Abakcus

← All articles

An essay in fifteen parts

Is mathematicsInventedorDiscovered?

Eight thinkers, eleven conversations and one question that will not go away. Followed honestly, it leads past physics and philosophy to the oldest answer there is: the theorems were true before us because they were known before us.

By Ali Kaya45 minute read8 voices5 solids, no sixth

Part I · The question

A proof older than Rome

Sometime around 300 BC, in Alexandria, a teacher named Euclid wrote down an argument that fits on the back of an envelope. Suppose, he said, that someone hands you a complete list of the prime numbers. Multiply them all together and add one. The number you get leaves a remainder of one when you divide it by any prime on the list, so none of them divides it. Yet every whole number greater than one is divisible by some prime. So there is a prime that is not on the list, and the list was never complete. However long you make it, it falls short. The primes do not end.

Euclid’s argument, in four lines

  1. Take any finite list of primes: p₁, p₂, …, pₙ.
  2. Form the number N = p₁ × p₂ × … × pₙ + 1.
  3. Dividing N by any prime on the list leaves remainder 1, so none of them divides N.
  4. N has some prime factor. It is not on the list. So no finite list holds every prime.
A torn papyrus fragment with Greek writing and a small geometric diagram
One of the oldest surviving pieces of Euclid’s Elements, a papyrus copied in Egypt around AD 100 and found in the rubbish heaps of Oxyrhynchus. The diagram belongs to Book II.

Empires have risen and vanished since then. Languages have died. Every physical theory Euclid’s contemporaries believed has been thrown out. The proof has not needed a single correction. It is as good this morning as it was in Alexandria, and it will be as good in ten thousand years, in any country, in any script. You can read more about the strange inhabitants of that endless list in our piece on interesting primes. Here I want to ask a different thing about it.

Was it true the day before Euclid proved it? Was it true before there were Greeks, or people, or a planet for them to stand on? Almost everyone, asked this plainly, says yes. Of course it was. The proof did not make the primes endless; it showed that they always had been. But look at what has just been conceded. A precise, detailed, unalterable fact was in force when there was nobody to state it and nothing to write it on. It was not made of atoms. It was not anywhere. And still it was the case.

That is the puzzle hiding inside a question that sounds like a parlor game. If mathematics is invented, the way chess and sonnets are invented, then it should be ours to change, and it should not have existed before us. If it is discovered, the way a continent is discovered, then it was somewhere before we arrived, and we owe an account of where. Neither answer is comfortable. Robert Lawrence Kuhn has spent years putting the question to some of the sharpest minds alive on his series Closer To Truth. I have gone through eleven of those conversations, with the eight people below, and this essay is my attempt to follow the argument to the end and say where I think it comes out.

← InventedDiscovered →

George Lakoff

Cognitive linguist

Built by embodied minds, out of metaphor

Stephen Wolfram

Physicist, computer scientist

A human slice of something necessary

Gregory Chaitin

Mathematician

Began in the mind of God, ended unsure

Silvia Jonas

Philosopher of mathematics

Platonist, 87 percent

Max Tegmark

Physicist

We invent the language, discover the structures

Roger Penrose

Mathematical physicist

A Platonic world, there before us

Mark Balaguer

Philosopher of mathematics

Off the line: no fact of the matter

Paul Davies

Physicist, cosmologist

Off the line: who chose the equations?

Positions are my reading of each interview, not a measurement.

They do not agree. That is what makes them worth hearing together. But I will say now where this is going, so that you can check my steps. Each of these thinkers, the believers in discovery and the believers in invention alike, ends up leaning on something eternal, necessary and beyond the physical world. They give it different names: a Platonic realm, a mathematical multiverse, an abstract necessity, a selector. The founders of modern science had a shorter name for it. I think they were right, and by the end I hope to have shown why. This is a companion to an earlier piece on what mathematics is good for. That one asked what the thing does. This one asks what it is.

Part II · Roger Penrose

Fourteen decimal places

Roger Penrose, smiling, at a science festival in 2011
Roger Penrose in 2011. Photo: Cirone-Musi, Festival della Scienza, CC BY-SA 2.0.

Penrose begins with a chair. It is the sort of object we mean when we say something is real: solid, there, able to hold you up. Ask what it is made of. Fibers and cells, comes the answer. And those? Molecules. And those are atoms, and an atom is a nucleus with electrons around it, and the nucleus is protons and neutrons held together by gluons, and those are made of quarks. Now ask the last question. What is an electron? What is a quark?

At that depth there is nothing left to point at. The only answer physics has is a piece of mathematics. An electron is the kind of thing that satisfies the Dirac equation, a line of symbols that a quiet young man in Cambridge wrote down in 1928. You cannot say what that means without mathematics, because there is nothing else it means. Follow the solid chair all the way down and it turns into equations. Penrose’s point is that this is where the description of reality always leads.

And the equations are not rough sketches. Penrose passes on Feynman’s image for how well the quantum theory of the electron matches experiment: as if you had computed the distance from New York to Los Angeles and got it right to within the thickness of a human hair. Gravity does better still. Newton’s law was accurate to about one part in ten million. Small disagreements showed up in the orbit of Mercury, Einstein replaced the theory, and the replacement has since been checked against a pair of neutron stars circling each other, one of them a pulsar that ticks like a clock. Over decades of watching, theory and observation agree to roughly one part in a hundred trillion.

1 in 10¹⁴

Agreement between general relativity and decades of pulsar timing, as Penrose cites it. Fourteen digits, from equations that fit on a postcard.

The size of the number is impressive, but the important thing is the order in which events happened. Newton, Penrose says, had data good to perhaps three figures. The theory he drew from it went on being right to seven. When Einstein worked, seven were known. He built general relativity mostly out of his own head, guided by a few principles that went back to Galileo and by very little new evidence. His equations then turned out to be right to seven further digits that no instrument of his day could see.

Newton, 1687

Built on about three digits of observation, later confirmed to about seven.

Einstein, 1915

Built when seven were known, later confirmed to about fourteen.

Black: the decimal digits the author had to work with. Gold: the digits the theory got right anyway. Figures as Penrose gives them.

