The Unabomber's Code

This is interesting. Ted Kaczynski wrote in code:

In a small journal written in code, he documented his thoughts about the crimes he was committing. That code was so difficult, a source says the CIA couldn't crack it -- until someone found the key itself among other documents, and then translated it.

Look at the photo. It was a manual, pencil-and-paper cipher. Does anyone know the details of the algorithm?

Posted on December 6, 2006 at 12:53 PM • 30 Comments

Comments

Ryan TomaykoDecember 6, 2006 1:39 PM

IANA cryptographer but as a programmer my first impression is that the whole thing is fabricated. Assuming the commas are insignificant--which they seem to be from the two scanned pages--I would have dropped them after about the 10th number.

On the other hand, what possible motivation would someone have for faking a code-book?

The whole thing smells a little funny.

Kevin DavidsonDecember 6, 2006 2:31 PM

It might depend on just how much time the CIA [NSA?] spent on breaking it.

What do they do these days, scan the numbers into an omnibus computer decrypter, or do humans still do this stuff?

Mike La SpinaDecember 6, 2006 2:55 PM

It looks like part of a modified DES cipher algorthim to me. DES uses a 14 x 4 matrix on its sbox configuration. He underlined the 4th line

Jon A. SolworthDecember 6, 2006 3:02 PM

I don't think its so surprising. Ted Kaczynski did after all do a PhD in math at U of Michigan and was hired as an Assistant Professor at Berkeley. One might also say that his mind worked in unusual ways---thus making it difficult to get a grip on the cipher.

Alice McGregorDecember 6, 2006 4:43 PM

I've seen many codes like that one, and even invented a few derivitves of my own. The following text, index-encoded, would look very similar to the two pages linked to:

03 01 13 07 21 15 08 28 09 00 09 15 22 26 03 00 09 15 17 20 08 18 05 22 09 03 24 25 20 06 16 22 18 14 15 16 09 24 14 27 28 23 21 08 10 01 01 ...

... it's a bit lengthy (it's a lot less lengthy and secure if you don't index encode it), but still usable as a pen-and-paper cipher with a good enough brain, or three alphabet discs: two rotating and one scrambled and replaceable. It even has nearly even letter distribution in the encoded text, so standard rotational cipher attacks don't work, it acts like a one-time-pad if the keyphrase is long enough, and auto-keying improves the evenness of key distribution.

FYI the above decodes to "I HAVE BEEN ROLAND, BEOWULF, ACHILLES, GILGAMESH."...

Israel TorresDecember 6, 2006 5:35 PM

Reminds of the codes by "CodeMaster" that are published in the back of "Boys' Life" magazine... particularly the one entitled Top and Bottom code. Especially neat because of how it appears to change each run...

>cm_TopAndBottomCode.exe
codemaster Top-and-Bottom Code v2.0
SPEC: BOYS' LIFE - MAY 2005 - PAGE 50 - CODEMASTER - Top-and-Bottom Code
http://tools.israeltorres.org
usage: cm_TopAndBottomCode.exe [/encode | /decode] numbers-to-decode
usage: cm_TopAndBottomCode.exe /decode 14-26-25-23-37-22-18-23-23-23-16-17

>cm_TopAndBottomCode.exe /encode I HAVE BEEN ROLAND, BEOWULF, ACHILLES, GILGAMESH
75 19 00 73-12-55-14 62 19 41 23 00 53-24 49-23 85-23 44 14 00 68-23-26 24-43-21
-89-14 48 14 83 15-00-35 24 29 23-92 24-76-18-90-12-27-21 49-26-00 31 14 89-22 9
7 12-95 19-91 21 59-21-71 23 72-26 00 73-17-27-19 55-21-77-17-37 14 63-16 73-23
24 26-61 12 00

>cm_TopAndBottomCode.exe /encode I HAVE BEEN ROLAND, BEOWULF, ACHILLES, GILGAMESH
81-19 00 23-12 97 14 30 19-69 23 00 55-24-87-23-33 23-70 14-00-86 23 74 24-19 21
39 14 88-14-75-15-00 55 24 75-23 50 24-72 18 64-12 67-21-63 26-00-93-14 95-22-6
9 12-23-19-31 21 71-21 95-23 60 26 00-85 17-57-19-75-21 75 17 11 14 75 16-57 23
26-26-49 12 00

>cm_TopAndBottomCode.exe /encode I HAVE BEEN ROLAND, BEOWULF, ACHILLES, GILGAMESH
87-19-00-85 12 51 14 86 19 73 23 00 33 24-63 23 71-23 30 14 00 16-23-56 24 83 21
-53 14-40 14-91-15-00 75 24 35-23-94-24-66 18-12-12 83-21 51 26-00-93 14-13-22 3
9 12-39-19 57-21-21 21 97-23 24-26 00 11-17-21-19 71 21 49-17-97-14 25-16-39 23-
30 26-37-12 00

This also can be done easily with pencil and paper as long as you know how it needs to be done.

Israel Torres

Davi OttenheimerDecember 6, 2006 5:37 PM

Can't believe he used spiral-bound paper and pencils. I thought he was against technological progress and modern technology...

Wonder if the key is in The Secret Agent.

Davi OttenheimerDecember 6, 2006 5:43 PM

"He wrote about everything. He wrote about what he had for lunch on May 5, 1979, where he got the food, how he prepared it and what did it taste like..."

