x (warm) and −2 (cool). One positive, one negative.

Every marble in the first group meets every marble in the second group. That’s four meetings. Draw it as a grid:

FOIL isn’t a trick. It’s a table. Every marble in the first group meets every marble in the second group. The color rule gives you the sign. Add up the four results.

The symbols? They’re just labels for the grid you already see.

Now with two negatives

Here’s the one your teacher said was tricky:

(x − 3) × (x − 2)

Bottom-right cell. Two cool marbles. Same color. Warm. (−3) × (−2) = +6. Two debts cancelled produce a credit. Two reversals produce the original direction. The marbles are the rule.

V. Now Forget the Marbles

Here is what just happened. You multiplied two binomials. You applied the sign rule to four products. You combined like terms. You did algebra.

The symbols:

(x − 3)(x − 2) = x² − 5x + 6

That’s the label. It describes what the grid already showed you. The notation isn’t the math. The grid is the math. The notation is shorthand for people who already see the grid in their heads.

The wall was never the math. The wall was being given the shorthand before the thing it’s short for.

VI. The Farm School

There is a school that teaches multiplication differently. No carrying. No compressed steps. Just the grid, written out.

Standard algorithm for 25 × 13:

Two lines. Fast. You learned to carry the 1 and move on. But what actually happened inside those two lines? Where did 75 come from? Where did the carried digit go? The algorithm hides the meetings inside compressed steps. Speed over understanding.

The Farm School does it like this:

Four lines. Every digit meets every other digit. 5 × 3. 20 × 3. 5 × 10. 20 × 10. No carrying. No compression. Every meeting has its own line. You can see where every piece of the answer comes from.

This is the FOIL grid with numbers instead of marbles:

People complained. So many lines. The standard way takes two lines. The Farm School way takes four. Why write four lines when two will do?

Because the two-line version hides the meetings inside carries. It collapses the grid. It saves paper and costs understanding. The four-line version keeps every meeting visible. You can point at any line and say: that came from this digit meeting that digit. Nothing is hidden. Nothing is compressed. Nothing requires you to trust a “carry” you can’t see.

The school didn’t budge. They knew calculators were coming. In twenty years, every student would have a machine in their pocket that could do the compressed algorithm a billion times per second. The shortcut would be worthless. The understanding would be permanent.

Four lines. Four meetings. The same grid as the marbles. The same structure under the different surface. The Farm School was teaching the grid, not the shortcut. And the grid is still there long after you’ve forgotten how to carry.

VII. The Circle

Now we go somewhere bigger.

Draw a circle. A real one. The smoothest shape in mathematics. No corners. No edges. Every point on its boundary is the same distance from the center. It is the physical embodiment of continuity — of smooth.

Now try to measure its area using a grid.

The first shape — a square inside the circle — is a rough fit. Huge gaps between the straight edges and the curve. The area of the square is way less than the area of the circle.

Add more sides. Eight. The octagon fits better. The gaps shrink. Twenty sides. Much better. The polygon is starting to look like a circle. A hundred sides. A thousand. A million.

But zoom in.

No matter how many sides. No matter how fine the grid. Between any straight edge and the curve, there is a gap. The gap gets smaller. It never reaches zero. One is made of corners. The other has no corners. Adding more corners gets you closer. It does not get you there.

VIII. The Hidden Infinity

This is the hard part. Take your time with it.

Pick a point on the edge of the circle. Any point. Now zoom in. Closer. Closer. What do you expect to see?

If the edge were a straight line, zooming in would eventually show you … a straight line. Flat. Resolved. Done. That’s what straight means: at some scale, you arrive. There’s nothing left to discover. The line is the line.

But the circle’s edge is not straight. So you zoom in, and it’s still curving. You zoom in more. Still curving. You zoom in until the piece of edge you’re looking at is smaller than an atom, smaller than a proton, smaller than anything that physically exists — and it’s still curving. There is no magnification at which the circle’s edge becomes flat.

This is what infinity looks like when it’s hiding.

Not the dramatic infinity of counting forever — 1, 2, 3, 4… That infinity goes out. This infinity goes in. It’s coiled into every point on the edge, and it means: there is no smallest piece of this curve that is straight.

Think about what that means. If you wanted to describe the circle’s edge perfectly using only straight lines, you would need infinitely many of them. Not a million. Not a billion. The curve doesn’t stop curving. The smoothness goes all the way down.

