Me and smokie explored the hypothesis that the 340 plaintext was originally written in 15 or 19 columns (by columns) and then taken off by rows (unkeyed columnar transposition) into a 17 by 20 grid (LRTB, RLTB, LRBT and RLBT) while occasionally skipping and/or adding a null letter up to a total of 10 nulls and/or skips in any combination. That would explain the bigram peak at period 19 whilst being being unable to solve the cipher with a regular period 19 untransposition since the nulls and/or skips act as a key.
We first made and optimized a solver within AZdecrypt that would be up to the task. It is basically a hill climber that works on top of the AZdecrypt substitution solver that figures out where the nulls and/or skips are. We made a bunch of test ciphers that were all solved successfully though the 340 remains unsolved throughout the test and the conclusion is that the hypothesis does not fit the 340. Period 15 returned better results than period 19. In total at least 6 months of CPU time was spent between me and smokie and the test was very thorough.
– Is it possible that the 340 is a different kind of period 15/19 transposition such as diagonal transposition which the test did not cover?
– Is it likely that there are more than 10 nulls and/or skips given the increased amount of bigram diffusion?
The test and its output can be found here, under the 1st and 2nd trials tab on the bottom: https://docs.google.com/spreadsheets/d/ … =509220377
Example output of the solver:
Score: 20183.77 Ioc: 0.06946 Multiplicity: 0.20520 Ngrams: 482 PC-cycles: 626 Period(15) Nulls(234,270) Skips(14,36,55,127,133,244,265,295) VBOSANDINTURNNESO AUSNOTWILLMOOLENC EISTHERVALIDEINRE DBLANDISANAUTHATS EVEROUNDSHALFONSO NEINSESTHATWASESS ENTSEEBEDANIMITAN DITSOMEATTHERSTAN DISAWTHISOUCHOUTA SMASWEHISMINTOFTH EDAIDONLYIMBOSAND HEROVENTIBATTHINT HEFASTITHISTHINSO CHISNTONDRSYOUORI FHEERCUSTRYTCELLE ONGRESTRAWNRANDLE SSABOUTINTHUSURUI WENEWWITHOURSTOFC REALYIDBEFOREITHE RSALLONYOUTSIDONA LLYNEH
Example case of the hypothesis:
Original plaintext:
ilikekillingpeopl ebecauseitissomuc hfunitismorefunth ankillingwildgame intheforrestbecau semanisthemostdan gerousanimalofall tokillsomethinggi vesmethemostthril lingexperenceitis evenbetterthanget tingyourrocksoffw ithagirlthebestpa rtofitisthatwheni dieiwillbereborni nparadiceandallth eihavekilledwillb ecomemyslavesiwil lnotgiveyoumyname becauseyouwilltry
Into 19 columns (by columns):
I buieiaometkehrtblb L enlfsnmonhssenhene I cilotiescaotniecoc K atirhmttenfpiniota E uinreahtigfadphmgu K ssgemlihtewriaaeis I emwsoonrititervmve L ioitsfgisttoiaeyey L trlbtagleihfwdksyo I iededlilvnaiiiilou N sfgcalvieggtlclauw G suaantennyiilelvmi P onmugosgborsbaeeyl E mtesekmeeultendsnl O uhieriextrthrdwiat P canmoltptrhaeaiwmr L hntaulheeoetblliey E fkhnsserrcbwolll
Add 10 nulls (question marks):
ibuieiaometkehrtblb lenlfsn?monhssenhen eicilot?iescaotniec ockatirhmttenfpinio taeuinreahtigfadphm gukss?gemlihtewriaa eisiemwsoonr??itite rvmvelioitsfgisttoi aeyeyltrlbtagleihfw dksyoiiededlilvn?ai iiilounsfgca?lviegg tlclauwgsuaantennyi ilelvmiponmugosgbor sbaeeylemtesekmeeu? ltendsnlouh?ieriext rthrdwiatpcanmoltpt rha?eaiwmrlhntaulhe eoetbllieyefkhnsser rcbwolll
Untranpose period 19:
ileotgeraditislrrer beicauivekillbtthoc unckeksmysiceaehaeb iliausiveylllenr?tw efltiseeyooaveddebo isoin?mlliuumyswall antrrgwitinwilniill o??heesoresgpelawil mmimamoildfsomotmee oethlotbeguntuprytn sttinstdcamehclekhc eihrfalaaus?ahfesan gt?ggi?ngeinnkhsoff e?illltokemthretpaw isevvesmroantnnidrt tiningeilusbhinpiit h?enbeetlsleeihatof agyouxphebncomaeiwi gir?tter
And as you can see this process makes gibberish of most of the cipher unless the nulls are correctly removed before untransposing the period.