Like a blog but encoded. That would be a challenge. Bruce, howabout a few entries in ciphertext? First one to post the answer in the comments section gets a signed copy of your restaurant guide. ;)

lorenzoDecember 7, 2006 2:33 AM

interesting quote:

"And there are notes of his attempts to outwit law enforcement, as in one journal entry where he says that he intentionally put two human hairs acquired in a bus station into a bomb "to deceive the policemen, who will think that the hair belongs to whoever made the device."

I know that DNA samples aren't everything, but for the media definitively are.. so to do a "mediatic kill" of someone this would be more than enough.

Another thought is that biometric security based on dna/retina/whatever aren't that reliable if they're based on something we can leave behind (saliva, hair, skin, pictures of retina, fingerprints) everywhere we go.

Another quote:
"Theodore John "Ted" Kaczynski (born May 22, 1942), aka The Unabomber, is a convicted terrorist best known for his campaign of mail bombings."
...convicted terrorist? This could be the first time I see the definition of "terrorist" applied correctly in a while. He *was* a terrorist. And did have a trial. Those who end up in guantanamo today aren't terrorist until (and if) they will be judged. I think that if they took the unabomber and put him in jail based on the definition of "enemy combatant" and suspect of terrorism, people would have been upset.

Oh well.

Clive RobinsonDecember 7, 2006 3:19 AM

@lorenzo

"Another thought is that biometric security based on dna/retina/whatever aren't that reliable if they're based on something we can leave behind (saliva, hair, skin, pictures of retina, fingerprints) everywhere we go."

And CCTV grabs the rest of the current stuff, such as gait / facial structure / Height / hand structure etc.

And now with backscatter techniques (X-Ray / IR / Microwave / acoustic) the last little surface bits get picked up. And possibly some other info like medical abnormalities etc...

Soon there will be RFIDs in your underwear and other clothing to tag and identify you as well.

So just colate it all together and,

"Welcome to the goldfish bowl"

OddscurityDecember 7, 2006 6:40 AM

It also looks a bit like the book ciphers used by secret services, way back when. Without knowing what book was used, and what index page, it would be tough to decode.

BobDecember 7, 2006 8:59 AM

Since Ted did his thesis/research in wave harmonics/acoustics, I would guess that he stuck with what he knew and used a fourier series. The key they found was probably what he used to start hte series.

Israel TorresDecember 7, 2006 10:02 AM

@Ilya Levin:

ohoe
oeyn
kbtj

from furthest top right down to next top left down (repeat) will produce:
"enjoy the book"

Which we certainly do.

Israel Torres

RFIDDecember 7, 2006 10:51 AM

I don't think its so surprising. Ted Kaczynski did after all do a PhD in math at U of Michigan and was hired as an Assistant Professor at Berkeley. One might also say that his mind worked in unusual ways---thus making it difficult to get a grip on the cipher.

Clive RobinsonDecember 7, 2006 12:25 PM

The cipher does not have to be that complicated to remain un-solved.

Providing you follow a few basic rules, you might have a modicum of success. For the simple reason that "to break an unknown cipher takes resources". So you get into an ROI situation.

Clasicaly to break an unknown cipher you first need to be able to spot statistics in the cipher text to give you clues to the system type.

You then need to spot a reliable change in the output statistics or find recognisable known plain text to confirm your trial guesses.

To do this in a reasonable time or with reasonable resources you usually need to have either,

1) a depth under the same key, or

2) known plaintext to work backwards from.

Failing the above a lot of time and resources and do the statistics game, in which case you need a real good reason to justify the costs.

In the past one (fairly) reliable method was to split the process into two or more parts that,

1) Flatten the frequency distrubution of the plain text.

2) Encode with your chosen key using both transposition and substitution.

You could also optionaly,

A) add nulls to plain text
B) Insert "flattening data" in cipher text
c) Split message in two at some random seeming point and nose to tail the two parts.

Most of the above have been done at one time or another with quite good success (at the time ;)

One "agent" system actually did the reverse and used a One Time Pad to encode followed by a second stage to make the frequency distibution look like a transposition cipher. Apparently this was for two reasons, the first so that the traffic did not stand out and the second to waste the time of "enemy cryptographers".

Israel TorresDecember 29, 2006 11:10 PM

FTW.
"Agents discovered the first of many clues to solving the puzzle in one of Kaczynski's notebooks, on a page labeled, "Unscrambling Sequence."

On the handwritten page, arrows appear to show the direction that a page packed with numbers should be read. Then, in a complex series of decoding steps that Kaczynski calls "phases," numbers are added, subtracted, and married, revealing a new set of numbers that translate to letters and word fragments."
-- http://cbs5.com/topstories/...

Israel Torres

AldarionJanuary 8, 2007 2:27 PM

How does that "top and bottom code" thing work?
I tried contacting Israel Torres, but I didn't find his email anywhere.

resJanuary 23, 2007 11:56 AM

Aldarion, it's just a "kid's code", of course:
The alphabet is written in two rows with A-M on top and N-Z on bottom. A letter is encoded as xx yy, where xx is even or odd to indicate the row (odd=top, even=bottom), and yy is one of 13 distinct numbers to indicate the column (14,24,22,15,23,26,17,12,19,18,28,21,16 from left to right). 00 indicates an end-of-word, and dashes are ignored.) E.g., the word 'I' is encoded as xx 19 00, where xx is any odd number.

Sherry PearsonJune 16, 2007 2:58 PM

I believe the unabomber is also the Zodiac. Take a look at his codes and the police composites. They look so much alike. They are the same person.

JJ GulkaDecember 29, 2007 3:50 PM

Since he did his thesis/research in wave harmonics/acoustics, I would guess that he stuck with what he knew and used a fourier series. The key they found was probably what he used to start hte series.

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc..