Join 3,523 readers in helping fund MetaFilter (Hide)


I hesitated / before untying the bow
July 12, 2012 6:59 AM   Subscribe

In 1992, renowned sci-fi author and futurist William Gibson (Neuromancer, Virtual Light) released Agrippa (A Book of the Dead), a self-playing poem contained on a floppy disk for old Macintosh computers that, once its text had scrolled up the screen one time, would be rendered unreadable on purpose. Now, 20 years later, a PhD student at the University of Toronto is enlisting the aid of cryptographers in hopes of figuring out how the program works.

Agrippa Previously
Gibson Previously
posted by Potomac Avenue (24 comments total) 17 users marked this as a favorite

 
Here's what it would have looked like.
posted by Potomac Avenue at 7:01 AM on July 12, 2012 [3 favorites]


Well, if I'd written a program like that, I'd have just overwritten the disk with random data. I doubt very much there's anything useful left on a disk that's been run.

It's been recorded, so meh, no big deal if the actual floppy is no longer extant. I see little reason to go wildly out of one's way to try to preserve art that deliberately self-destructs.

I am a little surprised that nobody copied the disk before running it, though. Or used the write-protect tab.
posted by Malor at 7:30 AM on July 12, 2012 [1 favorite]


I am a little surprised that nobody copied the disk before running it, though.

They did. You can get your own bit-level copy of the original.
posted by jedicus at 7:38 AM on July 12, 2012 [3 favorites]


He should have just put it on a SyQuest disc. Default self-destruct.
posted by davebush at 7:41 AM on July 12, 2012 [11 favorites]


Saw this on slashdot yesterday. A poster there noted that the resulting file had the ASCII combinations of GTAC.
posted by humanfont at 7:45 AM on July 12, 2012 [3 favorites]


It's a lovely poem, and I'm glad that someone screenshotted it and I understand the impulse to want to crack the code, but I also understand why Gibson made it self-destructing in the context of the poem itself. I've been reading his work lately (starting with the first two books in the Sprawl trilogy, which I've already read--I'll also re-read Mona Lisa Overdrive when I find my copy--and right now I'm about 2/3rds of the way through All Tomorrow's Parties), and I'm very much enjoying the way that he describes very particular details of the worlds that his characters inhabit, and this work very much plays to that strength.
posted by Halloween Jack at 8:17 AM on July 12, 2012 [1 favorite]


William Gibson has commented a bit about it on twitter:

Did they ash him for help? [Nope]

I was never entirely convinced that the text had actually been encrypted, myself. The whole project was more than a little fugitive, dodgy.
posted by rajbot at 9:08 AM on July 12, 2012 [1 favorite]


i love how coy gibson is about this, at least publicly. the tweets have an air of dry amusement (if tweets can have an air).

I remember at the time being a little horrified by the idea on one hand, and on the other suspecting that it was a stunt. I didn't have much exposure to Gibson at the time, just to Gibson fans, and the hype around his stuff was annoying as shit. Since then, having read almost all his fiction output (I aborted my first try at Zero History) and a lot of his non-fic, 'stunt' is not the word I'd use these days. he's one of those writers who I think has suffered greatly by being over-rated and under-appreciated.
posted by lodurr at 10:09 AM on July 12, 2012 [1 favorite]


He should have just put it on a SyQuest disc. Default self-destruct.

I go Jaz because I have very large files I never want to be readable again.
posted by bongo_x at 10:45 AM on July 12, 2012 [11 favorites]


WORM: "write once, read maybe"
posted by mosk at 10:54 AM on July 12, 2012 [1 favorite]


Hey, I figured it out! "There's plenty more money to make figure this out and prepare to meet july 19 fifty sixth and sixth hot dog stand outside ____ cafe ask for mister ___"

what the


But seriously... Malor is right. Why do people believe that the program "encrypted itself", as opposed to just trashing itself? This should not be hard to figure out via analysis of the original binary.
posted by qxntpqbbbqxl at 11:01 AM on July 12, 2012 [2 favorites]


It would certainly have been easier to just overwrite it with random noise, but this is a poem, a work of art. I do not find it at all hard to believe that an actual encryption scheme of some kind was used.

It's a shame that the hardcopy version with the photosensitive inks was apparently never actually made. It would be simultaneously wonderful and terrible to have a book that changed irreversibly every time it was opened. Sure, you could read it under low light, or maybe red light, to prevent it from changing – but then you'd never get to see the other versions of the poem evolve as old words faded and new ones reappeared. But if you allowed the changes to progress, there would be no going back.

Much like life and memory, of course – that was the idea. I wish it had been made, though the floppy-disk poem is pretty excellent as well.
posted by Scientist at 11:11 AM on July 12, 2012


Why do people believe that the program "encrypted itself", as opposed to just trashing itself? This should not be hard to figure out via analysis of the original binary.

Depends on the trashing. I was under the impression that a properly encrypted file is indistinguishable from random noise. If it was overwritten with random noise, how could you tell it wasn't encrypted?
posted by jedicus at 11:13 AM on July 12, 2012


Jedicus, that's true, but in this case they also have the program that performs the encryption. It's the difference between seeing a magician's stage show from the audience, or seeing it from every possible angle and being able to pause and replay it.
posted by qxntpqbbbqxl at 11:21 AM on July 12, 2012


Jedicus, that's true, but in this case they also have the program that performs the encryption. It's the difference between seeing a magician's stage show from the audience, or seeing it from every possible angle and being able to pause and replay it.