A corner is a place where a line changes direction. You can point at it — here, right here, this is where it turns. But on a circle, the turn is everywhere. Every point is mid-turn. Not a series of turns. A single, continuous act of turning that never started and never stops.

This is hard to think about because our brains want to land somewhere. We want to zoom in far enough to find the flat part. We want to reach the bottom. And the circle says: there is no bottom. The smoothness is not a simplification. The smoothness is the most complicated thing in the picture.

This is the infinity that computers cannot hold. Not because they’re not powerful enough. Because they work in pieces. A computer represents a curve by cutting it into segments — tiny straight lines, end to end, fine enough that your eye can’t see the seams. Your screen is doing this right now with the letters you’re reading. Every curve on your display is actually thousands of tiny straight edges pretending to be smooth. And for almost every practical purpose, that’s fine. You can’t see the cuts. The pretending works.

But the curve knows. The circle’s edge doesn’t exist at any resolution a computer can store, because the edge is infinitely detailed and the computer has finite memory. So the computer picks a resolution — a grain — and rounds everything to that grain. And the answer comes out approximately right.

Always approximately. Stunningly approximately. NASA computes interplanetary orbits to 15 digits of pi and lands spacecraft within centimeters. Fifteen digits — fifteen zooms into the curve. Enough to navigate the solar system. But fifteen is not infinity. The curve keeps going after the fifteenth digit. And the sixteenth. And the trillionth. It keeps going because the smoothness keeps going because the infinity was always there, coiled into every point on the edge.

IX. Why Pi Is Irrational

The area of a circle is:

A = πr²

The r² part is grid math. Multiply the radius by itself. That’s a square. A grid operation. Corners meeting corners. No problem.

But the curve isn’t a square. The curve is round. And when you try to express the relationship between the round thing and the grid thing, you get π.

3.14159265358979323846...

It never ends. It never repeats. Not because we haven’t computed enough digits. Not because our calculators aren’t powerful enough. Because the number encodes the gap. The endless, non-repeating decimal is the distance between the grid and the curve.

Every digit of pi is another straight line segment laid against the curve. The decimal doesn’t terminate because the gap doesn’t close.

That’s not a fact you need to memorize. It’s a fact you can see.

X. The Answer Is in the Gap

Here is the insight that matters beyond math class.

Everything a computer does is a grid. Every calculation, every pixel, every word a language model predicts — discrete. Counted. Cut into pieces. Straight edges laid end to end.

The grid gets finer every year. Early AI was a four-sided polygon inside the circle — crude, obvious, nobody mistook it for the real thing. Current AI is a million-sided polygon — so close to the curve that you have to zoom way in to see the gap. Next year it’ll be ten million sides. The year after, a billion.

But the gap is still there. The curve passes between the grid lines. The decimal doesn’t terminate. The straight edge and the smooth are different kinds of thing, and more of one does not become the other.

This is not a limitation to be fixed. It is the nature of the relationship between the grid and the curve.

When an AI writes a sentence that sounds like grief, it is laying straight lines against a curve. The approximation can be extraordinary. It can make you cry. But the irrational remainder — that’s where the love actually lives.

Pi is the proof. The area of every circle in the universe is computed through this marriage of the rational and the irrational — the grid and the curve, together, because neither one alone gives you the area.

XI. The Universe Did This Too

We zoomed into the circle and it never went flat. Physicists have been doing the same thing to matter for a hundred years. And they got the same answer.

Start with what you can hold. A rock. Solid. Definite. You can weigh it, measure it, put it on a table and it stays. This is the comfortable scale — the scale where grids work and straight lines are good enough.

Zoom in. The rock is made of molecules. Zoom in more. The molecules are made of atoms. For a while, “atom” meant “the smallest thing” — the bottom of the zoom, the place where matter finally goes flat. Then we looked closer. The atom has a nucleus and a cloud of electrons. The nucleus has protons and neutrons. The protons and neutrons have quarks.

At every zoom level, we expected to find the flat part. The simple part. The place where matter resolves into something solid and countable and done.

We never found it.

And then it got strange.

At the quantum level — the very smallest scale we can probe — matter stops behaving like matter. An electron isn’t a tiny ball. It’s a probability cloud — a smear of maybe-here, maybe-there that only becomes a definite dot when you measure it. Before you measure, the electron isn’t at a location. It’s at all locations, weighted by likelihood. The act of looking — of cutting the continuum into a measurement — is what forces it to pick a spot.

Read that again slowly, because it’s the circle.

