Compete to crack secrets lost to history—for fun, clout, and (no) profit
During the Civil War, Edwin De Leon was a Confederate propaganda agent in Paris. He had $25,000 (~$700,000 today) of secret service money to spend on favorable coverage in the European press.
In November 1863 a Union warship captured the Confederate ship carrying his official letters. The mailbag was thrown into the sea to keep it out of Union hands. It washed up on a beach, someone handed it in, and the Union, in a move calculated to humiliate, printed the letters in its own newspapers.
Letters 1 to 12, before this incident, had gone in plain English. Number 13 came back in cipher. No plaintext or key to it survived in any archive. Until now.
OPINION NATIONAL → L'Opinion Nationale, a paper of the
1860s) that used to attack him.
What happened next: He did not know he had already been fired. His boss signed the recall on 9 December, and it was crossing the Atlantic as he wrote. Letter 14, the following month, asked for more money. Letter 15, on 3 February, carries no cipher at all, and states that he cheerfully accepts the withdrawal of “a commission I have never exhibited, of a title I have never used, and of a salary which I have never accepted.”
Reel 9 frame 97, Despatch No. 13, Paris, 23 December 1863. LOC Pickett Papers MSS16550.
Nothing in the Confederate State Department's own surviving ciphering kit produces these symbols.
But the alphabet was rebuilt in 2026 from 197 symbols.
How did I crack this?
One of the cipher words (“⊡8⊐4⊐⊡4”) repeats its symbols in a shape that exactly one standard English word fits:
OPINION. For a man whose job was bribing European newspapers, the word beside it stood out as
possibly being NATIONAL, and the rest chipped away from there.
There is a lot of that. Unread ciphers sit in open archives, catalogued and scanned. Some give way in an afternoon. Others have consumed thousands of collective hours and returned nothing.
Think you can crack one? Prove it!
Somebody went to the trouble of making sure you could not read it. That is most of the appeal.
Most attempts to decipher go nowhere. Some produce a reading that looks right yet cannot be proved. Some look like a breakthrough, until hours or days later a footnote tells you it was solved and published decades ago.
Now and then one gives, and holds up when you try to break it yourself: the cipher gibberish stops being gibberish, and a man dead for a century tells you something he never wanted you to hear.
The documents are mostly public. Lots of ciphers haven't been cracked because nobody has bothered, or worse, because somebody bothered once, gave up early, and filed it under hopeless.
Both sides sent coded messages between 1861 and 1865. What survives is scattered: some in the government's published war records, some in a handful of specialist books, the rest sitting in archive catalogues, not opened since.
It is the best-documented war of its century, the telegraph pushed both sides into sending far more than they could protect, and the systems they trusted are breakable now. The corpus is also bounded, so it can be finished: catalogue the keys and the code dictionaries, test each one against every message on the same wire, and work outward until the traffic runs out.
Documenting it is the work. New decipherments are simply a byproduct.
The problems are listed here. The submissions page will go live shortly.
You get a documented cipher nobody has read, and enough of a paper trail to start the same day.
Maybe some fame, if you are the first to solve it.
Every problem is either open or closed, and carries exactly one status underneath that, chosen from a fixed list, so a failure cannot quietly turn into a success later. “I tried and it didn't work” and “I ran 4,000 attempts and none scored better than random” are different results. They do not get the same weight.
| Outcome | Meaning | n |
|---|---|---|
| OPEN, meaning meaningful work remains or the record still needs classification | 74 | |
| PARTIAL | Substantial original result, but the registered target's meaningful reading or mechanism remains incomplete | 28 |
| GATED | Blocked on sources, images, transcription, or controls | 19 |
| PAUSED | Intentionally stopped but resumable | 20 |
| QUEUED | Valid target prepared but not yet attacked | 2 |
| NEEDS_REVIEW | Legacy record requires human classification | 1 |
| ACTIVE | Currently being attacked without a substantive result yet | 0 |
| PARTIAL_CRACK | A mechanically controlled fragment survives, but the registered postcard cluster does not have a complete reading | 4 |
| SOURCE_FREEZE_ATTACK_BARRED | Source preparation exists, but the predeclared release conditions were not met and no attack is authorized | 0 |
| CLOSED, meaning the registered target is resolved, exhausted, or deliberately triaged out | 165 | |
| NEW_SOLUTION | This project independently produced a new, defensible ciphertext-to-plaintext solution for the registered target | 38 |
| SYSTEM_RECOVERED | A lost key, alphabet, phase, or mechanism was reconstructed using historically known material | 4 |
| PLAINTEXT_FOUND | A contemporary clear copy or translation supplied the content; the cipher was not broken | 31 |
| PRIOR_SOLUTION | An equivalent cryptanalytic solution was already available | 18 |
| DEAD_END | A proper bounded attack produced a documented negative | 9 |
| SOURCE_LIMIT_CLOSED | The source boundary was controlled as far as the evidence permits, with no defensible complete plaintext or predictive key | 61 |
| MECHANISM_CLOSED_SOURCE_BOUNDED | The mechanism or reading is secure within explicit physical-source limits, without silently upgrading the underlying source | 2 |
| TRIAGED_OUT | The target was deliberately closed as nondiscriminating or not worth further attack | 2 |
| Total registered problems | 239 | |
These are registered problems, including partial results, source leads and documented dead ends. A correspondence series counts once; its individual messages are not extra problems. A closed record is not necessarily a solve.
Every one of these was learned the hard way, by ignoring it first: