Can this be decrypted?
August 4, 2006 10:52 AM   Subscribe

Is Scientific American's spell-checker broken? (via justkevin)

This post title has the structure of ciphertext:

Gbalf Xozmn Ram Rqzyk Wtacu Lkugc Aaxjx Owkyu Dkoxk Zamdg Bnuio Nmrxk Zmqyf Nqeog Ziqxf Gutxe Nkmxd Gzmqj Brqge Kxkfs Qqzui Nactg Djfnq Eenaa Xjnk

Are there strategies for analyzing and decoding this text?
posted by Blazecock Pileon to Sports, Hobbies, & Recreation (8 answers total)
Best answer: Yes.
posted by [expletive deleted] at 11:00 AM on August 4, 2006

I could swear that post in the blue was deleted before......
posted by inigo2 at 11:03 AM on August 4, 2006

Yeah, it was. Matt thought it was just jibberish.
posted by [expletive deleted] at 11:06 AM on August 4, 2006

[expletive deleted] could you work through the process you used to get to the 'gm' key? I get the Vigenere bit after that, but not how you worked out the key on the way...
posted by togdon at 11:39 AM on August 4, 2006

Nevermind, I see that they're his initials. Although it doesn't seem that you worked it out that way, so if you managed to do it some other way I'm still curious...
posted by togdon at 11:41 AM on August 4, 2006

Todgon, someone already pointed out that there are two repeated sequences, zmq and aaxj. Because of the nature of the vigenere cipher, sequences of letters, especially three letter sequences, happen whenever a word or sequence in plaintext reappears at the same frequency as a multiple of the key length, resulting in repeated lengths of ciphertext.

Both people who found this assumed the key was 13 characters, but this didn't jibe with the letter frequency, which should be flatter with such a long key, so I assumed the key was very short, along the lines of 2 characters, which was a multiple of the distance between repeated segments.

Next, I went about testing short keys on the aaxj sequence. I used a Vigenere Table. I would use the ciphertext and guessed plaintext to work back to a key. My first guesses were things like that, what and text. I spent about ten minutes guessing, and eventually I tried ould, after realising that I'm not just looking for common four letter words, but sequences. It was my first non-word guess, and I got it.
posted by [expletive deleted] at 2:06 PM on August 4, 2006 [1 favorite]

Togdon, sorry.
posted by [expletive deleted] at 2:10 PM on August 4, 2006

Oh, I left out something important. K shows up with great frequency, and typically in odd numbered characters in the sequence. If k is e, then the cipher is two letters and starts with g, so I was left with o_l_ on the aaxj block. I wondered what word that could be for a second, then I clued in that it was probably could, would, should, etc.

If that failed, I would have just brute forced it, starting with ga, gb, gc, etc, until I found it.
posted by [expletive deleted] at 3:41 PM on August 4, 2006

« Older Help Me Take a Craft   |   QuickBooks Print Designer Newer »
This thread is closed to new comments.