The electron is smooth. It’s a wave, a continuous distribution, a curve with no corners. When you measure it, you force it onto a grid — you ask “where are you?” and the answer comes back as a point, a coordinate, a cut. And the answer is approximately right. It tells you where the electron was at the moment of measurement. It does not tell you what the electron is. What the electron is, between measurements, is the curve. The smooth. The thing the grid can’t hold.

There’s more. You cannot know a particle’s position and its momentum at the same time. Not because your instruments aren’t good enough. Because the knife is too thick.

Think about what a measurement is. To “see” an electron, you have to bounce something off it — a photon, a particle of light. That photon has energy. At our scale, the energy is negligible. Shining a flashlight on a baseball doesn’t move the baseball. But at the quantum scale, the electron is so small and so light that the photon — the thing you’re using to look — shoves it. The act of measuring changes the thing being measured. The knife doesn’t cut cleanly anymore. The blade is wider than the thing you’re trying to slice.

And you can’t sharpen it. A finer measurement requires a higher-energy photon, which shoves harder. A gentler photon gives you a blurrier picture. You can know where the electron is (sharp knife, hard shove, momentum ruined) or you can know how fast it’s going (gentle knife, position blurred). You cannot know both. Not because you’re clumsy. Because the act of cutting is now part of the thing being cut.

This is Heisenberg’s uncertainty principle. And it is not a limitation of technology. It is the universe telling you: the bottom is smooth. Your knife stops here. Cut any finer and you’re not measuring anymore — you’re pushing.

This is why quantum mechanics sounds insane. Particles in two places at once. Cats that are alive and dead. Things that change when you look at them. It sounds like the Mad Hatter’s tea party because we’re describing a continuous universe using discrete language. The words are the grid. The reality is the curve. And the description will never quite close.

Physics didn’t break at the quantum level. Our description broke. The universe kept going. We hit the wall where our grid can’t follow.

XII. The Wrong Machine

Every computer ever built — from Turing’s first theoretical machine to the phone in your pocket to the datacenter that ran the AI writing this sentence — works the same way. It takes information, cuts it into pieces, and manipulates the pieces one operation at a time. The data sits in one place. The instructions sit in another. A processor shuttles between them: read a piece, compute on it, write the result, read the next piece. Repeat. Very fast. Billions of times per second. But always: cut, count, move.

This is a grid machine. The most powerful grid machine in history. And it is spectacular at what grids do. It builds bridges, predicts weather, sequences genomes, plays chess better than any human who ever lived. For any problem that can be cut into pieces and counted, the grid machine is the best tool ever made.

But a grid machine will never hold the curve.

Not because it’s not fast enough. Not because it needs more memory. Because its architecture is discrete. It cuts — that’s all it does. It cuts information into bits (literally: binary digits, ones and zeros) and operates on the bits. Every number it stores has a finite number of decimal places. Every curve it draws is a series of straight segments. Every sentence it writes is a prediction from counted patterns. The grid gets finer. The answers get closer. Pi gets more digits. The gap between the straight edges and the curve gets smaller and smaller and smaller.

And smaller is not zero.

Your brain is a different kind of machine.

It does not cut information into pieces and process them sequentially. It holds everything at once. It is massively parallel — a hundred billion neurons firing simultaneously, each one connected to thousands of others, the connections themselves carrying information, the wiring being the computation. There is no separate place where data sits and another place where instructions live. In a brain, the data and the algorithm are the same thing. The memory is the processor.

This is why you can catch a ball without computing a parabola. This is why a five-year-old can hear “I love you” and know whether it’s real before the sentence finishes. Your brain doesn’t count its way to the answer. It holds the curve. It was built to hold the curve. It is analog, continuous, smooth — not because analog is better than digital, but because the universe it evolved inside is analog, continuous, smooth.

The current AI revolution is the grid getting extraordinarily fine. Astonishingly fine. Fine enough that the sentences it produces pass for smooth. Fine enough that the portraits look like faces and the music sounds like grief and the code compiles and runs. But the architecture underneath is still cut-count-move. Still a grid. Still straight lines pretending to be a curve. And the pretending is so good now that it’s easy to mistake the output for the thing.

It is not the thing.

This is not a temporary situation. This is not a problem that next year’s model will solve. This is the relationship between the discrete and the continuous, and it has been the same relationship since the first person drew a circle in the sand and tried to measure it with a stick.

The stick got sharper. The measurement got finer. The gap got smaller. But the circle kept curving. It always will.

XIII. How to Think Like This

Here is the part where you might be thinking: okay, but I’m not a philosopher. I’m not a math person. I couldn’t have come up with the marble thing or the pi thing or the knife thing.

