Le chiffre indéchiffrable

July 5, 2005 5:49 AM

Plgjoekz xh jiw lwe zqsd meecebefi aqxaxgw xb pzchiottazlq (pbq kvqetnpavckxg) fqrut fegqeifrm nvednsvu ix xzt 9hu kifiuea, efijn dnzx gu tug Vskwcsem gaehrt ic qahogbvaquggd. Lpsxgr li Nxgrpebi vxr awx acvrpt dlw rwcpij (we qgvopgesq i wlgoaieb tgamnttzpbrvim gaevrz), Kadvnp Bkxahhn Jidpsb jan hgcs fw gwcthtiow wpfyqij, xn 1553. Oglkwg'h wzxpwbeavadmgc vnzrwhsrf tri hdkrz sx ihr valydp frkxs ihnv wkw kfinvhwgeq dy dlw dpiqsmh kra pbsygsfamgc os vhyww ivnb gsbe ogfyvw wwz, irv uoe vho jaggg bmet ia uefif wialvws yrcrc, ef jboziszaone msvt qbcpv qe huen. Gzpfymw Tpbocgo wmrqrawxjlya cbeuzsq Dmytnrte psj ivr Jvaiifj devacu gpi Ugizgax Asg, phb ml laf mezx ktqemx mctvn Fbmwsfvkl Cpsvpsum jtjripws hvu moxzdr n liupdr nadij, xb 1863. Gpi hdllclzlsqsgqg uxpugrc, wmrv na xzxs bpe, biepwa wrw df gje veki qblik qrrckkfdt pl i psnprtsyr oxhuwyl p cbopexwg.
posted by Plutor (69 comments total)

Um. Oui?
posted by NinjaPirate at 5:51 AM on July 5, 2005

There are two types of MeFites: Those who will laboriously decipher this, and those who just can't be arsed.

No offense to Plutor, but I am in the latter category. I applaud your tenacity, though.
posted by Faint of Butt at 5:54 AM on July 5, 2005

"Drink ovaltine"
posted by gimonca at 5:56 AM on July 5, 2005 [1 favorite]

I have been trying, it's just that I'm dumber than a builder's elbow.
Kudos for using the "goodluck" tag though; I appreciate the benevolent sarcasm.
posted by NinjaPirate at 5:57 AM on July 5, 2005

That's nothing compared to my rot13 skills.
posted by furtive at 5:59 AM on July 5, 2005

(more addictive than decypher)
posted by NinjaPirate at 5:59 AM on July 5, 2005

"crytography" It's about cold?
posted by peacay at 6:06 AM on July 5, 2005

YOUR MOMMA !
posted by elpapacito at 6:07 AM on July 5, 2005

i thought it was just a vignere cipher. boy do i feel dumb. i also thought that "xn" was "in" because it's right before a year, but then "xb" came before a different one. oh well, that's as much effort as i'm willing to put in.
posted by timory at 6:11 AM on July 5, 2005

The Vigenère cipher was invented by Giovan Batista Belaso in 1553 and was first cracked by Kasiski in 1863.

I tried above using the tags as the cipher key but no luck.

Perhaps the links build the key?
posted by Rothko at 6:22 AM on July 5, 2005

The message reads:

Although it was the most important advance in cryptography (and cryptanalysis) since frequency analysis in the 9th century, every part of the Vigenere cipher is misattributed. Blaise de Vigenere did not invent the cipher (he developed a stronger polyalphabetic cipher), Giovan Batista Belaso had done so centuries earlier, in 1553. Belaso's implementation harnessed the power of the tabula recta that was squandered by the obvious key progression of those that came before him, and for the first time in seven hundred years, an unbreakable code could be used. Charles Babbage successfully cracked Vigenere for the British during the Crimean War, but it was kept secret until Friedrich Kasinski revealed his method a decade later, in 1863. The polyalphabetic ciphers, such as this one, remain one of the last codes breakable by a layperson without a computer.

