Is there any well-structured catalogue of previous z340 hypotheses, explorations and research already undertaken, so as not to repeat fruitless research?
Or indeed, upon which to continue or build upon?
I’m minded to explore the possibility of null symbols in the cipher, and didn’t want to undertake a load of work only for someone to point me towards alreadylookedfornulls.com
CK
Good starting point:
http://zodiackillerciphers.com/wiki/ind … servations
I have tried to condense all my work in AZdecrypt and its "Substitution + simple transposition" solver could likely solve your grid cipher hypothesis for example.
I have a table of contents for my thread, but it is a work in progress.
Nulls have been explored, but it is a very difficult subject and it is possible that the subject has not been explored enough.
For route transposition, or simple columnar transposition, a null inserted during transcription will cause a misalignment in the plaintext when the message is untransposed. It only takes a few transcription skips or nulls to make a message unsolvable. Skips or nulls around the middle of the message, and skips or nulls arranged vertically in a column will cause the message to be more difficult to solve, while skips or nulls at the top or bottom of the message, or arranged horizontally in a row will cause the message to be not so much more difficult to solve.
Skips or nulls also cause a distortion in the periocity of the message, and can cause a shift of periocity. For example, let’s use a 17 x 20 inscription rectangle, inscribe the plaintext LRTB, then read TBLR and transcribe into another 17 x 20 transcription rectangle. The period 1 bigrams in the plaintext are now period 20. Skip a plaintext during transcription. The period 20 bigrams in the area of the skip will be period 19. The period 40 ( formerly period 2 ) bigrams will be period 39. They can match up to other period 39 bigrams that are not in the distortion area, and cause a spike at that period.
The more skips or nulls, especially closer to the middle of the message, the larger the distortion area, and the lower the period bigram repeats you will get. If you randomly place about 10 or 12 of these in and around the middle rows, you will not get P19 stats. Jarlve wrote a computer program that we tested for about 6-8 months, and we tried to solve the 340 with up to about 12 skips or nulls. It didn’t work. However, more work could be done in this area. It takes many, many hours with a very fast computer to try everything.
We thought that the skips / nulls idea was the most simple and plausible explanation for P19, but for the message still not being solveable.
Also, there has been discussion and experimentation regarding the + symbol as a null, some or all of different combinations of the pivots symbols as nulls, and the regional bias symbols as nulls.
The + symbol was a good idea because it doesn’t cycle with any other symbols. Removing the + symbol and trying to untranspose did not work though.
Not much has been done with removing the pivot symbols, except that removing them or combinations of them generally shift period 39 stats to period 38, which is 19 x 2, but it also reduces P19 stats.
The regional bias symbols, the W, the theta, and a couple of others that exist only in the top and bottom four or five rows, or only in the middle area of the message and also don’t cycle with other symbols. Of the 1,953 combinations of two of the 63 symbols, only a handful of those combinations including the W and theta combination, when removed, result in increasing P19 stats, and most of those involve the regional bias symbols. I can try to find the discussion of that if you want.
I like the idea of the last 8 symbols in the message, the zodiac signature, as nulls, plus maybe the W and theta symbols.
One issue is trying to figure out how to use our time as efficiently as possible and finding people with the same energy at the same time and with the same idea to work together. And then if choosing to focus on one observation or pattern, usually ignores the others. It is hard to figure out what patterns or combinations of patterns to work on. This is really unchartered territory. There is no other message ( if it is a message ) like this one, not even remotely, and it is short.
You should definitely download Jarlve’s program and experiment with it. There is a robust skip / null transposition solver in it, which can solve a message with a handful of skips or nulls in seconds or maybe minutes.
I can see how quickly this can become disheartening.
(Before reading your post above) I ran some quick null experiments on my PC.
On the hypothesis that 9 of the 63 symbols were nulls – it seemed a simple method for the Zodiac to try and break up bigraphs by hand when enciphering on paper, possibly as a second pass having counted bigrams from a first encipherment.
There’s over 23 billion combinations of 9 from 63, which was too high for a quick experiment, so I brute forced all 9 from the top frequency 21 symbols, seeing which improved bigram counts the most. I then removed poor candidates from the 21 and replaced them with other symbols, repeating until I had the best set of 9. Not necessarily optimal but possibly close.
Alas, I only got 37 bigrams, not many longer patterns, and no hopeful solutions pumping through zkdecrypt.
The other thing I’ve tried is taking the skip-19 rearrangement directly (due to the high bigram count), and then repeating the 9-null optimisation. It’s a bit of a stretch to imagine someone doing all this work on paper (particularly without making a critical mistake), but bigrams of course go up.
Highest was 47 bigrams, with a message 294 symbols long.
And decrypt is nonsense with consistently lower scores than a truncated 408 solution.
What is interesting is that the + symbol contributes far more to creating bigrams than disrupting them. None of the optimal null patterns included removing +, and when I forced his manually it didn’t work well.
Why 9?
63 symbols vs the 408’s 54
The assumption being that ALL the extra 9 symbols over the 408’s 54 were nulls. I figured that if there were a few more or fewer it might not prevent the identification of potential solutions.
