Crack the crypto in Agrippa, win every William Gibson book ever published

Quinn DuPont writes in with "A cracking challenge to cryptanalyse a William Gibson poem ('Agrippa', written in 1992). The winner will receive a copy of every William Gibson book published. Project is academic (non-commercial)."

Gibson's poem is a beautiful work, and it came on a floppy disk that erased itself after displaying the poem's text a single time. Of course, it was cracked almost immediately (..f. all DRM, ever) but that wasn't really the point. The challenge site includes a System 7 emulator, an image of the floppy, some of the sourcecode for the app (which was apparently written in Lisp?!), and more.

Based on the pioneering work of Alan Liu and his team at The Agrippa Files, working in collaboration with Matthew Kirschenbaum at the Maryland Institute for Technology in the Humanities and the Digital Forensics Lab, a a bit-for-bit copy of this application has been recovered, along with numerous archival documents.

The first person to successfully crack the code will win a copy of every William Gibson book ever published (except Agrippa). Every runner-up will have their name (if provided) posted on this website. To win you must submit a technical description of your cryptanalysis below, under Creative Commons usage rights (the results of which will be used to further research on Agrippa). The technical description should explain what kind of encryption is used (if any), how it functions, and how it was reversed or cracked (and what the key is, if there is one). Should there be no encryption at all (a possibility), or should the application merely “scramble” or “destroy” the data, this must be technically demonstrated or proved. Since the plain text is known, the cryptanalysis is purely for fun and academic curiosity, and thus the description should provide technical details.

Cracking the Agrippa Code (Thanks, Quinn!)

• [Jaye Sunsurn]

  I thought the original encryption was routed around… ie someone found a way to port the output to a file on its first run (the only time you could see it before the decryption method wiped itself making a re-read somewhat difficult) therefore there was always a cleartext avoiding the need to crack the encryption.  Though cracking it is still cool.

• [Bonobo]

  Anyone who saw “Hackers” knows that the password for The Gibson is “GOD”

• [Joseph Francis]

  You said I could win every book he ever published. Then in the fine print you say I could win a copy of every book he ever published. Hmmm. That still comes out to thousands and thousands of books, which is a pretty good deal, but a lot of them will be virtually identical. I dunno. Is it worth it?

• [Snertly]

  What? No one kept a copy of Copy2PC?

• [Deidzoeb]

  I can’t even decrypt (..f. all DRM, ever)

• [Hamish Grant]

  This was mentioned in a /. post back in 2008.

  The text of the decrypted poem is here – in a differential file showing the version posted on Gibson’s site versus the version on the disk.

  http://agrippa.english.ucsb.edu/images/emulation/diskimage-analysis/freek/agrippa.diff

  here’s the extracted text formatted more nicely.

  http://agrippa.english.ucsb.edu/images/emulation/diskimage-analysis/freek/agrippa-poem-extracted-from-diskimage.txt

• [Quinn DuPont]

  For anyone wondering, the cleartext of the poem is indeed known (dumped from a debugger), but the encryption itself has never been broken or explored. Given that the cleartext is known, hopefully this will make cryptanalysis easier!

• [Matthew Kirschenbaum]

  The cleartext of the poem was circulating on the net within a day (literally) of the work’s public debut in December, 1992 (as detailed on the Agrippa Files site).  That’s not the point of Quinn’s challenge though, which is instead about recovering its alleged crpytographic functions–one more piece of the mechanism.

• [jon_anon]

  Corey, you might make this a bit clearer: the crypto was never cracked, as such. Someone found a way to get the cleartext off the floppy without actually cracking the cryptographic system.

• [Hamish Grant]

  If the text was on the disk already in a format that was accessible, was the encryption ever actually employed and was decryption ever even necessary?

  • [Quinn DuPont]

    We don’t know! Hopefully we’ll discover how the encryption functions (if it does at all–there are some odd hints suggesting otherwise)

  • [SamSam]

    The poem itself is the encypted text. In theory. The challenge is to decrypt the poem to something else, not to decrypt some code into the poem.

• [elix]

  Hm… after the poem is read (in emulation, at least), it inserts 3,000 base pairs into the file. I wonder if those actually code to any actual genes, or if it’s just a particular method of salting the binary with noise.

  Also, in case people haven’t seen, mouseover the “Cracking the Agrippa Code” banner image on the linked site from the post.

• [petr]

  I love LISP, lots of irritating silly parentheses.

• [Spencer Cross]

  I’m going to hazard a guess that most people who will participate in this likely already own (or have at least read) all of Gibson’s books.

• [Fashpo]

  Wow, can’t wait to win all of those translated Gibson novels, a Gollancz first edition of Neuromancer and the very rare Ikea Neuromancer edition: http://www.fashionpolice.dk/wgb/arkangl/ikea.jpg

• [Jeremy Mesiano-Crookston]

  I solved it! The Agrippa code is:

  “Break lines of ordinary prose into fragments with a line break to make poetry”

  Watch! I’ll prove it by decoding Agrippa:

  I hesitated
  before untying the bow
  that bound this book together.

  Now I apply my patented method:

  I hesitated before untying the bow that bound this book together. 

  PrestOOO!