And no the links don't build the key: it's fairly arbitrary as far as I can tell...
posted by jon4009 at 6:23 AM on July 5, 2005

I tried to decypher the text above, using the tags as the decryption key, but no luck. Perhaps the links build the key?
posted by Rothko at 6:23 AM on July 5, 2005

Kadvnp Bkxahhn Jidpsb =
Giovan Batista Belaso, who "introduced the notion of using a passphrase as the key for a repeated polyalphabetic cipher" [here] in 1553.
posted by pracowity at 6:29 AM on July 5, 2005

wtf?
posted by matteo at 6:30 AM on July 5, 2005

Oops. Should have checked the latest comments.
posted by pracowity at 6:30 AM on July 5, 2005

>:O
posted by jimmy at 6:35 AM on July 5, 2005

Well done link, and a fascinating topic.
posted by doozer_ex_machina at 6:36 AM on July 5, 2005

Wow, jon4009, that was much faster than I was expecting. Did you do a real Kasinski test? Or did you do it the easy way and find some seed words based on the dates and the capitalization (like pracowity did) and get the key from there? Yes, the key was arbitrary. I didn't even think to base it on the tags or a link.

(And I've been misspelling Vigenére and cryptography all damn morning.)
posted by Plutor at 6:42 AM on July 5, 2005

This is just too insular - a link you can't read unless you already possess substantial core knowledge in this field. Doesn't help that it's really long, either. While I like cryptograms more than the average guy, I don't like having to solve them just to read the damn article.
posted by JHarris at 6:45 AM on July 5, 2005

Way cool post, thanks Plutor. It makes me want to write a reply in invisible ink.
posted by nickyskye at 6:45 AM on July 5, 2005

I can't even do the cryptoquotes in the newspaper. When Sunday's Foxtrot cartoon required doing a cryptoquote in order to get the joke, I went straight to google for the answers.

Note: Paige is not in the solutions, so you have to figure that one out on your own. Also I believe the second or third wasn't in there either, so I... well, nevermind.

Also, jon4009: four minutes.. wow.
posted by odinsdream at 6:46 AM on July 5, 2005

This is just too insular

Damn right. How would you like it if I made a long post in Lithuanian or Sanskrit? "Hey, you just need to put in a little work with a dictionary and grammar—come on, it'll be fun!" (I anticipate responses along the lines of "That'd be awesome!" but I disbelieve them in advance. It would be just as annoying as this.)
posted by languagehat at 6:53 AM on July 5, 2005

I totally disagree. This was a creative and fun way to start the work week. A+ post.
posted by Rothko at 6:56 AM on July 5, 2005

shove het omhoog uw ezel, cryptojongen
posted by quonsar at 7:05 AM on July 5, 2005

Alas I'd love to say I was some kind of crypto expert, but no...

After clicking each of the links in turn, the last one seemed the most accessible to me. I googled for some kind of decrypter/cracker and found the applet linked above.

Given the ciphertext, the applet will attempt to calculate possible keys. What threw me initially here was that the applet doesn't handle punctuation or numbers at all, so I eventually figured out I had to remove those from the ciphertext (I probably should have RTFM). Once I'd done that, in the list of several hundred possible keys it generated I spotted a real word in amongst what appeared to be random gibberish. After a little trial and error where I rotated the other letters around and changed one letter, I struck gold and got the key text!

I felt kinda guilty after reading the message about it being breakable without a computer though, sorry :) FWIW, I thought it was a great post.
posted by jon4009 at 7:07 AM on July 5, 2005

A+++++++ LINK WOULD READ AGAIN
posted by killdevil at 7:17 AM on July 5, 2005

I had no intention of breaking the code, but I still think it was a clever post. A bit long for one so arcane, but, nevertheless, good.
posted by OmieWise at 7:26 AM on July 5, 2005

oh, jon4009-you still get the points. Using a computer in this case is the same as using a pencil. It was, after all, a post on an internet board.
posted by OmieWise at 7:26 AM on July 5, 2005

How would you like it if I made a long post in Lithuanian or Sanskrit?

It's not really like that. It was a puzzle, a riddle, and everyone here has the basic skills to solve it. And the post linked to English-language clues, including a link to the tool with which jon4009 (apparently no cryptography expert) cracked it.

It's more like posting something about the history of crossword puzzles but with the post encoded in British-style clues. It would be hard to solve if you weren't used to such things, but it would be fun for those who like crosswords and harmless (but perhaps educational) to those who don't.
posted by pracowity at 7:27 AM on July 5, 2005

Neat post. While working as a photocopyist at a law firm a few summers ago, I invented an extremely inefficient way to encrypt text using 26 matrices. Vignere's method is a hell of a lot faster, but yields the same basic result.

