"The E. A. Poe Cryptographic Challenge"...

While looking at Elonka's list of unsolved cipher mysteries while composing my post on the d'Agapeyeff cipher, my eye was drawn to the list of solved cipher mysteries she appended to it, and in particular to "The E. A. Poe Cryptographic Challenge".

Edgar Allan Poe often used codes and ciphers in his stories, most famously in "The Gold-Bug" (which incidentally inspired a very young William Friedman to take up an interest in cryptography). He also asked readers of one popular magazine to send him their ciphers to crack: which he (allegedly) managed to do for the hundred such that arrived.

However, in 1839 Poe published two tricky cryptograms allegedly by "Mr. W. B. Tyler" (probably a Poe pseudonym) which nobody at the time was able to break. These were rediscovered in 1985 by Professor Louis Renza, who then tried to raise their profile: before too long (in 1992), Professor Terence Whalen managed to solve the first one, which turned out to be nothing more complex than a simple monoalphabetic cipher.

The second (still-unbroken) cipher attracted the attention of Professor Shawn Rosenheim, who not only described it in his book The Cryptographic Imagination: Secret Writing from Edgar Poe to the Internet (Johns Hopkins, 1997), but also put up a $2500 prize to attract solvers' attention, with the help of Jim Moore of bokler.com who built a website to promote it.

And then, after Rosenheim and Moore had fielded hundreds of fruitless emails and responses, a software engineer from Toronto called Gil Broza finally cracked the second cipher in October 2000: his decryption is detailed here.

For followers of the Voynich Manuscript, this makes for fairly depressing reading: neither of the "W. B. Tyler" ciphers were, even by the standard of Milanese ciphers circa 1465, particularly tricky, yet Broza had to work really quite hard to solve the second one. He worked out his own transcription, wrote his own software... and then still basically had to break into it by hand, a process made even more difficult by the presence of errors in the ciphertext (which were probably introduced in the typesetting). And people wonder why modern supercomputers can't unravel the secrets of Voynichese - a cipher that is ten times harder than the second Poe Cipher.

The real mystery about Poe is actually the manner of his death: but that's an intriguing story for another day... :-)

Review of "PopCo"...

I'm a lousy fiction reviewer, probably for two main reasons: (1) creative writing classes taught me how to spot when writers are cheating (in order to make me a more honest writer myself); and (2) years of Voynich Manuscript-related research has made me constantly alert for infinitesimal details upon which the answer might just hinge.

Put these two together (a lie-detector and an adrenaline-fuelled eye for detail), and you have a completely unfair toolkit for reading novels, simply because novels are very rarely actually "novel" - they're more often an assembly of ideas.

Take Scarlett Thomas' "PopCo" (FourthEstate, 2004), for example. Superficially, it's like a 500-page anagram of my life (BBC Micro / chess / maths / philosophy / Godel's Incompleteness Theorem / videogames / business / marketing / cryptography / cryptology / secret history / Voynich Manuscript / etc), together with a load of other untaken doors (Bletchley Park / SOE / crosswords / vegetarianism / vegan / Go / low-level drug-use / homeopathy / etc), and it's written quite well: so I really should be engaged by it, right?

Problem #1 is one of construction: the first tranche is basically Douglas Coupland (specifically Microserfs), the second tranche Iain Banks (his fiction rather than his science fiction), then a bit of Martin Gardner's puzzle columns and Simon Singh's The Code Book: there's a kind of teenage girls' magazine section along the way, and a rather clunky historical pirate romance, before it all flips out into Thomas' fictional take on Naomi Klein's No Logo... Yet to me, a book needs to be more than merely a collage of influences, a narrated scrapbook: but perhaps that makes me too old-fashioned for contemporary fiction. If you wanted to be kind, you might compare it with Kurt Schwitters' Merz, carefully arranged collections of found objects (forged Merz pieces get placed on eBay all the time): but sorry, Thomas is no Schwitters.

Problem #2 is the lack of parents. The other day, while watching (the original TV series of) Batman on BBC4, my four-year-old son asked me where Batman came from. Well, I said, a man called the Joker killed both Bruce Wayne's parents, and when a bat bit him in the caves beneath his mansion, he somehow gained a super crime-fighting ability. OK... so where did Spiderman come from? Well, I said, after both Peter Parker's parents died, he was bitten by a radioactive spider, and gained amazing spider-like powers. My son paused, looking back at the screen. But what about Robin, he asked. No, don't tell me, I know: both his parents were killed... Before he had a chance to say "(and he was bitten by a radioactive robin)", I suggested we look Robin up on Wikipedia (though sadly he was basically correct). In PopCo, the main character Alice Butler is basically Crypto Girl, a sort of Elonka-lite: her mother dies and her dad runs away, and she gains her m4d cryptological and prime factorisation sk1llz from her grandad. Put it that way, and it all looks a bit comic-book thin, doesn't it?

Problem #3 is that I'm wise to novelistic conceits. I know that in a cryptological novel, someone called A[lice] is going to communicate with someone called B[en], who will pass on what she says to someone called C[hloe]: and this kind of spoils it. Incidentally, Ron Rivest denies that he used "Alice" and "Bob" (in his 1978 paper introducing RSA public-key cryptography) in any kind of homage to the film "Bob & Carol & Ted & Alice" (which is actually a bit of a shame). It would also have been cool if PopCo's Alice had been born in 1978 and openly named in crypto homage to Ron Rivest's paper, but I think she's too old (is she 29? I can't find the page, rats!).

Problem #4: cringeworthy logic/maths puzzles. To give texture to her story, Thomas brings together loads of lateral puzzles and mathematical ain't-that-amazin' fragments, the kind of thing that you sometimes hear being trotted out at student parties. For example:-Two men go into a restaurant and order the same dish from the menu. After tasting his food, one of the men goes outside and immediately shoots himself. Why? (p.109) The explanation given for this in PopCo is ludicrous (it involves an albatross and a dead child, don't get me started): but why is one not simply a food-taster for the other? Fugu: mmm, delicious... hey, what's that trainee doing in the kitchen... aaaarrgggh!

Problem #5 (probably the biggest of all for a Voynichologist) is that PopCo uses the Voynich Manuscript as a MacGuffin (or do I mean a "Philosopher's Egg MacGuffin"?). Alice's grandfather spends years on the VMs, and even gets her to count the words and letters on each page (and later to factorise large numbers): perhaps washing his car would have been a better way to earn pocket money. Alice says that she's learnt so much from the journey, from the search for the heart of the VMs: but really the manuscript is no more than occasional wallpaper for the narrative. The Beale Papers also make a brief appearance: my guess is that Scarlett Thomas would have used them as the central hook, had there been more than a paltry $20million dollars' worth of treasure linked to them: the alternative "Stevenson/Heath" pirate cipher mystery Thomas constructs is a bit thin when held up against real ones, regardless of the size of its haul.

...and so on. I feel in a bad place: I really wanted to like PopCo, but all I can do is whinge (and I haven't even moaned about her merging Alberti's and Vigenere's cryptography, etc). Other reviewers (such as here and here) seem basically to like the book: and compared to Dan Brown's Digital Fortress (where I wanted to kill all the main characters by the end of Chapter One, all the minor characters by the end of Chapter Two, and the publishers by the end of Chapter Three) it's Shakespeare.

Cryp-lit like this requires a certain kind of technical devotion from the reader, and if you are a diehard crypto-geek PopCo is something you really ought to read. But only if you've read the good stuff (like Neal Stephenson's excellent Cryptonomicon) first.