I would add that we discussed P19 for two and a half years before trying this option. Diagnosis of a single message so short is extremely difficult, but we agree that P19 could be evidence of some type of transposition. Crytography books, especially military cryptography books of the time, show many examples of route transposition, or simple columnar transposition ( no key ). Scytale is also an option, and also a classical cipher that could create P19.
We took our time to decide what to do because it was such a big investment in time. We discussed this and tried different other transpositions, but without more than a few nulls, including incomplete inscription rectangles, multiple inscription rectangles, some diagonal schemes requiring each untransposed row to be solve independently, etc. We also used a myriad of different creative ways to try to detect the cipher type, and attempted to create P19 with a lot of different varieties of schemes. A lot of smaller inscription shapes, 38 wide x two row rail fence and bifid made it very difficult to duplicate P19 stats, so if this is a transposition, there must be one or more large inscription rectangles.
I think that we agreed that several nulls or skips would cause misalignments in any untransposed message, making it unsolvable, and that one inscription shape with a few nulls or skips was a plausible and simple explanation that needed to be eliminated. A few months before we decided to take the plunge, the group here discussed the regional bias symbols and what they could be. One of the possibilities was nulls because removing some of the regional bias symbols actually increases P19.
The dialogue starts at the 9th post here, but final resolution to try the experiment starts on the next page or so. There are about 40 pages of off and on dialogue showing how the program was developed and tested.
viewtopic.php?f=81&t=3196&start=790
EDIT: See 8th post here.
How about a periodic transposition + null solver for AZdecrypt. The user would be able to determine the period and the amount of null characters to add or remove.
viewtopic.php?f=81&t=3196&start=800
A couple of weeks ago I tried to make messages with nulls in the same locations as the regional bias symbols W and theta, which are 11 positions. It was very difficult, but not impossible to produce P19 stats. I feel comfortable saying that with all of the testing, work and time, the cipher is very unlikely to be a route transposition cipher with one perfect inscription rectangle, up to 10 nulls or skips or any division or combination of nulls and skips totaling up to 10, and then homophonic substitution.
One thing that we didn’t do, or at least i didn’t do, was experiment with incomplete inscription rectangles ( last row not full width of rectangle ) and see if the program would fill in wildcards in those empty positions to get a solution. I am also not clear how the + symbol, if it is a polyphone, could hinder the solution of a complete inscription rectangle transposition with nulls and skips. None of our test messages had any polyphones, but the homophonic substitution solver does solve messages with one polyphone taking up to about 30 positions good enough to determine that there really is a message. Jarlve may want to clarify.
Strangely I am not devastated that this didn’t work. I think that we eliminated a plausible, simple explanation for P19 and Jarlve developed a way to detect nulls and skip locations with a homophonic substitution solver, something that may have never been done before, ever. Amazing job as usual!!!
I will try to think if I need to add anything else, but the dialogue helps a lot to show the thoroughness of the work.
Nice summary smokie. Thanks for working with me on this one, you came up with the idea of shifting the positions of the nulls & skips slightly instead of picking a new random position which helped the solver allot + all the bugs you found. It is a very good solver which also works for period 1 hypotheses with the added benefit that it can handle allot more nulls & skips then. Adding support for other transpositions would also be relatively simple.
The test/hypothesis did not cover polyphones. And there would have to be a significant amount as it can be seen that it is not really a problem with solving the 408.
I think that we eliminated a plausible, simple explanation for P19
Exactly!
Thank you both very much for this really remarkable work. Being able to exclude this theory is a really important aspect. I have followed the threads on this subject, but have not gone into it in depth. In any case, what you have done is extremely impressive!
First, let me start off with the usual disclaimer that, "I’m not sure if this has already been discussed." And so…
I personally think the P19 observation, if not yet another coincidence, could be the most significant contribution in possibly solving the cipher. I’ve followed the progress of others closely over the last couple of years, but I’m not satisfied that a simple approach has been offered that would result in a cipher with the P19 tendencies exhibited by the 340. But I have one.
For those with limited cipher knowledge (like myself), a Route Cipher is a transposition cipher where the key is which route to follow when reading the ciphertext from the block created with the plaintext. The plaintext is written in a grid, and then read off following the route chosen. Cipher 101 books usually start off with an example with a top to bottom, left to right columnar route. This route will create a period equal to the number of rows. So, my suggestion is this…
Perhaps the message was only 19 rows long. Z uses a simple route cipher, which creates a period 19. Maybe Z is aware that this is a weakness of a route cipher constructed in this fashion, so he adds a "signature" line (Incidentally, the line looks pretty hokey anyway with the cross hairs right in the middle followed by Z O delta A I K).