Obviously that is an assumption. I also thought some of the extra could be used to flatten the frequencies, leaving fewer for nulls. Or it could be a lot more nulls than 9.
At worst I suppose it could be a simple 26 symbol non-homophonic ("monophonic" ?) cipher with 37 null symbols added. That’s a lot of noise to filter out.
Trying 37 from 63 is also in the order of 3×10^18, which is heading towards intractable. Plus the resulting short ciphertexts with so many symbols removed would have a huge degree of freedom to assign nonsense plaintext solutions to.
So perhaps I won’t try that.
On the hypothesis that 9 of the 63 symbols were nulls – it seemed a simple method for the Zodiac to try and break up bigraphs by hand when enciphering on paper, possibly as a second pass having counted bigrams from a first encipherment.
I have tested this hypothesis also (at least up to 20 null symbols) and AZdecrypt 1.15 comes with a hill climber that can figure this out (substitution + units) as well as null rows or columns.
What is interesting is that the + symbol contributes far more to creating bigrams than disrupting them. None of the optimal null patterns included removing +, and when I forced his manually it didn’t work well.
I make homophonic cryptograms, and one day experimented by making 62 symbol messages, and then randomly inserted 24 of the same symbol ( to represent the + ) into the cryptogram. The + symbol did not create very many bigram repeats at any period, and of course did not cycle well with the other symbols. However, in the 340, the + symbol does exist in several of the period 19 bigram repeats.
What sort of research has been done already regarding cipher symbols mapping to pairs of letters such as digraphs, like one symbol meaning "ee"? I’m not sure where this would end, since you could also have symbols for common 3-letters, or even whole words and phrases.
And the flip side of that, of groups of symbols coding for single plaintext letters?
What sort of research has been done already regarding cipher symbols mapping to pairs of letters such as digraphs, like one symbol meaning "ee"? I’m not sure where this would end, since you could also have symbols for common 3-letters, or even whole words and phrases.
Nomenclator cipher.
daikon looked at bigram nomenclators: viewtopic.php?f=81&t=2782&hilit=nomenclator
There is no proper homophonic nomenclator solver out there (that can also substitute symbols for words etc). So the work here remains limited. Though all nomenclator ciphers that have been posted here have been solved since they usually also contain many regular 1 to 1 assignments.
And the flip side of that, of groups of symbols coding for single plaintext letters?
Verbose cipher.
Can probably be ruled out imo since it creates many bigrams.
If only 25% or less (give or take) of the letter to symbol assignments are nomenclator/verbose then the cipher will solve anyway without special hill climber (340 multiplicity).
—
I would personally not take anything for granted and create test ciphers anyway. Tell us what you did and we try to solve it and look at the stats.
Doranchak posted up 20000 scytales on an excel spreadsheet. I have it on my computers i dont know if it is available anywhere else on this site. It has some interesting results but that could be by default of so many..its just a numbers game. Lots of things like this T + T + U + U + kept my small mind amused for a while after azdecrypt popped out with HAHAHAHA..
On the hypothesis that 9 of the 63 symbols were nulls – it seemed a simple method for the Zodiac to try and break up bigraphs by hand when enciphering on paper, possibly as a second pass having counted bigrams from a first encipherment.
I have tested this hypothesis also (at least up to 20 null symbols) and AZdecrypt 1.15 comes with a hill climber that can figure this out (substitution + units) as well as null rows or columns.
Quick example:
221 character plain text encoded to 43 symbols, then added 20 null symbols:
W+<<(]7[4]'JG].;L KAB/ID#<T%5OIY]2> ,VT(_]4Q$:(-ZSG9X *_V)<-T@U1N3T^HH/ W(T+61KA:T%L^D-5 (DJ[PRT92",Z=+_$3 :WTKO!]S:3YT];>&U R21V'0L96B%7.CAG- TTP)(OQ,=Y7N$BGUZ D57_V'"%+D6C*_-) 1TY(3K*,T0V=B7_& T8$7W59TP5IEUTDT' G^+6*%EU5#:"YI/)% 0FVDDP/]0>LM-52J XJ7(,T"=_$R5Z4C3$ [AY'0:25KWU"#Z%C7 OV295D4>6ST]0L+1T B(#7<,A7-.RG?8]TN ^THH8!W/MKP)*59= _+>4O!.Y'"5T6JPDT
Solved in under a minute with standard 5-grams so more null symbols could be attempted:
Score: 20095.48 IOC: 0.06481 Multiplicity: 0.19815 Repeats: 511 PC-cycles: 4442 Seconds: 59.07 Unit: symbol Mode: remove Key((,],7,[,.,/,I,D,T,5,O,>,S,R,!,&,C,8,F,M) COMMUNITYCOSTUMEA REAHUGEPARTOFWHAT MAKESBLISSCONSOSP ECIALINORDERTOHEL PCOPLAYERSANDCONT ESTANTGETABETTERH ANDLEONWHATSALLOW EDATTHECONVENTION WEVEUPDATEDANDCLA RIFIEDTHERULESAND PROCEDUREAROUNDCO STUMESATABLISSCON TWOTHOUANDNIN
Used the Substitution + units solver with the following settings: unit: symbol, mode: remove, multiplicity weight 1, key length start and stop set to 20 and everything else unchanged.
