Where exactly does mathematical beauty live? Is math a truth discovered or a game invented? And why does the rote drill taught in school feel like a different thing entirely from the "beauty" mathematicians talk about? Four books, four answers.
2026 · Book Recommendations · Issue 20
Most people remember math as formulas and exam questions, yet mathematicians speak of beauty. Each of this issue's four books seizes a different mechanism. Hardy's A Mathematician's Apology gives the criterion: a pattern is beautiful when it is deep, surprising, and inevitable—"beauty is the first test." Lockhart's A Mathematician's Lament argues school teaches math backwards; the real beauty lies in the act of asking and proving things yourself. Cheng's How to Bake Pi dissects abstraction—decluttering the mind so you can see the structure different things share. Zhang Jingzhong's The Mathematician's Eye uses ordinary triangles and areas to show what that "eye" sees, and insists math can be made easy.
| Book | Author | Year | The one thing it nails |
|---|---|---|---|
| A Mathematician's Apology | G. H. Hardy | 1940 | A mathematician is a "maker of patterns"; beauty is the first test of good mathematics |
| A Mathematician's Lament | Paul Lockhart | 2009 | Math is an art of the imagination, taught in schools as a hollow shell of rules without reasons |
| How to Bake Pi | Eugenia Cheng | 2015 | Math isn't about numbers but about how things work—abstraction is decluttering the mind |
| The Mathematician's Eye | Zhang Jingzhong | 1990 | Look at ordinary things with a mathematician's eye and a hidden order appears; math can be made easy |
Hardy (1877–1947) was Britain's greatest number theorist, yet he wrote this little book at 62, convinced his creative powers were gone. He set out to answer a plain question: how does a man who gave his life to "useless" pure mathematics justify himself? His answer is not utilitarian but aesthetic—mathematics is worthwhile, first of all, because it is beautiful.
His central claim: a mathematician is of the same kind as a painter or a poet, a "maker of patterns," differing only in that the medium is ideas. And the first test of a mathematical pattern is its beauty—made of depth (the more fundamental the structure, the more beautiful), unexpectedness, inevitability (once seen, it could not be otherwise), and economy (the most achieved with the least).
He offers two proofs over two thousand years old as exemplars: that √2 is irrational (the Pythagoreans) and that the primes are infinite (Euclid). Both are short, both proceed by contradiction, both reach conclusions that are surprising yet, once grasped, feel self-evident. This, Hardy says, is the anatomy of beauty—and the proofs are "as fresh and significant as when they were discovered."
Deeper still, he is a mathematical realist: 317 is a prime not because we think so, but because it is so. For Hardy the mathematician is a discoverer, not an inventor; the patterns exist in an objective mathematical reality and we merely catch sight of them. Beauty, then, is not decoration but the mark of truth.
Hardy stakes all value on "pure uselessness," deliberately belittling applied mathematics—which reads as extreme today. The book is also steeped in a late-life melancholy ("mathematics is a young man's game"). But precisely because his stance is extreme, his distillation of beauty comes through unusually clear. Read it at a distance.
Hardy's "beauty is the first test" transplants directly into engineering and architecture. Try next week: take a system design or a piece of core code and ask Hardy's question—is it ugly? Redundancy, special-case patches everywhere, logic that only resolves after winding detours—that is ugliness; depth, surprising simplicity, the symmetry where one change moves the whole—that is beauty. Treat "ugly solutions have no permanent place" as an early signal: ugliness is often the precursor of wrongness. Concretely—list three "ugly spots" in your current architecture and force yourself to find a more symmetric alternative for each, even if you don't ship it.
Lockhart opens with a nightmare parable: suppose music were taught the way math is—children spend years memorizing notation and scales and copying terminology, never allowed to touch an instrument or hear a single piece, on the grounds that they must "master the fundamentals before they earn the right to enjoy music." The result would be a nation that loathes music. This, he says, is exactly what we do to mathematics.
His central claim: mathematics is an art, an art of the imagination. Real mathematical activity is facing a problem—often one you posed yourself, say "when does the rectangle inscribed in a triangle have the largest area"—then guessing, trying, getting stuck, and finally arriving at a beautiful argument. The joy is entirely in that struggle and its release, which is exactly the part school deletes.
The disease lies in teaching only the what and stripping out the why: students memorize formulas and drill problem types, never learning where the rules came from or why they hold. A proof, which ought to be "the poetry of mathematics"—a convincing story about why it could not be otherwise—is degraded into a two-column fill-in-the-blanks ritual. Mathematics is reduced to an empty shell.
His constructive proposal: math class should look more like art class—give students real, enticing problems and let them explore, err, argue, discover. A beautiful proof is not a format but a successful explanation. Beauty lives in the moment of "I figured out for myself why."
Lockhart's critique is sharp but his constructive side is idealized: he says little about how to implement this under large classes and standardized tests, and dismisses all mechanical practice as harmful—overlooking that some fluency is necessary scaffolding for free exploration. Read it as inspiration, not an operating manual.
As a parent of a school-age child, this book's reminder is the most direct. Try next week: stop adding drill, and offer a "Lockhart problem" instead—no single right answer, hands-on: e.g. "with 12 equal matchsticks, which enclosed shape has the largest area?" or "climbing a staircase one or two steps at a time, how many ways are there?" Guess, try, and argue alongside her; the point is not getting the answer but the moment she states a reason of her own. Remember: when a child asks "why," that is where mathematics begins—don't cut it off with "just memorize it first."
