Shannon's Bet: The Cipher That Cannot Be Broken

Jul 26

# Shannon's Bet: The Cipher That Cannot Be Broken

Every cipher so far in this series eventually fell — not always to brute force, but to some leftover pattern a patient analyst or a fast enough machine could exploit. In 1949, **Claude Shannon** — better known as the founder of information theory — asked a more fundamental question: is there a limit to this? Can a cipher be built that leaks *nothing at all*, no matter how much time or computing power an attacker has?

His paper, *Communication Theory of Secrecy Systems*, gave cryptography its first rigorous mathematical foundation. And remarkably, the answer to his question was yes — with a catch that would define the rest of cryptographic history.

## Defining "Unbreakable" Precisely

Before Shannon, "secure" was a loose, almost aesthetic judgment — a cipher was considered strong until someone clever broke it. Shannon reframed security in terms of **information**: does the ciphertext reveal *any* statistical information about the plaintext at all?

He defined **perfect secrecy** as a mathematical condition: for every possible plaintext and every possible ciphertext, the probability of that plaintext producing that ciphertext is exactly the same as the prior probability of that plaintext occurring at all. In plain terms — seeing the ciphertext teaches an attacker *nothing new whatsoever* about what the message might say. Not "very little." Nothing.

This is a much stricter standard than "hard to break." A cipher can require billions of years of computation to crack and still, technically, leak a sliver of information. Perfect secrecy allows for none.

## The One-Time Pad: Proof It's Possible

Shannon showed that perfect secrecy is achievable — with the **one-time pad**. The method is simple: generate a truly random key exactly as long as the message, use it once to combine with the plaintext (typically via XOR or modular addition), and never reuse it.

Because the key is truly random and as long as the message, every possible plaintext is equally consistent with any given ciphertext. A three-letter ciphertext could just as easily decode to "YES" as "NO" — nothing about the ciphertext favors one over the other, given a perfectly random key. There's no frequency trap here, no repeating structure, nothing for a Kasiski-style attack to grab onto, because there's no reused pattern at all.

## The Catch That Changed Everything

If a provably unbreakable cipher exists, why doesn't everyone just use one-time pads? Because Shannon's proof came with unavoidable conditions:

- The key must be **truly random** — not computer-generated pseudorandomness, but genuine entropy.

- The key must be **at least as long as the message itself.** Encrypting a terabyte of data requires a terabyte of random key.

- The key must be used **exactly once** and then discarded.

- Both parties need a way to **share this enormous key secretly** in advance — which is, ironically, the exact same problem encryption is trying to solve in the first place.

In other words: perfect secrecy is real, but it simply relocates the hard problem rather than solving it. Instead of "how do we hide a message," it becomes "how do we secretly transport a key as large as the message, every single time." For anything beyond occasional short, extremely high-stakes communication (the one-time pad did see real use — Cold War diplomatic channels, for instance), this is completely impractical.

## Shannon's Real Legacy: A Different Kind of Bet

Shannon's paper effectively split cryptography into two philosophies going forward:

1. **Information-theoretic security** — provably unbreakable, but only achievable under the one-time pad's brutal logistical constraints.

2. **Computational security** — not proven unbreakable, but secure *in practice* because breaking it would require an infeasible amount of computation, even with the best-known methods and hardware.

Nearly every cipher used today — AES, RSA, elliptic curve systems — is a bet on computational security. They don't claim mathematical perfection. They claim that reversing them, with everything currently known about mathematics and computing, would take longer than the universe has existed. That's a different, humbler kind of promise than Shannon's perfect secrecy — but it's the only kind that scales to a world sending billions of encrypted messages per second.

Shannon didn't hand cryptographers an unbreakable cipher to use everywhere. He handed them a **ceiling** — a precise definition of the best that's mathematically possible — and made clear that everything built afterward would be a deliberate, calculated compromise beneath that ceiling.

*Next in this series: how that compromise split cryptography into two camps — symmetric ciphers like AES, and the public-key systems like RSA that finally solved the key-sharing problem Shannon's one-time pad could never escape.*

Shannon's Bet: The Cipher That Cannot Be Broken

One Key or Two? Symmetric Ciphers, Public Keys, and the Problem Nobody Could Solve

Machines That Think in Letters: Enigma, Bletchley Park, and the Birth of Computing

Your Site Title

email@example.com