That's a fair point.
posted by jedicus at 11:45 AM on July 12, 2012


Not sure I'm following you. Are you saying that access to the source code would allow you to tell whether the alleged cryptext was actually encrypted? My math is way not up to cryptography, but my understanding is that that's in fact the point of how modern secure cryptography works: even with the source code you can't tell. All you've got to do is...lose the key.
posted by lodurr at 11:50 AM on July 12, 2012 [1 favorite]


lodurr, I think you might be conflating two separate things. Modern cryptosystems strive to make sure that if I show you an encrypted block of data, you can't statistically distinguish it from random noise. They also are designed to make sure that even if you know the exact details of the encryption scheme used (the source code, essentially), you won't be able to use that information to recover the plaintext from some encrypted text (assuming no weaknesses in the cryptosystem and that you also don't have access to the key).

But if you can look at the program and see exactly what operations it's performing on a block of text, you'll definitely be able to tell whether what it's doing is encryption or just overwriting with random garbage, even if you might not be able to tell the difference between the two possibilities from the end product. Not that decompiling and stepping through this particular program will necessarily be easy, since it's designed for a 20 year old processor that probably not too many people are enormously familiar with at this point.
posted by figurant at 1:51 PM on July 12, 2012


OK, I thought you were saying that you'd be able to decode the cryptext. That seemed odd, since otherwise you seemed to be making sense to me. Which isn't necessarily saying much on this topic.
posted by lodurr at 1:52 PM on July 12, 2012


If the poem was actually being encrypted with a symmetric cipher (where the same key is used to encrypt and decrypt), then you probably can decode the encrypted information. The program would need to have a copy of the key in it somewhere, or else a way to derive it, in order to perform the encryption in the first place, and that can be recovered in the decompilation. Not that performing this decryption would be too productive, because we already know what the poem says, but it would demonstrate conclusively that the program was encrypting the text.

An asymmetric cipher, like RSA or other, is also a possibility, but I'm betting the program was just designed to overwrite the poem with garbage.
posted by figurant at 2:12 PM on July 12, 2012


I don't really get this. What, exactly, is the challenge? If you have an emulated Apple system, you can easily trace exactly what happens, and from there figure out if it's getting encrypted or just randomly destroyed.

Has no one done this before?
posted by ymgve at 2:52 PM on July 12, 2012


Scientist: It's a shame that the hardcopy version with the photosensitive inks was apparently never actually made. It would be simultaneously wonderful and terrible to have a book that changed irreversibly every time it was opened. Sure, you could read it under low light, or maybe red light, to prevent it from changing – but then you'd never get to see the other versions of the poem evolve as old words faded and new ones reappeared. But if you allowed the changes to progress, there would be no going back.

An Argentine publisher called Eterna Cadencia published an anthology of writing by young Latin American authors where the ink will fade over the course of a few months once the book is removed from its packaging.
posted by Kattullus at 3:52 PM on July 12, 2012 [1 favorite]


It's a shame that the hardcopy version with the photosensitive inks was apparently never actually made.

I really love that idea. It reminds me of another project I read about ages ago, which was a collaboration between an author and a glassblower. Each piece consisted of a unique short story (or poem?) handwritten on paper, which was rolled up and incorporated into a glass sculpture. So the owner has a choice: smash the glass sculpture, or leave the story locked away and unread forever.
posted by metaBugs at 2:37 AM on July 13, 2012 [2 favorites]


He should have just put it on a SyQuest disc. Default self-destruct.

I go Jaz because I have very large files I never want to be readable again.


Last week, I picked up an old Zip100 drive from the thrift store for 2 bucks. Took it home, hooked it up to my PC, and dumped the contents of a half-dozen Zip disks that I'd had in a shoebox in the garage. They were all written between 1998 and 1999, and I didn't get a single error. It was pretty fun. (The archive of my old website is, of course, embarrassing.)

Now I'm wanting to fire up the old C=64 that's been sitting in the garage forever and see how many disks are still readable.
posted by xedrik at 10:01 AM on July 13, 2012


I... wha? If you had asked me 10 minutes ago before I read this thread, I would have said I was 95% certain I remembered someone having cracked the security way back in the 90s.

I vaguely recall someone (genius IQ computer nerd; subcategory "90s Industrial Music"), upon learning that I was a Gibson fan, handing me a cracked disk with a wink and a nod. I was duly impressed, of course. Played it, read the poem, thought "That's interesting enough, I suppose." Then tossing the disk in a desk drawer never to be seen again.

Everyone seems so certain that it has never been cracked. It makes me question the validity of my own memories.

Although thinking back on it, the most likely explanation may be that it was a bootleg copy. Possibly even just a hand-coded version of the poem/program without the self-destruct-y bits. (Plus the requisite dollop of 90s hacker swagger, of course, with the big stompin boots and the oversized jacket bedecked with useless [yet merrily clinking] buckles and straps.)
posted by ErikaB at 11:01 AM on July 13, 2012


« Older Can Tumblr embrace ads without selling out?...  |  Letter from the Pulitzer Ficti... Newer »


This thread has been archived and is closed to new comments