Here is the quiet fact at the center of the most-used cryptography on earth: the number is public, and its two prime factors are the secret. You can publish the product. Print it on a billboard, hand it to your adversary, broadcast it to the world. What must never leak is the pair of primes that multiply to it. Whoever holds those two numbers holds the key; whoever holds only the product is locked out. The whole edifice rests on the gap between having the product and having the pair.
And the strange thing — the thing worth sitting with — is that no one has proved that gap is real. There is no theorem that says factoring is hard. We have tried for half a century to find a fast way to recover the primes from the product, and failed; that is strong evidence, but evidence is not proof. So if the hardness is unproven, why does anyone trust the lock? The honest answer is not what most people assume. The lock does not hold because the wall is infinitely high. It holds because of a race.
ICheap to make, dear to break
The asymmetry that runs the whole machine is this. To make a hard problem costs almost nothing. Generate two large primes — a fast test confirms each one in milliseconds — and multiply them. You now hold a number that may take longer than the age of the universe to factor, and you built it in the time it takes to blink. To break one runs the other way: ruinous, astronomical, growing faster than exponentially as the primes get larger.
So the two sides of the lock are not evenly matched, and they were never meant to be. The defender's move is cheap and the attacker's move is dear. That single asymmetry — trivial to forge, brutal to invert — is the engine. It means the defender can keep paying for new locks forever, while each old lock costs the attacker a fortune to pick. The defender is not richer or cleverer. The defender simply has the cheaper move, and makes it again and again.
The wall does not have to be unclimbable. It only has to be higher than the attacker can climb before the thing behind it stops mattering.
IIThe race
Picture it as a cat and a mouse. The cat forges secrets — two fresh primes, a new product — cheaply and constantly. The mouse grinds at the old products, trying to recover the primes that built them. And here is why the cat stays ahead: while the mouse is still tunnelling through last year's lock, the cat has already forged this year's. By the time the mouse breaks an old number, the secret it guarded is stale — the key retired, the message long since irrelevant. Real systems are built exactly this way. Keys expire and rotate on a schedule, precisely so that a break, when it finally lands, lands on nothing that matters anymore.
And when the mouse gets faster — when raw computing power finally catches an old key size — the cat has already moved the standard. This is not hypothetical; it is the recorded history of the field. Keys that were once safe at a few hundred bits were broken, so the standard climbed to a thousand, then two thousand, then four. Every time the mouse's machines caught up to yesterday's wall, the cat had already built a taller one. The defender forges harder problems as fast as the attacker forges faster tools, and the cheaper move keeps winning the race.
So the security was never the security people imagine. It is not a fortress, standing forever. It is a relay — the defender handing off to a fresh, harder problem before the attacker reaches the last one. The lock holds not because it cannot be broken but because it keeps being replaced, faster than it can be broken. Temporal security. A race the defender wins because forging is cheaper than breaking.
IIIWhen the rules change
The race holds on one assumption: that the mouse only ever gets faster, never fundamentally different. As long as the attacker is grinding the same way with bigger machines, the cat wins indefinitely — cheaper move, taller walls, fresh locks. The race breaks only if the mouse gains a new kind of move — not a faster version of the old grind, but a different physics of attack.
That move exists, on paper. A quantum computer does not climb the wall faster; it dissolves the part of the wall that factoring rests on, reading the primes out of the product by a means no classical machine can imitate. It is not a quicker mouse. It is a different animal. And so today's defenders are doing the most far-sighted thing the race has ever asked of them: forging an entirely new kind of lock — built on different hard problems, ones the quantum animal cannot dissolve — before that animal has even fully arrived. The cat is building tomorrow's wall against a mouse that is still mostly a rumor, because some attackers are recording today's secrets now, betting they can break them later. The race never ends. It only changes what is being forged.
The cat needs the mouse. A lock no one attacks is the dangerous one.
And that is the part worth ending on, because it is the truest. The attacker and the defender are not enemies in the way the story suggests — they are the two legs the whole field walks on. Every break teaches the defenders where the weakness was; every defense teaches the attackers where to look next. Cryptography is more trustworthy because people keep trying to break it in the open, not less. The locks we trust are exactly the ones that have survived being attacked honestly, in daylight, by everyone who cared to try. The health of the thing is in the contest continuing — both sides sharpening, both sides honest, the race staying open. The secret is two primes. The security is the race. And the race only means anything as long as it is still being run.
A note on what is and isn't claimed: this essay assumes the practical security model actually used, not a proof of hardness — because there is no proof of hardness. Everything here is conditional on factoring remaining costly in practice, which is an empirical bet, not a theorem. That bet has held for decades against enormous effort, which is why it is trusted; it is not why it is certain. The quantum shift is precisely the case where the bet's assumptions change, which is why the move to post-quantum cryptography is already underway. Treat all of it as how the lock survives in the world we compute in, not as a guarantee about every world.