Think about what that rules out. A common deflating view says mathematics is simply the filing system we impose on our observations, a human grid laid over a world that has no grid of its own. If that were so, a theory could never be better than the data it was fitted to. A summary cannot know more than the thing summarized. But the best physical theories keep knowing more. They are right in advance, about digits and phenomena nobody had looked at. Penrose says plainly that he sees no way to make sense of this if mathematics is only our way of organizing what we see. We are not imposing the order. We are finding an order that is built into how the world runs. Many of the equations that changed the world have the same biography: worked out on paper first, obeyed by nature afterward.

Eugene Wigner, a physicist with a Nobel Prize and no taste for mysticism, gave this situation its famous name in 1960. He called it the unreasonable effectiveness of mathematics, and he ended his essay with a sentence that does not sound like a physicist at all.

The miracle of the appropriateness of the language of mathematics for the formulation of the laws of physics is a wonderful gift which we neither understand nor deserve.

Eugene Wigner, 1960

Hold on to two of his words: miracle and gift. A gift is not something you find lying about by accident. Wigner reached for the vocabulary of giving because nothing else fit what he was looking at. We will come back to that.

Part III · Roger Penrose

The archaeologist

Penrose adds a historical observation that is easy to pass over. The Greeks developed geometry as a study for its own sake. They did not think of themselves as doing physics, because it had not occurred to anyone that the geometry of the world could have been different from Euclid’s. They reasoned about lines and numbers as pure thought, and built the method of proof while they were at it. You can still turn the pages of the first English Euclid and watch them do it. Two thousand years later that pure thought turned out to be the working language of the physical universe. Nobody planned this. The key was cut long before anyone knew there was a lock.

Ever since, mathematics has been something you can study for itself, with a life of its own, and Penrose says mathematicians experience it as being out there, with a reality independent of chairs and of us. He thinks in terms of three kinds of reality. There is the physical world. There is the world of our conscious experience. And there is the mathematical world, which he calls Platonic after the philosopher who first described a realm of perfect forms. The fact that there is no largest prime belongs to that third world. It did not become true when someone saw how to prove it. Kuhn presses him: it would have been true if nobody ever proved it? Exactly, Penrose says. And he gives a reason. If the physical world depends this precisely on mathematical laws, then the mathematics had to be there already, for the world to depend on.

Mathematicaleternal truthsMentalwhat we experiencePhysicalwhat there is to touchGOVERNSGIVES RISE TOGRASPS
Penrose’s three kinds of reality. Each arrow is a mystery: why matter obeys mathematics, how matter comes to think, and how thought reaches truths that are not made of matter.

His favorite comparison is to geology or archaeology. The mathematician is an explorer, turning up beautiful things that have been lying there for ages and showing them for the first time, including things no one would have dreamed of. He is honest about the proportions. Only a small part of the mathematical world, an extraordinarily fruitful part, has anything to do with physics. The rest is simply there, vast and strange, with no known use.

His best exhibit is the Mandelbrot set. Take a point c in the plane. Start at zero, square it and add c, square the result and add c, and keep going. Color the point dark if the numbers stay bounded forever. That is the entire recipe. Now look at what it draws, and then look closer.

z → z² + c

The Mandelbrot set. Click anywhere along its edge to go deeper. Magnification 1×

Every time you zoom there is more: spirals, filaments, small copies of the whole shape with their own decorations. Nobody designed any of it. Benoit Mandelbrot did not put it there, and neither did the programmer of this page. A dozen characters of arithmetic contain it all, and they contained it before there were computers able to show us. If this is an invention, it is the only invention in history whose inventor was astonished by it and could not alter a single pixel. Penrose asks how such a thing could be a mental creation, and I have never heard a good reply. The same feeling comes from his own notebooks, page after page of drawings by a man who is clearly looking at something.

Part IV · Max Tegmark

There is no sixth one

Max Tegmark gives the neatest formulation of the whole debate. Do not confuse the language of mathematics with the structures of mathematics, he says. The language we invent. The structures we discover.

His example is the one at the top of this page. Plato’s circle became fascinated by solids whose faces are all the same regular polygon, meeting the same way at every corner. They found five: tetrahedron, cube, octahedron, dodecahedron, icosahedron. They were perfectly free to choose the names. They could have called the twelve-faced one anything they liked. What they were not free to do was to produce a sixth. There is no sixth one. The last book of Euclid’s Elements ends by proving it. No vote was taken, and no culture, however inventive, has a different list.

This is the test I find most useful. Inventions bend to their makers. We changed the rules of chess when the old game was slow; we can spell a word differently next year. A discovery is something that pushes back. You cannot make the square root of two a fraction, and people have wanted to since the Pythagoreans. The proof that it is irrational does not ask for your consent. A mathematician is the freest person in the world when choosing symbols and has no freedom at all about what turns out to be true.

Tegmark has a nice picture for how culture enters without making mathematics arbitrary. Send two people to live in New York for ten years. They will not walk exactly the same streets. But both will find Central Park and Times Square. In the same way, he says, a civilization on the far side of the galaxy would find the integers, addition and multiplication, and the Platonic solids. Which remote side streets each one wanders into depends on taste. The city is the same city.

How big is the city? Tegmark imagines an atlas of every mathematical structure, ordered by complexity. Page one holds the simplest things, like the empty set. Further on you reach the cube and the dodecahedron, and much later the curved four-dimensional spaces and the abstract spaces of quantum theory that physicists use to describe our world. The atlas has no last page. Yet it is not a place where anything goes. To get in, a structure has to be free of contradiction, and proving that is hard work. He cites Hilbert’s old dictum that in mathematics, to exist is to be consistent.

Then Tegmark takes a step that most people will not take with him. He proposes that physical reality does not merely resemble a mathematical structure. It is one. Every physical quantity, he argues, is at bottom a number attached to a point: the strength of the magnetic field here, of the electric field there, like the temperatures and pressures on a weather map but fundamental. Give all the numbers and you have said everything there is to say. And he holds that all the other structures in the atlas exist in just the same way ours does.

I will come back to that last claim, because I think it is where his view runs into trouble. But notice something he says in passing and calls a deep mystery. Out of the endless atlas, the structure we find ourselves in is far simpler than it had to be. It is full of symmetry. It strikes us as beautiful. He does not know why. Neither does anyone else who starts from the atlas alone. A list of all possibilities cannot tell you why the elegant one is the real one.