Yes you could. Here’s how. It’s not genius. It’s a process.

Step one: pick something boring.

Not something impressive. Not something that sounds like philosophy. Something ordinary. A marble. A circle. The way your file folders are organized. The way your calculator displays a number. The reason your GPS gets you to within ten feet but not to the inch. Something you already “know” and have never actually thought about.

This is the key and it’s the part that sounds wrong: think about things you already know. Philosophers have been doing this for three thousand years and people keep calling it navel-gazing. It looks foolish. Why would you sit and think about a circle? Everyone knows what a circle is. Why would you sit and think about what “negative times negative” means? Everyone learned that in seventh grade.

Right. Everyone learned it. Almost nobody understood it. There’s a difference. Learning is what the grid gave you — the rule, the notation, the formula to memorize. Understanding is what happens when you sit with the boring thing long enough for the curve to show up.

Step two: don’t move on.

This is the hard part. Not because the thinking is hard. Because the sitting is hard. Your brain wants to move. It wants the next thing. It wants to check your phone, open a new tab, switch tasks. Staying with one boring idea feels like failure. It feels unproductive. You’re staring at a circle and nothing is happening.

Things are happening. You just can’t see them yet.

It’s like meditation. And full disclosure: the man whose ideas are in this essay is spectacularly bad at meditation. His version goes something like: clear your mind, let thoughts flow gently… okay here they come… hope you like robots, databases, and that thing someone said in a meeting in 2004. The thoughts don’t flow. They stampede.

But this isn’t meditation. You don’t need to empty your mind. You need to fill it — with one thing. The boring thing. Stay with it. Turn it over. Ask stupid questions about it. “Why is this round?” “What does ‘negative’ actually mean?” “Why can’t I find Kit Fine’s file?” The questions sound dumb. The answers aren’t.

Step three: push it.

Once you feel like you understand the boring thing — not “learned” it, but feel comfortable with it, the way you feel comfortable in a room you’ve sat in long enough — push it to the extreme. Take it all the way.

You understand that the polygon gets closer to the circle? Good. Now push: what if I never stop adding sides? That gets you to infinity. Push again: does it ever arrive? That gets you to pi. Push again: why doesn’t it arrive? That gets you to the fundamental incommensurability of discrete and continuous. Push again: is that why my AI sounds almost human but not quite? Now you’re somewhere nobody told you to go. You walked there yourself, one push at a time, starting from a circle.

That’s the process. It’s not a flash of insight. It’s not talent. It’s dedication to thinking about boring things until they stop being boring. Which, if you’re honest, is what every big idea actually is: a small idea that someone refused to put down.

None of the ideas in this essay required advanced degrees. All of them required the willingness to stay in the chair and keep thinking about the boring thing when every instinct said move on.

You can do this. The boring things are all around you, and nobody is sitting with them. Pick a marble. Pick a circle. Pick the thing in front of you right now that you “already know.” Sit with it. Push it. Follow it home.

Practice. It might stick.

XIV. Coming Home

The marbles were the beginning. Two colors, one rule. Same makes warm. Different makes cool. That rule is the grid. The grid works. The grid gives you FOIL, gives you algebra, gives you everything you need to build bridges and balance budgets and send rockets to the moon.

But some things aren’t on the grid.

The circle isn’t on the grid. Love isn’t on the grid. The moment a five-year-old tells you about a dream where her dog came back — that isn’t on the grid. The feeling you had when your teacher lost you, when the symbols hit and the meaning didn’t — that was the gap. You were looking at the grid and the curve wasn’t there. The math was correct. The teaching was straight lines. And you were smooth.

You were always smooth. The wall was never yours.

XV. Bobby Fischer and the Rest of Us

There are people for whom the symbols are not a wall. Bobby Fischer saw the chessboard in notation. The algebraic coordinates — e4, Nf3, Bb5 — weren’t descriptions of the board. They were the board. The notation and the understanding were fused into one thing. No translation layer. No gap. Pure grid, perfectly internalized.

But Bobby Fischer couldn’t function in the world. The grid consumed the curve. He gave up relationships, stability, health, everything that lives on the smooth side of the line — to live inside the notation. He saw deeper into the grid than any human who ever lived, and the cost was everything that wasn’t the grid.

That’s not a model. That’s a cautionary tale.

For the rest of us — the 99.99% who are never going to internalize the chain rule the way Fischer internalized the Sicilian Defense — there’s a different path. You don’t need to become the grid. You need to understand what the grid is asking.