Eugenia Cheng (b. 1976) works in category theory—"the mathematics of mathematics"—and performs as a concert pianist on the side. She wants to dispel a misconception: math is not about numbers and computation, but "the logical study of how logical things work." What matters is process and relationship, not calculating fast.
Baking runs through the whole book: recipes, ingredients, methods. The real skill in cooking is not any one dish but mastering transferable methods. Math is the same—its true objects are the general structures abstracted out of piles of concrete examples.
The book's core move is abstraction: abstraction is decluttering the mind, putting aside the details you don't need right now—like clearing away the utensils you won't use before cooking, so the counter is clean and the structure becomes visible. Once you abstract, you find that "7 plus 5," "5 hours past 9 o'clock," and "turning a shape two steps then three"—may all share the same structure.
This is the categorical view: don't fixate on concrete objects, study the relationships (arrows) between them. It lets you see cross-domain analogies—the same abstract structure recurring in arithmetic, in symmetry, in music. For someone drawn to cross-disciplinary mapping, it is a handy tool: abstract first, then recognize "this is the same thing I've met elsewhere."
For the sake of accessibility the first half is light and easy, but when the second half enters category theory in earnest the threshold rises sharply and the general reader can fall behind; the baking analogies sometimes trade precision for warmth, and specialists will find them a touch too sweet. Take it as a door into abstraction, not a textbook on category theory.
"Abstraction = decluttering the mind" is almost the definition of architectural design. Try next week: take a concrete problem you're stuck on (a distributed-consistency bug, an AI workflow) and run Cheng's abstraction—strip away the business details and ask only "what class of problem is this, really? Have I seen an isomorphic structure elsewhere?" Often you'll find your problem is the same abstraction as "how do a set of nodes agree on a single value"; recognizing that abstraction usually lets you borrow a mature solution from another field. Make "abstract first, then transfer" your first reflex on any new problem.
Zhang Jingzhong (b. 1936) is an academician of the Chinese Academy of Sciences, a mathematician who has poured his life into mathematics education. The Mathematician's Eye is slim, yet the great geometer S. S. Chern, having read it, recommended it be translated into English. What it demonstrates is exactly the "beauty" Hardy and Lockhart describe in English—shown here in a Chinese voice.
His core idea is "educational mathematics": not the study of how to teach math, but the project of reshaping mathematics itself so it is better suited to teaching and learning. In his view much math is hard not in essence but because predecessors never tidied it up; math can be made easy.
He offers "four easies" as method: once it's familiar, it's easy; once it's simplified, it's easy; once it's understood, it's easy; once it's made intuitive, it's easy. The book demonstrates with ordinary problems—why must the three altitudes of a triangle meet at a single point? How does the single notion of area unify many seemingly unrelated theorems? Look with a mathematician's eye and a startling order appears beneath the clutter.
The "eye" is a way of seeing: first make the familiar strange (wait—why should three altitudes meet at one point?), then make it unified (ah, the same structure underneath). The layperson sees a heap of scattered formulas; the mathematician sees one structure reappearing in different disguises. Beauty lives in that moment of "so that's why."
The book is aimed at young readers, so its depth and systematic coverage are naturally limited and seasoned readers will find it thin; some passages and typesetting show their publication era. But as a model of "an original Chinese work that brings math to life," its value lies not in difficulty but in demonstrating an approachable, hands-on sense of mathematical beauty.
"Making the hard easy" is the core craft of technical leadership and mentoring. Try next week: pick a concept your team finds hard (an algorithm, a system, an AI mechanism) and run it through Zhang's "four easies" as a diagnosis—is it hard because it's unfamiliar (lacks examples), not simplified (not decomposed), not understood (missing the motivation/why), or not intuitive (lacks a picture)? Treat the cause, instead of vaguely declaring "this is hard." Much "hardness" is not intrinsic—it's just untidied. Make these four a checklist for writing docs and onboarding people.
Hardy said "ugly mathematics has no permanent place." Use it as a technical sense of smell: a solution that survives only on layered special cases, patches, and detours has probably chosen the wrong abstraction. Self-check—can you state in one sentence "why it could not be otherwise"? If you can't, or the more you explain the more tangled it gets, that's often not because you're slow but because the design itself is unlovely, and very likely wrong. Beauty here isn't fastidiousness; it's an early warning.
Lockhart says: strip out the "why" and knowledge becomes a hollow shell. The test is simple—can you derive it from scratch and explain it to a layperson, rather than reciting the conclusion? If you can only repeat the result and not the reason, what you hold is an empty shell that fails the moment the problem is slightly deformed. The mark of having truly learned something is that you could rediscover it—even having forgotten the answer, you could think it up again.
Cheng and Zhang point from two ends at the same thing: get the abstraction right and the difficulty dissolves. Faced with a hard problem, don't grind—step back and ask, "stripped of all concrete detail, what class of problem is this? Have I seen an isomorphic structure in another field?" Much hardness is being stuck at the concrete level, not yet abstracted to the height where you can see it connects to a problem you already know. Find that abstraction and you've often borrowed a ready-made solution from elsewhere.