I’ve done a large segment slide test to find out more about localized areas of higher encoding disturbance (2-symbol cycles measurement) possibly related to the pivots.
Observations and interpretation,
Segment size 8 to 17 all contain the horizontal part of the second pivot situated in row 14. Segment size 2 to 7 contain the only repeating 3-gram repeat in the 340 and is situated in row 13 above the second pivot. There is a peak improvement at segment size 12. Both horizontal pivot segments rank high on all lists. Other interesting segments occured around "L++)WCzWc" situated in row 18.
It is interesting that both pivots and the 3-gram repeat do so well here. In the 408 there is a repeating 4-gram repeat in row 24 which was probably dropped down (vertically aligned) from row 8 (glurk’s observation I believe). In the 340 this also seems to occur and it ranks high as null. Since the horizontal parts of the pivots also rank high as null it could be considered that Zodiac also dropped these down, though not vertically.
Here are the top entry from each segment size, the format is: "segment size: segment position start, improvement value, segment string":
17: 224, 77.11, FN^f524b.cV4t++yB 16: 225, 81.64, N^f524b.cV4t++yB 15: 224, 83.18, FN^f524b.cV4t++ 14: 225, 82.35, N^f524b.cV4t++ 13: 224, 82.45, FN^f524b.cV4t 12: 224, 90.21, FN^f524b.cV4 11: 225, 86.17, N^f524b.cV4 10: 228, 70.14, 524b.cV4t+ 9: 228, 70.16, 524b.cV4t 8: 228, 81.92, 524b.cV4 7: 211, 78.70, +N|5FBc 6: 212; 78.68, N|5FBc 5: 212: 72.30, N|5FB 4: 212: 68.13, N|5F 3: 212, 58.97, N|5 2: 212, 46.64, N|
Here’s a visual of segment size 5, brighter is more disturbed:
Thanks smokie,
Here is the solution to smokie40:
spoiler alert: BYTHETIMERALPHUIN ISHEDBLOWINGTPECO NCHTHEPLATFOEMWAS CROWDEDTHERAWERED IFFERENCESNETWEEN THISMEETILGANDTHE ONEHELDIBTHEMORNI NGTHEAFAERNOONSUN SLANTEDINFROMTHEO THERSIDEOFTHEPLAT FORMANDMOSTOFTHEC HILDRENFEELINGTOO LTTETHESMARTOFSUN NURNHADPUTTHEIRCN OTHESONTHECHOIRBO TICEABLYLESSOFEGR OUPHADDISCARDRDTH EIRCLOADSRALHHSAT ONAFALLENTRFNDHIS LEFTSIDETOTHESUNO
You may make another one. I would like to stick to 1:1 substitution for now and increase the difficulty by other means as gradually as possible.
O.k. The next two messages will illustrate a principle that I was going to talk about some time ago. They are almost exactly the same except for one thing. I am hoping that the first one is easier to solve than the second one.