Part V · Mark Balaguer

Four addresses for a number

So far the discovery side is winning easily. Now the bill arrives. If mathematical truths are found and not made, they were somewhere to be found. Where?

Plato and Aristotle walking side by side in Raphael’s fresco The School of Athens; Plato points upward
Plato, pointing up, beside Aristotle. Raphael, The School of Athens, 1509–11.

Mark Balaguer is a philosopher who has spent his career on exactly this, and he is a useful guide because he has defended more than one side. Real mathematics, he tells Kuhn, is not the skill of solving problems. It is a theory about something, in the way physics and biology are theories about something. It studies structures: the natural numbers that start at zero and go on forever, and far stranger things. Theorems are claims about what those structures are like. The mystery is why structures that seem entirely abstract and intellectual should be relevant to how physical things behave. And beneath that lies a blunter question. The number seven is not in the physical world. Does it exist at all? He says there are four possible answers.

In the head

Mentalism

Numbers are ideas in human minds.

Then two plus two was not four before there were heads.

In the world

Physicalism

Mathematics is about physical things.

The world may be finite. The integers are not.

Outside space and time

Platonism

Abstract objects exist on their own.

Where is that, and how would we ever learn of it?

Nowhere

Fictionalism

There are no numbers. Theorems are useful stories.

Stories do not predict a pulsar to fourteen digits.

The first two do not survive long. If numbers were ideas in human heads, there would have been no arithmetic for the neutron stars to obey before there were heads, and the truths of mathematics would be facts of psychology that might differ from person to person. If numbers were physical, we would have to find the infinite, the perfect and the exact somewhere in a world that is, as far as we can tell, finite, grainy and approximate. Balaguer thinks the only serious contenders are the last two: the strongest kind of realism, and the strongest kind of anti-realism.

Platonism says mathematical objects exist but are abstract: not physical, not mental, not in space, not in time, and not able to cause anything or be affected by anything. Fictionalism says there are no such objects and that mathematics, taken at face value, is strictly false, like a novel that is well constructed and about nobody. Balaguer has argued for each. His conclusion, when Kuhn asks which one is right, is startling. Neither. It is not only that we cannot find out. He thinks there is no fact of the matter.

His reason is worth hearing, because it is the most honest thing anyone says about Platonism in all these interviews. When he teaches the view, his beginning students scratch their heads. Non-physical, non-mental, what? He remembers reacting the same way as an undergraduate. Then, he says, philosophers work on the idea for years, become comfortable with it, and start talking as though it were clear. But the beginners were right. We have been told only what an abstract object is not. Nobody has said what it is. It is not like a remote village that an expedition could go and find. The notion is so thin that we do not even know what it would be for it to be true.

I think Balaguer has diagnosed the disease precisely and drawn the wrong conclusion from it. The trouble with the Platonist’s heaven is real. A truth with no location, no cause and no knower is very hard to believe in. But that is a reason to ask what Platonism has left out, not a reason to give up on the plain fact that Euclid’s theorem was true before Euclid. Keep his complaint in mind. Something that is not physical, not spatial, not temporal, and yet real: stated that way it sounds empty. There is one very familiar kind of thing that fits the description and is not empty at all. We will get to it.

Part VI · Silvia Jonas

The philosopher who is 87 percent sure

Silvia Jonas weighs the arguments the way a judge weighs testimony, and she comes down for Platonism. Kuhn asks how confident she is on a scale of one to a hundred. She pauses and says 87. It is the most charming number in the series, and the reasons behind it are solid.

The first is about meaning. When I say that a skirt is red, what makes the sentence true is that there is a skirt and it is red. When I say that two plus two equals four, the natural reading is the same: the sentence is true because the things it talks about are there and are as it says. Platonism lets us treat mathematical sentences like every other true sentence we utter. Its rivals have to explain why the most certain statements we know are true in some special, watered-down sense.

The second is the one that moves most people: the nearly miraculous fit between mathematics and natural science. Philosophers call the formal version the indispensability argument. If mathematics were only a convenience, shorthand that makes theories tidier, we could in principle do without it. If instead it carries real explanatory weight, so that the science cannot be stated or understood without it, then we have the same grounds for believing in numbers that we have for believing in electrons. The strongest reply came from Hartry Field, who in 1980 actually rewrote Newton’s theory of gravity without referring to numbers at all, to show it could be done. It was a heroic piece of work. It was also enormously more cumbersome than the original, it has not been extended to the physics we now use, and it leaves the old question standing: if numbers are dispensable fictions, why does reasoning with them never lead us astray about the world?

Then Jonas names the objection that Platonists fear most. It is called the problem of epistemic access, and it is simple to state. We learn about things by being affected by them. Light bounces off the chair and reaches my eye. But abstract objects, by definition, do not affect anything. None of our five senses reaches them. So how do we know anything whatever about them? How did Euclid, sitting in Alexandria, obtain reliable information about a realm that sends no signals?

Her own answer draws on recent work in set theory. For over a century one question has resisted every attack: whether there is a size of infinity strictly between the infinity of the counting numbers and the infinity of the points on a line. This is the continuum hypothesis. It is now known that the standard axioms cannot settle it. Mathematicians can construct perfectly consistent universes of sets in which it holds and others in which it fails. Joel David Hamkins has argued that this simply is the answer: there is not one mathematical universe but many, a multiverse of them, and a theory describes a real universe whenever it is consistent. On that view, Jonas says, the access problem has a straightforward solution. To know that a consistent theory is true of something, you do not need a mysterious line of sight to its objects. You only need to see that it is consistent, and that you can check on paper.

It is an ingenious move. Kuhn spots the price at once. If there are endlessly many mathematical universes and we simply pick the ones that happen to match nature, then the match is no longer astonishing. The very abundance that solves the access problem dissolves the miracle that made Platonism attractive in the first place. Jonas grants that pluralism has consequences here. So the Platonist is squeezed. With one mathematical heaven, she cannot say how we see into it. With infinitely many, she cannot say why this world hangs on this one. Both difficulties have the same root. The theory contains truths, and it contains us, and it contains a universe, and it contains nothing at all that connects the three.