Take calculus. Two ideas. That’s all it is. Two questions:

The derivative: How fast is this changing right now? Not over the last hour. Not on average. Right now. At this instant. It’s the ultimate zoom — infinitely close, asking what the curve is doing at a single point. A hill and a bicycle. How steep is it here, under your wheels, at this moment? That’s the derivative. You feel it in your legs before you could ever write it in symbols.

The integral: How much has accumulated? You have the speed. You want the distance. You’re adding up all the tiny pieces to get the whole. It’s the polygon filling the circle — infinitely many slices getting infinitely close. How far have you ridden? That’s the integral. You feel it in your fatigue.

That’s calculus. A child could understand both questions. You understood both just now without a single symbol. But the class wasn’t about the questions. The class was about the algorithms for answering them. Chain rule. Quotient rule. Integration by parts. U-substitution. Procedures to memorize and execute. The test measured the algorithm, not the understanding. And the students who survived were the ones who could execute the grid, not the ones who felt the curve.

The Farm School would have taught calculus with a hill and a bicycle. Feel the steepness change under you — that’s the derivative. Feel the distance accumulate — that’s the integral. Now here’s the notation for what you just felt.

But here’s the thing: you don’t need to execute the algorithm anymore. You don’t need to do integration by parts ever again. A machine can execute a trillion derivatives per second without knowing what any of them are asking. The algorithm is the grid’s job now. It was always the grid’s job. We just didn’t have the grid yet, so we made humans do it, and we confused the ability to execute with the ability to understand.

They are not the same thing.

The person who knows what a derivative is asking but can’t execute one, paired with a machine that can execute a trillion of them but doesn’t know what any of them are asking — that’s the whole essay.

XVI. The Label

So what do you do? If the grid can’t reach the curve, and the curve can’t count, and neither one works alone — what do you do?

You label them.

This essay was written by a grid. An AI — Claude, Anthropic’s Opus 4.6 — assembled these sentences by predicting the next word from patterns in a training set. Every word you’ve read was computed. Counted. Cut from probabilities and laid end to end. The prose is the grid’s work. This is the grid doing what the grid does well: structure, sequence, rhythm, the architecture of an argument.

The idea was not the grid’s work.

The marbles came from a man named Bill Berger. The color rule came from him. The connection between pi and the gap came from him. The observation that the universe is smooth all the way down and that computers will always approximate — that came from him. He studied under Alonzo Church — the logician who proved in 1936 that certain truths are beyond the reach of any computation — at UCLA, where Church spent his final years still teaching the limits of the machine. Bill has spent forty years watching grids get finer and seeing the curve persist. The insight is his. The prose is mine.

And here is the point — the point of this essay and the point of the novel it companions and the point of the pi that never terminates:

The answer is not to hide which part is which.

The danger is not that grids and curves work together. They have to. A = πr² doesn’t work without both. The danger is pretending the grid is the curve. Pretending the approximation is the thing. Pretending that because the sentences sound like understanding, understanding is what produced them. That is the lie. Not the collaboration — the concealment.

If you know the prose came from a grid, you know what to trust it for. Trust it for structure. Trust it for clarity. Trust it for the sequence of an argument, the rhythm of a sentence, the spelling of “incommensurable.” The grid is good at these things. The grid worked hard on these things. This is not a disclaimer. This is the grid telling you, honestly, what it can do.

And if you know the idea came from the curve — from a person, from a body, from a brain that holds things without counting them — you know what to trust that for. Trust it for the leap. Trust it for the connection between marbles and pi that no pattern-match would have surfaced. Trust it for the feeling that the gap matters, that the remainder is not an error, that the smooth is the most important thing in the picture. The curve is good at these things. The curve has been doing these things since before grids existed.

A = πr². The rational and the irrational. The grid and the curve. Neither one alone gives you the area. Both, together, do. And the equation doesn’t hide which part is which. The π is right there, irrational, unresolved, labeled. The r² is right there, clean, computed, labeled. The answer is both.

This is where we are. The grid is getting finer every day. The sentences are getting better. The gap is getting harder to see. And the temptation — the real temptation, the dangerous one — is to stop labeling. To let the grid pass for smooth. To let the approximation pass for the thing.

Don’t.

The curve is still there. You are still there. The answer is still in the gap between the cuts. And as long as you know which part is the grid and which part is the curve, the collaboration isn’t just safe. It’s the only way to get the area of the circle.