Now, before anyone gets excited, attempts at solving this have yielded the usual crap. But I think the additional observation (Jarlve and Largo) that Z may have also possibly shifted columns could play into this as a second tier. That’s all I got. Happy New Year.
First, let me start off with the usual disclaimer that, "I’m not sure if this has already been discussed." And so…
I wish there was a ready-made signature for this disclaimer. Whenever I post a new idea, I ask myself the same question: has anyone already had the idea?
But I don’t think that’s bad at all. Even if an idea has already been posted, it may be forgotten over time. Perhaps there have been new perspectives and insights since the last time someone posted an idea. Ideas are always good…no matter if old or new!
I personally think the P19 observation, if not yet another coincidence, could be the most significant contribution in possibly solving the cipher.
I do also think that P19 and Flip+P15 are indeed one of the most relevant observations. However, with my tools I have also been able to create "normal", i.e. straight forward substituted ciphers with a strong nGram peak on P19. In contrast to z340, however, all these examples can be easily solved by AZDecrypt. Nevertheless, I agree with you: P19 is most likely a very strong indication. You have to keep in mind, however, that all sorts of things can be responsible for such a peak. It is also possible that P19 (unintentionally) leads to a wrong track.
For those with limited cipher knowledge (like myself), a Route Cipher is a transposition cipher where the key is which route to follow when reading the ciphertext from the block created with the plaintext.
In this case, the route is indeed the key. Very well explained!
Perhaps the message was only 19 rows long. Z uses a simple route cipher, which creates a period 19. Maybe Z is aware that this is a weakness of a route cipher constructed in this fashion, so he adds a "signature" line (Incidentally, the line looks pretty hokey anyway with the cross hairs right in the middle followed by Z O delta A I K).
I’m a little unsure here. On the one hand, I share your opinion that a simple solution is the most likely (PS, have a look: https://en.wikipedia.org/wiki/Occam%27s_razor . I’ve posted that before, but I really like this concept.).
On the other hand, I don’t think Zodiac was aware of or thought about any statistics. He probably recognized the vulnerability of z408 ( repeated bigrams) from various newspaper articles. But if he actually used a transposition, then I don’t think he thought about or knew about any periodic repeats. However, your idea may well be the right one: a signature hidden somewhere in the cipher…that destroys the correct order and gives us what is causing us so many problems. An additional row or column…maybe even a diagonal line.
But at this point I think that the work of Jarlve and Smokie has excluded many options. AZDecrypt has an extremely good Nulls/Skips detector, which would probably have provided the right solution, even if z340 contains such a signature. What would be great would be sample ciphers that AZDecrypt can’t solve. You could try your own ideas and let them compete against the current solvers.
Translated with http://www.DeepL.com/Translator
I have been a strong proponent of route transposition for a few years now because of the evidence. If someone rides a bicycle through a puddle, and leaves wet tire marks on one side of the puddle, then we know that direction he was going even if we didn’t see him actually ride through the puddle. However, attempts at solving a one inscription rectangle on transcription rectangle transposition, even with up to 10 nulls or transcription errors ( skipped letters ), with a powerful program, have not yielded even a partial solve.
I agree that the signature may be filler, but even so, filler in the top or bottom row does not have nearly the misalignment effect after untransposition that filler or skips closer to the middle of a message. Filler in rows, even near the middle, would cause fewer problems than filler in columns.
I still think that transposition is likely. There could be some sort of subtle zigzag pattern, or multiple inscription rectangles, or some other simple but creative way to do it, and unfortunately with 63 symbols, it is so far not detected.
Work can still be done as far as removing the signature. That’s something that hasn’t been exhaustively worked on.
However, attempts at solving a one inscription rectangle on transcription rectangle transposition, even with up to 10 nulls or transcription errors ( skipped letters ), with a powerful program, have not yielded even a partial solve.
It may be possible that a partial solve appeared, but was obscured by the rest of the candidate plaintext which resembles gibberish. It’s probably hard to detect among so much gibberish!
– Is it possible that the 340 is a different kind of period 15/19 transposition such as diagonal transposition which the test did not cover?
– Is it likely that there are more than 10 nulls and/or skips given the increased amount of bigram diffusion?