Part VII · George Lakoff

The case for invention: the body

Given all that, you can see the appeal of bringing mathematics back down to earth. George Lakoff, a linguist, does it more thoroughly than anyone. He tells Kuhn that the choice between invented and discovered is a false one, and then gives an account in which mathematics grows entirely out of the human body and brain.

It starts with what infants can do. Show a baby one puppet on a little stage, lower a screen, add a second puppet, and raise the screen. If there are two puppets, the baby is content. If, by sleight of hand, there are three, the baby stares. Long before words or counting, there is a sense that one and one make two. Lakoff’s claim is that everything else is built on such beginnings by a mechanism he has studied all his life: metaphor.

Small children spend weeks putting things into containers and taking them out. From this, he says, comes the idea that numbers are collections of objects, and that a bigger collection is a bigger number. Children walk, one step and another and another. From this comes the idea that numbers are points along a path, with the starting point as zero. Once numbers are places on a line, you can face the other way and walk, and there are the negative numbers. And here is his loveliest example. Why is minus one times minus one equal to plus one? Because we can rotate things in our minds. Multiplying by minus one is a half turn, swinging the positive side of the line around to the negative. Do it twice and you are back where you started. Ask what single operation, done twice, gives a half turn, and you get a quarter turn, and that is the square root of minus one.

−10+1i× i = 90°× i AGAIN
Multiplying by −1 is a half turn, so doing it twice brings you home: (−1)(−1) = +1. Half of a half turn is a quarter turn, and that is the square root of −1.

Even infinity, on his account, is a metaphor. An endless process, counting on and on, is understood as though it were a finished one with a final result, the way we speak of having walked a mile. He and Rafael Núñez argue that this one conceptual move underlies infinite sets, the decimal expansion of π and the limits of calculus. And once a metaphor is in place, it has consequences that its makers did not foresee and cannot adjust. That is how Lakoff makes room for the experience of discovery. You set up the metaphor, and then you find out what follows.

I admire this work, and I think most of it is true. It is a real account of how creatures like us come to grasp mathematical ideas, and why Euler’s identity can be explained with a walk around a circle. But notice that it answers a different question. It tells us how the human mind climbs up to mathematics. It does not tell us why the thing we climb to is the same for everyone, cannot be otherwise, and is already obeyed by the stars.

Kuhn puts the difficulty well. Grant every word, he says. For all but the last sliver of cosmic history there were no brains and no metaphors. The physical world ran regardless, and it ran on what we now describe with mathematics. Lakoff answers that the mathematics is not in the world. The world is as it is. His example is a spiral galaxy. We say there are logarithmic spirals in nature, but what is out there is a rotation and an outward motion. The spiral formula comes from combining our ways of understanding rotation and expansion. It is in our understanding of the galaxy, not in the galaxy. And it is no miracle that our mathematics fits, he adds, since we make it from the same faculties we use to perceive the world.

This is the weakest moment in a strong interview. If the galaxy really rotates at a steady rate while expanding at a proportional one, then it really traces that curve, whether or not anyone is there to name it. The relationship between the two motions is a fact about the galaxy. Calling it “our understanding” does not move it into our skulls. And the explanation of the fit is circular in an interesting way. It says our minds are suited to the world. Suited by what, and why should a mind suited to catching a ball be suited to the interior of a neutron star, which no human sense has ever met? Lakoff’s own best observation counts against him. He says a well-made metaphor has entailments you must then discover. So even on the embodied view there is a realm of what follows from what, and it is not up to us. That realm is precisely what we were asking about.

Part VIII · Stephen Wolfram

The case for invention: the artifact

Stephen Wolfram attacks from the opposite direction. Lakoff says mathematics is too human to be eternal. Wolfram says our mathematics is too particular to be the one eternal thing people take it for.

Look at the history, he says. In Babylon there was arithmetic for trade and geometry for surveying land. Mathematics as we know it is what you get by generalizing those two subjects over and over, guided by one method: the proving of theorems. At each step, from whole numbers to real numbers to complex numbers and beyond, mathematicians asked which favorite theorems they could keep, and extended in whatever direction kept them. So the subject has always stayed close to the places where theorems can be proved. Its shape reflects our path, he argues, not necessity.

Then he does something only a person who builds mathematical software for a living would think of. All of the roughly three million published mathematics papers rest on a few pages of axioms. Axioms are strings of symbols. So you can list every possible string and ask what mathematics each one would generate. Where is ours in that list? He checked for ordinary logic. Depending on how you count, it turns up at about the fifty-thousandth position. Not first. Not marked in any way. The other entries are not nonsense, either. Many are perfectly consistent systems on which a mathematics could be built. If a visitor from another star handed us one of them, he says, we would have no grounds for calling it illegitimate.

He goes further. Our mathematics looks orderly, with most of the questions people care about eventually getting solved. Wolfram suspects this is an effect of where we have chosen to walk. Ask questions at random and nearly all of them would be unanswerable; the undecidable is close at hand, and history has steered around it. Even the celebrated fit with nature is, he thinks, partly circular: the parts of nature that science has mastered are exactly the parts our kind of mathematics can handle. His verdict in the early interview is that human mathematics is one of the great artifacts of our civilization. Wonderful, hard-won, and an artifact.

That sounds like a plain vote for invention. But Kuhn talked to Wolfram again years later, and the later conversation is the more revealing one. By then Wolfram had given a name to what all those possible systems add up to: the totality of every possible computation, carried out in every possible way. He calls it the ruliad, and he puts it underneath both physics and mathematics. We are observers of a certain kind inside it. The physical world we perceive is the slice our kind of observer can take; the mathematics we do is another slice. He notes how remarkable it is that we can do mathematics at a high level at all, talking about the Pythagorean theorem without descending to the definition of a real number each time, just as we can talk about the flow of water without tracking every molecule.

So is mathematics invented or discovered? Something in between, he tells Kuhn. Invented, because the slice depends on what we are like; different observers would have a different mathematics. Discovered, because there is an ultimate reality under the slice that we did not make. And when Kuhn asks whether that underlying thing was always there, Wolfram describes it as an abstract necessity, something that could not have been otherwise and that nobody can stand outside of.

