Hi. Boy I’m awaiting a firestorm for this but hear me out…
(And please read it carefully before responding)
Curious if this approach has been tried. (I’m sure parts of it have been)
Take the z corpus and determine useful stats – letter frequency (single, bi and trigrams etc.), word frequency, common mistakes, etc. etc.
Take z340 and create some homophonic substitution codes that deliver 340 plaintexts that mimic the above stats
I think the key is to do this to the code as a series of random letter "jumbles" – randomly reordering the 340 symbols.
However – the "jumbles" should have letter positions individually "tagged" so that they can be reassembled into the original 340.
Now, scan the random jumbles that reflect the z corpus stats for useful strings -and/or- and here is where I ask you not to shoot me! – try "un-anagramming" the random jumbles back into common z words and stats etc. – preferably long words and many of them!
Finally, from the most promising of these candidates, and using the "invisibly tagged" locations that can be mapped back to z340 normal reading direction – map them back into place and here’s the real goal of this exercise: look for any possible transposition patterns. Maybe this could provide clues about the "space" of the 340.
The idea for this is to see if there are other ways to skin this cat. Look at it from another facet. Because there seem to be a couple of layers going on.
Personally I think if a little guess work (like un-anagramming from a random jumble) can lead to a solid, workable clue, then it will have been a useful, and well applied tool.
And since I’m starting this thread – I’m curious what other people think would be a useful "order of operations" for decoding
Hey Fisherman’sFriend,
Not sure what you mean.
Do you mean a random transposition pattern with a fixed length that is repeated over and over again? That would qualify as columnar rearrangement.
Such as:
1. Code 2. Plaintext 3. Plaintext after encode 1. 421534215342153 2. ABCDEABCDEABCDE 3. DBAECDBAECDBAEC
There must be some system or pattern because there are 340! possible anagrams of the 340. That’s 715 decimal digits. Just to illustrate that we need to be careful and set limits to our hypotheses.
340! = 5100864472103711080930193283927293363034467198288537603946858243798113070849885229213136372954649899020351011761059778806133788936493577434500361683415188982472401346338884861141422740713138880590833811293694271365431210201323723370857438895402320411929506183969598863382952818389369235832500911641281417808599643869688726988699026312600541139182490611593239476045067552375315738712391517997520501623882216426473863468587538818060875218240504105205706947492365188106425431862641010474387420418554164289649410408629854298787786204180850445367165509339805260836237886088058511128061771315622325534263276929061018352747735327201401241600000000000000000000000000000000000000000000000000000000000000000000000000000000000
N gram files, made from everything that Zodiac wrote, like what beijinghouse did. But with the text transposed before the n gram files are made. Apply and see if one type or period of transposition fits better than others. If one transposition shows best results, then apply the scheme and try to solve. I think that is the idea.
EDIT: Forgive if I am wrong.
N gram files, made from everything that Zodiac wrote, like what beijinghouse did. But with the text transposed before the n gram files are made. Apply and see if one type or period of transposition fits better than others. If one transposition shows best results, then apply the scheme and try to solve. I think that is the idea.
EDIT: Forgive if I am wrong.
Hi Smokie,
I think that sounds very similar to what I’m thinking.
Let me try to re-state it, maybe I can clarify it (as much for myself as well as for anyone interested)
(-Also keep in mind I can barely use my phone and am interested in z340 from a pen and paper standpoint – but obviously I have come to see the limitations!)
1. Establish N-gram stats based on all accepted Z writings
2. Random shuffle z340, but do so in a way that it can mapped back into the original and preserving the order of the appearances of each symbol)
3. Test for keys that score well in relation to step 1 WITHIN the randomly ordered 340 from step 2) – In some sense (because it’s a jumble) I think it’s worth testing more for letter frequency than bigrams / trigrams etc.
4. Here’s the controversial part – basically anagram the entire thing scanning for long important z words.
5. Take high scoring candidates from step 4 (For example you find paradice, slaves, christmass, etc. etc.)
6. Now map this back into the original z340 and look to see if any of these randomly unscrambled words point to patterns / layouts / etc.
7. From this point, just work on interesting candidates as per usual.
I think I’m wanting to advocate for making some unscientific assumptions, but then testing them scientifically.
I personally don’t think that would work. Jarlve pointed out how many different possible ways there would be to anagram the 340. It is just too many for a computer.
Last year we spent about 6 months working on a project that would untranspose the message at periods 15 and 19, the most likely periods if the message is in fact transposed, which we don’t know for sure. And we allowed for up to 10 nulls or transcription mistake skips. No results. The scope of that project was infinitesimal compared with what you have in mind, and it took six months.
There are some confounding patterns with the arrangements of the symbols with respect to each other, which may be clues about how he may have rearranged the message before encoding. If it is a message at all. Otherwise, a very interesting mathematical system of placing the symbols into the 17 x 20 grid. Those clues are probably the best way to approach the problem, I think. I think about it every day.
That is all I can offer. The computer programmers may be able to give better discussion. Thanks.