How does one do a Kasinski test by hand?
posted by sindark at 7:28 AM on July 5, 2005

I, too, think it was a clever post.
posted by kenko at 7:29 AM on July 5, 2005

I'm usually against obscure posts but I'll have to disagree with languagehat on this one -- this FPP was obviously encrypted and presented an obvious challenge to those so-inclined (not that I have any great interest in this field, but sometimes it's fun to problem solve). I'd only just worked out how to toggle that histogram solving link and I expect, even with a computer it would have taken me at least an hour.
So well done jon4009 !!
And thanks plutor - it was an innovative and interesting post.
posted by peacay at 7:42 AM on July 5, 2005

Neat post. But the capitalization, dates, and two letter words made it clear (er, no pun on "clear text" intended -- or was it?) it was enciphered English.

Reminds me of the time I deciphered a "personals" ad in a simple substitution cipher, and sent a Vigenère enciphered reply.

Never did here back from the person who took out the ad. Was my enciphered reply too hard to decipher, or did playing cipher one-upmanship just prove i was too nerdy?
posted by orthogonality at 7:48 AM on July 5, 2005

This was awesome.
posted by bshort at 7:50 AM on July 5, 2005

Eh, I tried rot-13ing it and gave up after that, but since someone posted the solution in the thread, who cares?

I thought it was intresting.
posted by delmoi at 7:52 AM on July 5, 2005

If this is such a great post, why is no-one discussing the content of the links?
posted by mendel at 8:15 AM on July 5, 2005

There were links?
posted by odinsdream at 8:29 AM on July 5, 2005

Because the post itself is what's great?
posted by kenko at 8:30 AM on July 5, 2005

.
posted by buzzman at 8:35 AM on July 5, 2005

Thanks, plutor. I'm always interested in a little cryptography stuff, and these links were no exception. I thought it was some link-spider putting up random stuff first when I saw the shopdonkey link. I always wondered for a while about how cryptographers could get at a message just based on frequencies of letters and sequences, and now I know a lot more about it. Don't pay any attention to these lazy, jealous critics.
posted by BlackLeotardFront at 8:49 AM on July 5, 2005

great post! I love crypotgraphy even though I'm really bad at it. Xr ten Xb eleven.
posted by chaz at 9:18 AM on July 5, 2005

Gorgeous and playful post. Informative.
posted by loquacious at 9:52 AM on July 5, 2005

If this is such a great post, why is no-one discussing the content of the links?

Good question. But I'll withdraw my objection, since everyone but thee and me thinks it's the cat's pajamas and the bee's knees.

*begins preparing post in Linear B script*
posted by languagehat at 10:00 AM on July 5, 2005

Languagehat, surely you'd be up for a Linear A post, right?
posted by fatllama at 10:15 AM on July 5, 2005

va bene, domani farò un post in italiano
posted by matteo at 10:26 AM on July 5, 2005

Main hindi-hi me likhunga. Lekin sanskrit nahin - zyada samay hota hai.

Probably wrong, as it's been years even for Hindi.
posted by metaculpa at 10:47 AM on July 5, 2005

Superb post!
I have no idea what this is about, but figure if it involves cryptography that I should be enthusiastic in a futile attempt to sound smart!
posted by spock at 10:53 AM on July 5, 2005

metaculpa : "Main hindi-hi me likhunga. Lekin sanskrit nahin - zyada samay hota hai."

Tsk tsk.

"Mein Hindi hi me likhunga. Lekin sanskrt mein nahin kyunki zyada samay lagta hai"
posted by Gyan at 10:58 AM on July 5, 2005

How does one do a Kasinski test by hand?

One just looks for repeated, same-letter sequences, and then uses the offset between such sequences to try and find a guess at the key length.

Thus if 'IVE' is repeated 5 times, twice 6 apart, twice 12 apart, and once 18 apart, a 6 letter key would be a decent guess. The you just start guessing 6 letter keys.

Before you do that, though, it would behove one to find the index of coincidence to determine if the cipher seems to be mono- or poly-alphabetic (assuming it's a simple substitution cipher at all).

I = 1/[n(n-1)] * Sum_(i=0 to 25) [ n_i(n_i - 1 ) ]. (n_i is the letter freq. for each letter from a = 0 to z = 25.

For English text, it will range from 0.0385 to 0.065, the low end is indicating a poly-alphabetic substitution key as long as the text (or near), and the high end indicating a mono-alphabetic substitution.

I'm not a cryptographer, but I was once made to study it against my will.
posted by teece at 11:12 AM on July 5, 2005

Dhanyavad, guru-ji. That lugna idiom always got the best of me.
posted by metaculpa at 11:36 AM on July 5, 2005

[ihvu ic kgdr]
posted by nervousfritz at 11:36 AM on July 5, 2005

Languagehat, surely you'd be up for a Linear A post, right?

Linear B has Unicode. Linear A has been proposed but not, as far as I know, accepted.
posted by languagehat at 11:41 AM on July 5, 2005

05487 78156 87196 17226 74783
38745 48719 26794 46052 48745
98700 54791 63740 91487 38409
25879 97534 48793 87504 12973
32084 15894 38462 47161 60450
posted by boo_radley at 12:12 PM on July 5, 2005

jon4009: a break is a break, no matter how it's done. That was an amazing tour de force.
posted by warbaby at 12:18 PM on July 5, 2005

Thanks for the lesson jon4009 and Plutor
posted by tke248 at 12:37 PM on July 5, 2005

I absolutely hated this post when I first read it, even when I clicked the links. Then I started to click around, got interested, and completely turned around.

Great post.
posted by bigtimes at 12:54 PM on July 5, 2005

I thought someone had changed the layout of Plutor's keyboard to Dvorak or something without telling him.

I applaud the post, though I exhausted my patience before cracking it.

oh, and LanguageHat, don't even joke about posting in Sanskrit. That's a place I never want to revisit.
posted by Busithoth at 1:13 PM on July 5, 2005

?L TRIDWN AN TKTBW BALLG? AL"RBI?
posted by Hal Mumkin at 1:48 PM on July 5, 2005

Heh. fun. I tried decoding it using "metafilter", "plutor", "asshat" and the key words...

Then finally tried "pancakes".
posted by taz at 1:56 PM on July 5, 2005

???????? ????!
posted by languagehat at 3:01 PM on July 5, 2005

Goddammit, and I replaced it on preview too. That's it -- I'm never posting in Sanskrit again.

*hears sigh of relief from Busithoth's direction*
posted by languagehat at 3:02 PM on July 5, 2005

Nice. I've never cracked a Vigenére by hand, either. Now go and solve the final panel of Kryptos.
posted by blag at 3:38 PM on July 5, 2005

Everything I know about cryptography I learned from "Cryptonomicon" (in other words, I can define "one time pad," but that's about it).

None the less, cool post, if only for the discussion.
posted by brundlefly at 3:46 PM on July 5, 2005

Cryptography is teh new hotness! I guess the person that makes a 30x30 grid of letters and makes their post a word jumble will be lauded as well.

On the upside, Metafilter should be the only ranking for pzchiottazlq in a day or three.
posted by Ynoxas at 7:07 PM on July 5, 2005

this _Ucking ROXX!!!! (I am also of the second category of mefites.)
posted by prettyboyfloyd at 9:48 PM on July 5, 2005

I'm glad that (some of) you guys enjoyed the post. If you're interested in learning more about code-making and code-breaking, I strongly suggest picking up a copy of The Code Book by Simon Singh. The Vigenére Cipher is but a small portion of our cryptographical heritage.

I apologize to those who thought this post was too insular. I'm glad that it was solved quickly and the cleartext was posted, but even without that, the links still worked.

(Just as a reference for anyone who might be interested in posting Vigenére'd HTML anywhere else, the third link will encrypt text, and more importantly, it will preserve HTML, punctuation, and capitalization.)
posted by Plutor at 4:52 AM on July 6, 2005

Cool post. Solved it the hard way. Thanks for the fun.

Yum. Giddy up.
posted by rush at 9:41 AM on July 6, 2005

[This is stellar.]
posted by SteelyDuran at 3:41 PM on July 6, 2005

On the upside, Metafilter should be the only ranking for pzchiottazlq in a day or three.
posted by Ynoxas at 9:07 PM CST on July 5 [!]

Well, I was partially right.
posted by Ynoxas at 9:18 AM on July 8, 2005