I find this the most instructive turn in the whole series. Here is the most radical critic of the idea that our mathematics is special. He has catalogued the alternatives and measured our place among them. And at the foundation of his system he has put something that is eternal, necessary, immaterial, all-encompassing and the source of both the physical world and the truths of mathematics. He would not use the word, but that is a list of divine attributes with the name removed. The argument for invention, pushed as hard as it will go, does not arrive at a world where mathematics is merely ours. It arrives at a necessary being and declines to say whether it thinks.

Part IX · Gregory Chaitin

The mathematician who confessed

Of all Kuhn’s guests, Gregory Chaitin is the one who says out loud what the others circle. He begins sensibly enough. Some mathematics, he says, feels so fundamental that if you had not found it someone else would have, because it has to be there. Other mathematics feels contrived, an exercise performed to get a paper published. He will not name the fields he has in mind. So perhaps some is discovered and some invented.

Then Kuhn asks whether anything in mathematics is truly discovered, sitting there forever, and Chaitin does something rare in an interview. He tells the truth about what people in his profession privately believe. Philosophers, he says, regard Plato’s world of perfect ideas as hopelessly old-fashioned. But look into the unconscious of working mathematicians, himself included, and you will find what he calls a medieval, theological belief in exactly such a world. Otherwise, he asks, what have they been doing with their lives? Is it all a game made up as we go along? He would like to believe he has not thrown his life away on that.

Kuhn, who never lets a comfortable answer stand, says he wants to know what is true, not what would make Chaitin happy. And Chaitin goes on. Pure mathematics, he says, is independent of the physical world. It is a separate reality. He does not know where it is. Do not ask him where the positive integers live. But they do not live here. It is, he admits, like a religion: a belief in an invisible world, better and purer than ours. And then, saying he means to be provocative, he gives the invisible world its traditional name. Mathematical truths, he says, are “thoughts in the mind of God.”

Kurt Gödel as a young man, wearing round glasses
Kurt Gödel as a student in Vienna, 1925.

He has a story to go with it. An astrophysicist finds himself seated beside Kurt Gödel at a formal dinner at the Institute for Advanced Study in Princeton. Gödel was the greatest logician since Aristotle and famously hard to talk to, so the man seizes his chance. To break the ice he describes a splendid new result in astrophysics, something he is proud of, sure it will win the great man’s interest. Gödel listens. Then he says that he does not believe in empirical science; he believes only in truths known independently of experience. End of conversation. Chaitin tells it as a joke, and then says that in a way this is what every pure mathematician does. He mentions that Gödel, in papers published only after his death, wrote that pure mathematics is the one place where the outlook of medieval theology survives, and wanted it kept that way.

It should be said that Gödel meant it. He was a convinced theist. He left behind a formal proof of the existence of God, built with the tools of modern logic, and he argued in a major lecture that mathematics is not a human creation but describes a reality that exists independently of us. The man who proved the deepest theorem about the limits of proof held that mathematical truth is found, and he was not embarrassed about where it is kept.

Chaitin himself does not end there, and he deserves a fair report. His life’s work, which extends Gödel’s, shows that there are mathematical facts that are true for no reason any proof can capture. From this he draws a practical moral. Perhaps mathematicians should behave more like physicists. If computer experiments strongly suggest that something is so, and it would be very useful, and no proof is in sight, perhaps it should be adopted as a new axiom, the way physics adopts a new law. That sounds like inventing mathematics as we go, and he knows it. He started, he says, from the conviction that mathematics is perfect and in the mind of God, and was forced against his will toward the opposite view, by reasoning that itself relied on the perfect reality he started with. He ends by saying he is not quite sure where he stands, and that the story is more like a novel than a mathematics paper.

I think the novel has a better ending than he gives it, and it is contained in his own results. I will come to that under the objections.

Part X · Paul Davies

Who chose the equations?

Paul Davies is a physicist, and the question he brings is not about numbers in the abstract. It is about why there is a world that follows them. All scientists agree, he tells Kuhn, that we do not live in any old universe. It is ordered, ingenious and intelligible, and it permits beings who can ask about it. Could it have been otherwise? And if so, why is it like this? He walks through the answers on offer, and his tour is the best short map of the territory I know.

The first answer is that there is no answer. Most of his colleagues, he says, would prefer to stop worrying and accept the universe as a package of marvels that simply is. Pressed, they say that at bottom there is no reason. Davies cannot swallow this. Science depends from top to bottom on the world being rational. To say that this rationality rests on nothing, that a meaningless universe just happens to imitate a meaningful one flawlessly, seems to him very hard to believe. I agree, and I would put it more strongly. A person who says the order of nature is a brute accident has no right to be surprised by anything, and no reason to expect the next experiment to come out like the last.

The second answer is that it could not have been otherwise. Perhaps a final theory will show that only one set of laws is logically possible, a formula short enough for a T-shirt with nothing left to choose. Einstein is said to have wondered whether God had any choice in making the world; this answer says no. Davies thinks it fails. It is easy to write down other self-consistent laws describing other possible worlds. Any final theory would still rest on a particular kind of mathematics that might have been different. The problem moves up a level and stays a problem.

The third answer is the multiverse. Maybe there are countless universes with different laws, and we naturally find ourselves in one that allows us. Davies grants that this may be part of the truth and denies that it is an explanation. You must take the unseen universes on faith. You still need something to generate them, and laws governing the generator, and the framework of quantum theory and relativity besides. You have assumed most of what you set out to explain. He adds a mischievous argument he first offered as a joke: in a big enough multiverse, simulated worlds would vastly outnumber real ones, so we should conclude that our own laws are fake, which undercuts the reasoning that led us there.

The fourth answer is the traditional one. The universe looks designed, nobody can deny that it looks designed, and so perhaps it was. Davies treats this more respectfully than most of his colleagues would, and states the usual objection: then what explains the designer? He notes that careful theologians reply that God is not one more item needing a cause but a being that exists necessarily, containing the reason for its own existence. He calls this a deep argument and says he is not persuaded it works.

His own preference is for a universe that somehow explains itself from within, with mathematics, physics and observers bound into one self-consistent loop. He is candid that this is words and not yet equations, and that set beside the others it looks ridiculous too. What I value in Davies is not that proposal but two observations he makes on the way.

The first is about purpose. Science, he notes, has been trained to avoid it, yet it is the most ordinary kind of explanation there is. Ask what his wife is doing in the kitchen, mixing ingredients and heating the oven, and the true answer is that she is baking a cake. The end explains the steps. He thinks the universe as a whole may need that kind of explanation, and I would only add that we know of exactly one sort of thing that acts for the sake of an end. It is a mind.

The second is what he calls the selector problem, and it is the sharpest point in all eleven conversations. Suppose you believe, with Tegmark, that mathematical structures are simply there. Mathematics alone gives you every consistent possibility, the beautiful and the ugly, the fertile and the sterile, without preference. But not everything possible is actual. Something marks the difference between a set of equations that is merely consistent and a set that has a universe answering to it. Where did that rule come from? Why this one? Even the people who say that every mathematical possibility is real have chosen mathematics as the criterion, Davies points out, which is itself a choice. Why not all the beautiful universes, or all the good ones?

Equations are descriptions. A description of a fire warms nobody. However complete the blueprint, something has to build. Abstract objects, as Balaguer reminded us, do not cause anything; that is part of what makes them abstract. So the Platonic realm, even if we grant it, is the wrong kind of thing to produce a world. It can be what the world was made according to. It cannot be what made it. For that you need something that can select among possibilities and bring one about. We have a word for the power to choose among possibilities for a reason. The word is will.

Part XI · The argument

Thoughts need a thinker

Let me collect what the eleven conversations have established, using only what the speakers themselves grant.

Mathematical truths were true before any human being existed and would be true if none ever had. That is Penrose, Tegmark and Jonas, and it is conceded in practice by everyone who trusts a proof. They do not change. They are not made of matter and are not located in space. They are necessary: they could not have been otherwise. That is Wolfram’s own word. They are not inert curiosities, because the physical universe conforms to them to fourteen decimal places. That is Penrose again. And they are intelligible: minds can come to know them, though no signal travels from them to us. That is the fact Jonas has to explain.

Eternal. Unchanging. Immaterial. Necessary. Sovereign over the physical world. Knowable by the mind. Read the list again without thinking about mathematics. It is not a new list. Every item on it was written down by theologians many centuries ago, as an attribute of God.

Now recall Balaguer’s complaint. An abstract object is defined entirely by negatives, and nobody can say what it positively is. He is right about the Platonist’s version. But there is one thing in our experience that is real, is not made of matter, has no position in space, and yet is not empty or mysterious in the least, because each of us is acquainted with it every waking moment. It is a thought. Your idea of a triangle weighs nothing and is nowhere, and it is as real as anything you know.

That suggests what Platonism left out. A truth is something that can be understood. A proposition is something that can be meant. These belong to minds the way a color belongs to a surface. To say that there are thoughts, complete and exact, that no one is thinking and no one ever thought, is like saying there is a pain that nobody feels. The Platonist’s heaven is strange for exactly this reason. It is full of ideas and has nobody in it.

So where are the truths of mathematics? They cannot be in human minds, which came late, are few, and make mistakes. They cannot be in matter. And floating free they make no sense. What remains is that they are in a mind that did not come late, is not limited, and does not err: a mind that is itself eternal and necessary. The integers are not homeless. Chaitin said it as a provocation, and I think it is the plain truth. Mathematics is the content of the mind of God.

William Blake’s The Ancient of Days: a bearded figure kneels in a blazing sun and reaches down with a pair of compasses
William Blake, The Ancient of Days, 1794. The compasses again.

Geometry, which before the origin of things was coeternal with the divine mind, supplied God with patterns for the creation of the world.

Johannes Kepler, Harmonices Mundi, 1619

Part XII · The tradition

A very old idea

None of this is my idea, and it is not a late attempt to rescue religion from science. It is the view from inside which science was made. The people who created mathematical physics did not hold their faith in one hand and their equations in the other. They looked for mathematical laws because they believed the world had a lawgiver, and they expected to be able to understand those laws because they believed their own minds came from the same source. The history of mathematics reads differently once you notice it.

Medieval illumination of the Creator bending over the unformed world and measuring it with a large pair of compasses
God as architect of the world. Frontispiece of a Bible moralisée made in Paris around 1225, now in Vienna.
  1. c. 360 BCPlato, TimaeusA divine craftsman orders the cosmos by number and assigns the regular solids to the elements. Plutarch later credits him with the saying that God is always doing geometry.
  2. c. 390AugustineSeven and three are ten always and everywhere. Unchanging truth above the human mind points to an unchanging mind.
  3. c. 1225Bible moraliséeThe frontispiece shows the Creator bending over the unformed world with a pair of compasses.
  4. 1619KeplerGeometry is coeternal with the divine mind and supplied the patterns for creation.
  5. 1623GalileoThe book of the universe is written in the language of mathematics.
  6. 1713NewtonThe system of the world could only proceed from the counsel of an intelligent and powerful Being.
  7. 1714LeibnizEternal truths have their reality in the divine understanding.
  8. 1883CantorHolds that the infinite numbers exist in the mind of God, and writes to theologians to say so.
  9. 1886KroneckerGod made the integers; all else is the work of man.
  10. 1930JeansThe Great Architect of the universe begins to appear as a pure mathematician.
  11. 1963DiracGod is a mathematician of a very high order.

Start with Plato, since his name is on the theory. In the Timaeus the world does not arise by itself. A divine craftsman, looking to eternal patterns, brings order to disorder by means of number and proportion, and the five regular solids are his building blocks: four for the elements and the dodecahedron for the whole. The physics is fanciful. The structure of the thought is not. From the very beginning the Platonic forms came with a mind that knows them and a maker that uses them. The lonely, unthought abstract object of the modern textbooks is Plato with the most important part taken out.

Augustine put the part back and made the argument explicit. In a dialogue written around the year 390 he asks his friend to consider that seven and three make ten. Not just today, and not just as far as we have checked. Always. No one can make it otherwise, and all who can reason see the very same truth, not each a private copy. So this truth is not part of my mind or yours. It stands above our minds, because we do not judge it; we are judged by it, and corrected by it when we add wrongly. Yet it is plainly the kind of thing that belongs to intelligence. Augustine concludes that there is an eternal and unchangeable wisdom in which such truths have their home, and that this is what people have always meant by God. Thirteen centuries later Leibniz said it in one sentence: the eternal truths have their reality in the divine understanding, and without God there would be nothing real even in what is merely possible.

The same conviction runs through the scriptures of the great monotheistic traditions, and the scholars who kept mathematics alive through the Middle Ages, in Baghdad and Córdoba and Paris, knew those lines by heart. The Book of Wisdom says to the Creator: you have ordered all things by measure, number and weight. The Qur’an says that the sun and the moon move by an exact reckoning, and that everything was created in due measure. A thirteenth-century French Bible opens with a picture that has been reproduced ever since. God stoops over the formless world holding a pair of compasses, drawing its boundary. The medieval mind saw no tension between piety and geometry. Geometry was the trace of the Creator’s hand.

Engraving of Kepler’s model of the solar system: nested spheres separated by a cube, a tetrahedron and the other regular solids
Kepler’s nested Platonic solids, from the Mysterium Cosmographicum, 1596.

Then come the founders. Kepler believed, as a young man, that the spacing of the planets was fixed by nesting the five Platonic solids inside one another. He was wrong, and found out by testing the idea against the best observations in Europe. But the faith behind the guess, that the heavens were laid out on a geometrical plan a human mind could recover, carried him through years of calculation to the three true laws of planetary motion. He said he was thinking God’s thoughts after him. Galileo wrote that the universe is a great book lying open before our eyes, that it is written in the language of mathematics, and that without that language one wanders in a dark labyrinth. A book has an author. He did not think that needed saying.

Newton, having reduced the motions of the heavens and the fall of an apple to one law, added a closing essay to the Principia to state what he took it to show: this most beautiful system of sun, planets and comets could only proceed from the counsel and dominion of an intelligent and powerful Being. James Clerk Maxwell, who did for electricity and light what Newton did for gravity, had a verse of the Psalms carved over the door of the Cavendish Laboratory: the works of the Lord are great, sought out by all who take pleasure in them. You can follow his own hand at work in the equations on a Warsaw wall.

The mathematicians were no different. Georg Cantor, who first tamed the infinite, held that his transfinite numbers exist in the highest degree of reality as ideas in the divine intellect, and he corresponded with theologians about it. Leopold Kronecker, his fiercest opponent on nearly everything, drew the line between discovery and invention with a toast: God made the integers, and all the rest is the work of man. Gödel we have met. In the twentieth century the physicist James Jeans wrote that the universe was beginning to look less like a great machine and more like a great thought, and that its architect appeared to be a pure mathematician. Dirac, no churchgoer, said that one could describe the situation by saying that God is a mathematician of a very high order who used very advanced mathematics in constructing the universe. Even Paul Erdős, who liked to tease the Almighty, could not describe a perfect proof without saying that it came from The Book, the volume in which God keeps the best proof of every theorem. You do not have to believe in God, he told his colleagues, but you should believe in The Book.

That last remark says more than Erdős intended. The working mathematician’s experience is of reading something already written. The only disagreement is about whether to admit there is a writer.

Part XIII · The test

What the divine answer explains

A view earns its keep by what it explains. Take the open problems one at a time and see what becomes of them.

Where is mathematics? This was Balaguer’s question, and the reason he gave up. Truths are in a mind, which is the one place a truth can be without strain. Not in ours, but in a mind that is eternal as they are eternal and necessary as they are necessary. They are abstract from our side, since we cannot touch them. They are not abandoned.

How do we know it? This was Jonas’s problem of access, and for the bare Platonist it is close to fatal: no causal thread runs from the abstract realm to the human brain. But suppose the human intellect and the order of number have a common author. Then our ability to follow a proof is not a lucky break. A mind made by the mind in which mathematics resides would be expected to find mathematics intelligible, as a child raised in a language understands it without being able to say how. Einstein wrote that the eternal mystery of the world is its comprehensibility. On this view it is what one would predict. We were made to read, and given something to read.

Why does the universe obey? This is Wigner’s miracle and Penrose’s fourteen digits. If matter and mathematics were strangers that happened to meet, the fit would be beyond belief. If the world was made by a mind according to what that mind knows, the fit is exactly what you would expect. Finding geometry in the orbits is then no stranger than finding grammar in a letter. And the order of events that so impressed Penrose, equations first and confirmation later, stops being a puzzle. Our reason and the world’s structure agree in advance because both come from the same reason.

Who chose? This was Davies’s selector, and it is where every impersonal theory fails. The Platonic realm contains all consistent possibilities and prefers none. The ruliad contains every computation and prefers none. Something actual had to distinguish this world from the infinity of others that were equally possible. Abstract objects cannot do that, since they do nothing. A will can. Here the old theology has a distinction that the modern discussion lacks. Mathematical truths are necessary; they belong to what God eternally knows, and even God does not make two and two equal five. The existence of this particular universe is not necessary. It is chosen. That is why mathematics alone can never tell you what the world is like, and why Kepler had to check his beautiful theory against Tycho’s numbers. You learn what is possible by thinking. You learn what was chosen by looking.

Why is it beautiful? This was the mystery Tegmark confessed. Out of an atlas with no last page, the structure we live in is simple, symmetrical and lovely to the minds that study it, and the surest guide to a new law of physics has repeatedly been its elegance. An index of all possibilities has no reason to favor the elegant ones. A maker does. We do not find it odd that a cathedral is beautiful.

Why is mathematics one? Wolfram, in the later interview, remarks on the deep correspondences that keep turning up between branches of mathematics that seem to have nothing to do with each other: algebra and geometry, number theory and analysis. He attributes them to everything being, underneath, one thing. I agree. Parts of a single thought fit together. The unity he is describing is the unity of a mind.

Six problems, and one answer covers all of them. A philosopher would say that this is what a good explanation looks like. Wigner’s word was simpler. He called it a gift.

Part XIV · The other side

Objections

I do not want to win by leaving out what can be said on the other side. None of Kuhn’s guests, with the possible exception of Gödel in Chaitin’s story, would sign what I have written. So here are the best objections I know, several of them from the interviews themselves.

