The Enigma Machine — How It Worked and Why It Failed

A German cipher clerk in 1941 believed his message was secure because Enigma had 158 million million million possible settings. But the machine had a tell: no letter could ever encrypt to itself. That single quirk, built into the reflector’s wiring, let cryptanalysts rule out thousands of settings per second. The cipher wasn’t broken by cracking its math — it was dismantled by exploiting what it couldn’t do.

How it worked: three shifting layers of substitution

Enigma transformed a plaintext letter into ciphertext through three electromechanical stages, each dependent on the machine’s momentary state.

First, the plugboard (Steckerbrett): up to 13 cables swapped pairs of letters before input reached the rotors. A standard setting with 10 cables created 150,738,274,937,250 possible configurations. The swaps were reciprocal: if A mapped to N, then N mapped to A.

Second, the rotor stack: three (or four, for naval use) rotors, each internally wired in a unique scrambled pattern. The rightmost rotor advanced with every keypress; the others followed via mechanical notches, creating a 16,900-length cycle before repetition. Pressing “E” twice in succession rarely produced the same output.

Third, the reflector (Umkehrwalze): a fixed rotor that bounced the signal back through the stack in reverse. This ensured the encryption process was self-inversing — type the ciphertext, and plaintext lit up. But it also enforced a fatal rule: because the reflector could not send the signal back on the same wire, a letter never encrypted to itself.

Where it shows up: 1939–1945, the war of cribs and captures

1932: Marian Rejewski at the Polish Cipher Bureau deduced Enigma’s internal wiring using permutation group theory — a feat made possible only because German operators transmitted the message key twice (e.g., “ABCABC” encrypted).
Late 1938: Germany increased rotors from 3 to 5 and plugboard cables from 6 to 10. Polish bombas — six linked Enigma clones — could no longer keep up; settings recovery jumped from 2 hours to days.
July 1939: Poland handed reconstructed machines, cryptanalytic methods, and intelligence to Britain and France weeks before invasion.
1940: Alan Turing’s Bombe, enhanced by Gordon Welchman’s diagonal board, automated the testing of cribs. It didn’t find the key — it eliminated the trillions of configurations that violated known constraints.
Feb–Dec 1942: Germany introduced the 4-rotor Enigma M4 for U-boats. Allied decryption collapsed. Convoy losses spiked — 78 ships sunk in November 1942 alone. The blackout ended only after U-boat U-559 was boarded in October 1942; sailors pulled codebooks from sinking water, hours before it went under.

By 1944, 211 Bombes ran at Bletchley Park and outstations, cracking over 90,000 messages per month. Decrypted intelligence, codenamed Ultra, reached Churchill and operational commanders within hours.

What’s contested: Was Enigma mathematically weak — or just fatally misused?

Some argue Enigma’s design was sound for its time — that its compromise resulted solely from German operational blunders like repeated indicators, predictable message formats (“WETTER,” “HEIL HITLER”), and weak keys (“AAA”). The math, in this view, held; the humans failed.

Others counter that the reflector’s self-inverse property created an inherent cryptographic flaw. No amount of operator discipline could fix a cipher that leaked one bit of information per letter: “this is not the plaintext letter.” That redundancy, exploited via cribs, made Enigma’s permutations algebraically vulnerable — not just operationally brittle.

The debate hinges on whether a cipher’s trustworthiness should depend on perfect use. Modern cryptosystems assume bad operators. Enigma did not.

Why this has to do with other realms

The Bombe’s logic shares DNA with protein folding prediction in dest alpha centauri. Both solve NP-hard constraint satisfaction problems at scale: the Bombe eliminated invalid rotor settings by propagating electrical contradictions through a crib-derived network; AlphaFold eliminates impossible protein configurations by enforcing physicochemical laws. One used relay circuits in 1943, the other deep learning in 2020 — but both demonstrate that intelligent search through astronomical spaces is feasible when you know what must be false. The same principle applies to exoplanet detection: we don’t see the planet, we see what its presence cannot allow in the star’s light curve.

An open question

If Enigma’s self-avoidance flaw were fixed — say, by replacing the reflector with a fourth stepping rotor — would German naval ciphers have remained unbroken until 1945?

Key Sources

Rejewski, Marian — "How Polish Mathematicians Deciphered the Enigma" (1984, reprinted in Annals of History of Computing) — primary account of the 1932 breakthrough.
Hodges, Andrew — Alan Turing: The Enigma (1983) — definitive biography, details the Bombe’s logical structure and wartime context.
Welchman, Gordon — The Hut Six Story (1982) — firsthand explanation of the diagonal board’s impact; declassified in part in 1986.
Bauer, Friedrich L. — Decrypted Secrets (1997) — technical analysis of Enigma’s cryptographic properties and vulnerabilities.
GCHQ — 1999 declassification memo — confirms timeline of Polish handover and early Bletchley achievements.

Abhishek