A New 340 Solution

Introduction Method Solution Resources

The Documentary

The recent documentary on The Hunt for the Zodiac Killer, which was shown on the History Network, teased that Kevin C. Knight, at the USC Natural Language Group, would be turning his profound knowledge and abilities toward solving the Zodiac Killer's 340 cipher. I was, needless to say, impressed; everyone knows about Knight's work on the Copiale Cipher.

I wasn't expecting very much from the History Channel's documentary; their reliance on specials about the SS and Hitler's Germany convinced me years ago that they were devoted to entertainment, not education. Their portrayal of CARMEL, the supercomputer which was going to tear into the 340, was silly, with blue lights behind the racks, scrolling Matrix-style text in which a dimly-outlined human face showed, and the like. The computer itself was of some interest, and so was the approach this very intelligent team was taking. Were they using a Beowulf cluster? A two-rack BlueGene/Q? How many pflops? Inquiring minds wanted to know -- and instead we got hysterically bad poetry. And some of the people you'd want on a 'dream-team,' particularly John C. King, Thang Dao , Pallavi Kanagalakatte Basavaraju , Dan Umanovskis , Taylor Berg-Kirkpatrick and Dan Klein , were nowhere to be found. Instead, we have people who an average audience will take to be geniuses -- young folk from Google wearing horn-rimmed glasses and the like. That, plus a poetry-writing 'supercomputer' of unknown provenance with dramatic lighting gives a clear picture of what is to happen at the end. The 'supercomputer' cranks ominously, and does nothing, and then Craig Bauer rushes in dramatically with another guesswork solution.

The Problem

Thang Dao's master's thesis clearly outlines the problem with any sort of computational approach, and Pallavi Kanagalakatte Basavaraju's work highlights the problem.

To explain -- Dan Umanovskis' zkdecrypto can solve the 408 cipher in a matter of minutes. In fact, now that one can download zkdecrypto on virtually any platform, anybody, anywhere, can solve the 408. But the 340 remains insoluble, and people have used some very clever methods (as shown in the San Jose theses listed above) in their attempts. This tells us something important, I think: the problem is not in our methods; it is in the 340 cipher. Obviously, Kevin Knight is brilliant, and had some sort of very powerful computational technology at hand. He solved the Copiale, and so we know he is very good at this. What seems to be needed is a new approach, one that puts more emphasis on the pressures facing the awful person behind the cipher.

What Faced the Killer

The Killer (I won't give him the honor of calling him by his preferred names, "Zodiac," or, as I hope to show, "Beltane Ally,"), had a great deal going on between August and November of 1969. A timeline will show what I mean.

  1. July 31, 1969, 408 mailed
  2. August 7, 1969, "This is the Zodiac" letter; when they crack my code "they will have me."
  3. August 8, 1969, Donald and Bettye Harden crack the 408.
  4. September 27, 1969, Lake Berryessa attack
  5. October 11, 1969, Presidio Heights attack
  6. October 14, 1969, Letter with Stine shirt-tail fragment
  7. November 8, 1969, the 340 is received.

It seems very unlikely, upon some reflection, to suggest that the 340 and 408 codes are fundamentally different. How would The Killer have, in the space of 99 days, (Thursday, July 31, 1969 to Friday, November 7, 1969) come up with some entirely new cipher? I think, instead that he realized two things:

  1. Lines of garbage interfered with the decryption process. Note that the proposed decryption of the 408 includes a line of junk, "EBEORIETEMETHHPITI".
  2. Using words like "KILL" provide decryptors with cribs.

What The Killer did

I imagine, then, that The Killer (I'll use 'him, he' to describe The Killer) would use the same techniques of encryption that the Hardens uncovered. He would, however, use a far more complex vocabulary, more abbreviations, more misspellings, and more junk lines.

And no crib words. That, after all, is what anyone would do in the same situation. He saw that a crib made it possible to decrypt the 408; thus, he would avoid cribs at all cost.

What a solution should do

And, as Ryan Garlick argues, there are several ways not to devise a solution to 340. Just as surely, there are several things a solution will require, as James Hodgkins pointed out in his addition to the knowledge brought forward by the Copiale Cipher; one must be familiar with the topic dealt with in the plaintext if the interpretation is to be sensible.

Further, as Garlick argues, the solution must follow rules, and be replicable by others from those rules. I should be able to tell you how I arrived at my solution, and without knowing the solution, you should be able to recreate the solution.

How to Proceed

What are the things we know - or, at least, I contend we know? And what do they say about how we should proceed?

  1. The 340 cipher is a homophonic cipher like the 408; a substitution cipher with ''variants.'' In other words, low frequency letters (e.g. q, j, etc.) map to only one cipher character, while high frequency letters (a, e, n, etc.) can be represented by several cipher characters. The 408 used seven symbols for 'E', four for 'T', but only two for a rarer letter like 'F'.
  2. Further, we know from Dan Olson of the FBI that

    Lines 1-3 and 11-13 contain a distinct higher level of randomness than lines 4-6 and 14-16" of the 340. This appears to be intentional and indicates that lines 1-3 and 11-13 contain valid ciphertext whereas lines 4-6 and 14-16 may be fake.

  3. Thus an analysis of the encrypted text should discard lines 4-6 and 14-16, plus the signature at the very end (Zodiac symbol plus ZODIAK, which is clearly not encrypted.)
  4. We also know that The Killer was fond of abbreviations, deliberate misspellings, and the like as a way of increasing the difficulty of decryption. Thus we can expect to see these, plus complex and unexpected words; and
  5. We can expect that the sorts of words that made the 408 document readily decrypted ("kill") to be absent.
  6. And we know that zkdecrypto-lite works on the 408, and so - if these ideas are correct, it should enable us to make progress on the 340 once the 'garbage lines' are deleted along with the signature.

Introduction

The reasoning behind this solution

Method

Exactly how this solution was derived

Solution

The solution itself, with explanatory material

Resources

Resources so you can

  1. recreate this solution

  2. improve on this solution

Who I am

Absolutely the least interesting and important thing here