“Then who made God?” This is Davies’s challenge, and he finds “God did it” empty when it is offered as the end of the story. The reply is the one he mentions: God is not a further contingent thing in need of a cause but a necessary being. Davies doubts that the idea of a necessary being makes sense. But look at who is raising the doubt. Everyone in this debate already believes in something that exists necessarily, uncaused and eternal. For Penrose it is the Platonic world. For Tegmark it is the totality of structures. For Wolfram it is the ruliad, which he calls a necessity in so many words. Nobody asks who made the fact that seven is prime. So necessary existence is not the believer’s special problem. It is common ground. Davies says it best himself, with the image of a tower of turtles: every explanation rests on something, and the arguments are really about where to stop. The only question is which stopping point explains the most, and I have just listed what this one explains.

“This is a God of the gaps.” The phrase is fair when God is brought in to cover something science has not yet worked out, and embarrassing when science then works it out. This is not that. I am not pointing to a hole in physics. I am pointing to the thing that makes physics possible at all: that nature is lawful, that the laws are mathematical, and that we can know them. No future experiment can explain why experiments work. That is not a gap in the building. It is the ground it stands on.

“Mathematics comes from the brain.” Lakoff’s account explains our grasp of mathematics and I accept nearly all of it. It does not explain what is grasped. On the view I am defending, his findings are exactly what one would expect: creatures intended to understand the world would be equipped, from the cradle, with the beginnings of number. That an infant already knows one and one make two is not evidence that arithmetic is a human construction. It is evidence that we arrive prepared.

“There are many mathematics, so none is special.” This is Wolfram, and in a different way Hamkins and Jonas. Suppose they are right, and there are countless consistent systems. Each of them still has its own necessary consequences, which nobody chose. What follows from axiom system number fifty thousand follows whether or not anyone ever looks. Multiplying the mathematical universes multiplies the eternal truths; it does not make any of them a human product. It only makes the mind that contains them larger than we had imagined, and makes Davies’s question more pressing. Out of so many, why did this one receive a world?

“Gödel showed that mathematics is incomplete, so it is not perfect.” This is the worry that unsettled Chaitin, and I think it points the other way. What Gödel proved is that in any consistent system rich enough for arithmetic, there are true statements that the system cannot prove. Consider what that implies about the invention theory. If mathematics were a game we made up, there would be nothing to mathematical truth beyond what the rules let you derive. Truth would simply be provability. Gödel showed that truth is strictly bigger than provability, in every system we could ever write down. There is more that is true than any set of human rules can reach. That is not what an invention looks like. It is what you would expect if our systems are finite attempts to chart something that exceeds them. Chaitin’s experimental mathematics, adding axioms as evidence accumulates, is then not a retreat from discovery. It is how explorers behave when the territory is larger than the map. Gödel drew this conclusion himself.

“You have not proved it.” No. A proof in the strict sense is available only inside mathematics, and this is a question about mathematics from the outside. What I have offered is the kind of argument we use everywhere else when we reason well: here are the facts, here are the rival accounts, and this one accounts for more of them with less left over. Jonas put her confidence in Platonism at 87. I will not put a number on mine. I will only say that I find it far easier to believe in an eternal mind than in eternal thoughts that nobody thinks.

Part XV · The verdict

Reading

So, invented or discovered? We need both words, in the proportion Tegmark gave. The language is ours. The numerals, the plus sign, the letter π, the order in which the subject happened to develop, the fashions and rivalries of each century, the choice of which axioms to try first: all of that is human, and Lakoff and Wolfram are good guides to how human it is. Our symbols all have birthdays. But what the symbols say is not ours. No one decided that the primes are endless. No one can add a sixth solid to the five. Those were found, and they were there to be found before there was anyone to look.

And I think “discovered” is still not quite the right word. A continent is discovered: it was simply lying there, and it means nothing. Penrose’s comparison was to an archaeologist, and that is closer. But think about what an archaeologist does on the day he brushes the sand from a stone and finds lines of writing on it. He does not suppose the marks arranged themselves. He does not think he is inventing them, either. He knows at once that someone was here before him, and he sits down and begins to read.

That is what I believe happens whenever a proof closes. Every mathematician reports the same sensation at that moment: a click, and the certainty that it could not have been otherwise and had always been so. The simplest account of that feeling is that it is accurate. The theorem was true before the proof, before the mathematician, before the Earth. It was true because it has always been known.

Euclid’s primes did not wait in the dark for Alexandria. Kepler was right about what he was doing, even in the years when he was wrong about the planets. We invent the alphabet. The text is not ours.

Mathematics is discovered by us because it was first thought by God.

Sources

  • Closer To Truth, Episode 409: Is Mathematics Invented or Discovered? Robert Lawrence Kuhn with Roger Penrose, Mark Balaguer, Gregory Chaitin and Stephen Wolfram.
  • Closer To Truth interviews with Roger Penrose, Max Tegmark, Silvia Jonas, George Lakoff, Gregory Chaitin, Stephen Wolfram (and his later conversation on the ruliad), and Paul Davies, Part 1 and Part 2. The speakers’ views are summarized in my own words; the conclusions drawn from them are mine and not theirs.
  • Images: Wikimedia Commons. All are in the public domain except the photograph of Roger Penrose, by Cirone-Musi for the Festival della Scienza, used under CC BY-SA 2.0.
  • Euclid, Elements, Book IX, Proposition 20, and Book XIII.
  • Eugene Wigner, “The Unreasonable Effectiveness of Mathematics in the Natural Sciences,” Communications on Pure and Applied Mathematics 13 (1960).
  • Plato, Timaeus. Augustine, On Free Choice of the Will, Book II. Leibniz, Monadology, §43–44.
  • Johannes Kepler, Mysterium Cosmographicum (1596) and Harmonices Mundi (1619), Book IV. Galileo Galilei, The Assayer (1623). Isaac Newton, General Scholium to the Principia (1713).
  • Wisdom of Solomon 11:20. Psalm 111:2. Qur’an 55:5 and 54:49.
  • Kurt Gödel, “Some Basic Theorems on the Foundations of Mathematics and Their Implications” (Gibbs Lecture, 1951). P. A. M. Dirac in Scientific American (May 1963). James Jeans, The Mysterious Universe (1930).
  • Hartry Field, Science Without Numbers (1980). George Lakoff and Rafael Núñez, Where Mathematics Comes From (2000). Max Tegmark, Our Mathematical Universe (2014).