I am currently working on a new test and can start attacking z340 today or tomorrow. Since I’m not that deep into the subject of nulls & skips, it’s possible that someone else already had my idea. If not, the following challenge cipher shouldn’t be solvable:
Fba:d9oEuDhir1AXL qclkUeZBw2d:DnQiC ubxUK5Ge=Mj7o4Q1p §3VhHYZc9Fyk$CRt2 bdDiQAp:eJv5jqbKh +7ZujQ;Ucej4sN;CM 5oX1b§dh0R=cPAeXD b9=E;qgKBIM7aXLQi Z:QYJ§x3CwFsNRLdu eU=4QwEHQo2p1TK:J Mjyb$Xku5+e0PhAcg anCqDUx57§=RiAhoZ 9cqE:I17C§3QbRsBd eJSuF5AKlTy$+QMjQ 0Gqohl;7NHX1DY=P2 kigcpXb4YeLaxQw§3 IZdjTIVyD$9mt4KI+ =UsiBZ0CHI;PdRQgY 9Xnka:DGxIABVi=NZ qtT3;7oB=FvZODAIK
EDIT:
It has 11 skips
Another note: I am not sure if the term "skips" is correct in my case. The transposition is misaligned in 11 places, so to speak. My test can solve the cipher, a first attempt with z340 was unsuccessful so far. However, I have only used 17×20 as base size until now.
Another note: I am not sure if the term "skips" is correct in my case. The transposition is misaligned in 11 places, so to speak. My test can solve the cipher, a first attempt with z340 was unsuccessful so far. However, I have only used 17×20 as base size until now.
Thanks for your cipher.
I can’t tell the period, can you? You removed/skipped 11 characters and added 11 other characters at the end of the cipher?
Seems interesting and I am happy that you are looking in to the matter. 11 skips creates allot of bigram diffusion.
Got this without removing or adding anything (simple transposition solver). I can’t read the plain text really but can see context "backing, vocal, sheffield, heavy metal".
Score: 22847.77 IOC: 0.0624 Multiplicity: 0.1852 Minutes: 2.51 Repeats: RSBACKINGVO ACKINGVO NINET (2) NETHE SBACK SCONS PT-to-CT cycles: 23135 Offset(X:148) Mirror(188*2) Diagonal(19*18,UTP,C:4) ANSESANDACNDHASCO NSFHLGUITIOEELLIO TTLICALSWHLSRICKS AVAGESBANDACKINGV OCALSEATIOLLENDRU MSBACKRBDETCALSPH ILCOLLEIIAVDARSBA CKINGVOCNNCSOOVIV IANCAMPBEAGGKSCFR SBACKINGVOLLUVABA YHHASBEENTDEFLEPP ARDEVITNSEGLISHRO CKBANNETODEDINNIN ETYSEVENMIFIVENIN SHEFFIELDIESTSCON SIDEREDPARNNNHHEN EWWAVEOFBRARNETHE AVYMETALMOEOYSHTS INCENINETERFTEDUX YNINETHEBAADNAMOA
Looks like it’s all from here:
https://en.wikipedia.org/wiki/Def_Leppard
Def Leppard are an English rock band formed in 1977 in Sheffield and is considered part of the new wave of British heavy metal movement. Since 1992, the band has consisted of Joe Elliott (lead vocals), Rick Savage (bass, backing vocals), Rick Allen (drums, backing vocals), Phil Collen (guitars, backing vocals), and Vivian Campbell (guitars, backing vocals), which has been the band's longest running line-up.
——————————–
I don’t believe in monsters.
Yep, you got it glurk.
While removing the last 11 characters AZdecrypt’s simple transposition solver gets it perfectly since I designed the diagonal transposition routine to handle incomplete rectangles:
Restart: 1 Hill climber: 181/5000 @ 500000 Score: 23375.71 IOC: 0.0627 Multiplicity: 0.1884 Seconds: 1.16 Repeats: GUITARSBACKINGVOCALS SBACKINGVOCALS SBACKINGVOCAL PT-to-CT cycles: 21800 Diagonal(17*20,UTP,C:5) DEFLEPPARDAREANEN GLISHROCKBANDFORM EDINNINETYSEVENTY SEVENINSHEFFIELDA NDISCONSIDEREDPAR TOFTHENEWWAVEOFBR ITITHHEAVYMETALMO VEMENTSINCENINETE ENNINETYNINETHEBA NDHATCONSISTEDOFJ OEELLIOTTLEADVOCA LTRICKSAVAGEBATSB ACKINGVOCALTRICKA LLENDRUMSBACKINGV OCALSPHILCOLLENGU ITARSBACKINGVOCAL SANDVIVIANCAMPBEL LGUITARSBACKINGVO CALSWHICHHASBEENT HEBAND
Impressive!
What does surprise me, though: I let AZDecrypts Transposition Solver process the cipher for about an hour and got no result. You got your result after 2.51 minutes, if I interpret it correctly. Did you use the default settings?