smokie41 in which I reveal my true name:
1 2 3 4 5 6 7 4 8 6 9 10 11 4 12 13 13
14 15 12 16 17 4 18 12 7 19 16 12 11 18 7 14 16
12 16 16 7 5 7 19 12 5 7 4 4 18 4 12 3 13
2 13 17 4 12 8 20 5 16 12 2 16 11 7 19 2 19
21 20 5 11 13 13 17 16 8 13 7 6 2 13 12 8 21
5 15 13 6 9 4 18 7 4 19 15 4 2 12 18 7 7
15 12 21 12 15 2 2 15 16 7 3 7 21 15 7 4 13
21 22 2 4 18 13 10 20 17 16 6 8 15 7 4 9 18
6 19 7 19 12 13 7 19 3 13 4 19 7 2 4 12 15
7 4 7 15 21 2 9 19 19 15 7 9 12 18 14 9 2
7 7 12 12 5 17 9 22 7 13 8 17 12 19 7 12 18
15 7 13 12 16 18 13 23 16 13 11 16 3 15 4 13 4
15 16 19 7 13 12 19 20 3 19 2 18 13 7 18 19 7
7 16 4 15 4 23 12 18 15 7 3 23 7 15 5 15 4
4 7 12 22 19 8 11 4 17 4 17 9 7 14 9 3 16
19 4 2 16 15 13 19 17 2 18 4 13 18 19 16 7 13
9 16 4 13 4 19 7 6 12 12 2 2 7 18 7 23 4
19 7 3 12 13 2 4 17 15 2 7 15 7 13 12 13 7
18 7 18 19 3 18 19 2 12 12 20 15 2 7 18 7 11
17 2 18 18 19 17 7 7 4 7 5 17 7 17 4 4 5
EDIT: I didn’t really reveal my true name in smokie41, but I did in smokie42:
1 2 3 4 5 6 7 8 9 9 8 10 5 11 12 8 13
14 11 1 5 3 4 8 7 6 4 14 15 13 16 17 5 8
16 1 18 19 1 9 9 8 10 17 10 11 8 19 11 6 18
10 9 13 20 8 18 20 13 4 11 16 9 13 9 13 13 19
12 5 20 17 1 11 20 17 8 12 20 12 8 4 17 17 10
13 6 11 1 18 4 14 1 21 4 17 4 6 4 1 11 6
9 17 19 6 4 11 11 10 4 8 8 11 1 5 5 14 18
12 8 5 20 4 3 13 6 13 11 19 12 13 18 7 17 8
15 17 22 20 4 8 8 8 23 8 13 6 6 6 6 1 17
9 13 11 6 6 1 9 11 10 16 8 8 4 5 11 10 5
16 8 11 13 3 3 18 20 13 9 19 9 5 9 16 8 15
1 17 7 13 11 2 8 11 1 8 1 1 17 11 5 1 8
10 9 5 17 8 13 3 6 13 4 18 15 10 10 1 6 1
1 9 8 13 15 16 8 4 20 17 12 11 16 6 15 16 20
10 7 4 11 8 5 1 17 1 8 5 6 1 17 16 13 20
8 9 1 17 13 9 7 1 4 14 12 3 4 5 2 16 4
17 20 5 20 16 8 1 6 7 23 3 17 16 9 7 8 8
20 5 1 11 3 7 11 10 5 23 4 13 10 14 17 8 19
14 1 9 4 1 4 8 8 11 5 4 17 8 3 8 10 6
10 14 17 12 7 8 10 10 13 17 14 5 9 13 11 11 1
Interesting ciphers smokie,
I am working on them and getting scores around 20000 with readable bits here and there, enough to discern that it is an excerpt from the book Lord of the flies. I am not exactly sure what you did but most of the lower order ngrams (bigrams) are still there while the higher order ngrams repeats been diminished. It seems that the order of the messages has been altered somehow in relatively large chunks.
Hey smokie,
Regarding the smokie39,
I believe that column 10 is a null or that you randomized its positions, at least the numbers seem to suggest it. Especially the top and bottom part of the column show higher randomness so I thought of flipping it but that does not improve the stats by much. You may have chosen its symbol makeup carefully to fit in with the frequencies of the cipher. I ask of you, is column 10 a null or not?
I’ve made some progress on the smokie41 and 42. Here is a – reasonable or workable – solve of the smokie41 scoring 21912.
Do you need another cipher? By the way, thanks for maintaining and organizing this thread and all your work on the Zodiac ciphers. I really don’t know where we would be without you.
spoiler: Score: 21912.64 Ioc: 0.07029 MEDHISPSHINANESST AREHIATTHECDRACES ASTHOUTACKWASBENT DOUBLEHEWASDOWNLI KEASPRINTERHISNOS EONLYAFEWINCHESFR OMTHEHUMIDEARTHTH ETREETRUNKSANDTHE CREEPERSTHATFESTO ONEDTHEMLOSTTHEMS ELVESINAGREENDUSK THIRTYFEETABOVEHI MANDALLABTUTWASTH EUNDERGROWTHTHERI WASONLYTHEFKINTES OINDICETIONOMATRA IAHEREANRACKEYTWI GAMDWHATOIGHTBETH EIMERESSIANOFONAS IDEOLAHOOFDELOWEM
Regarding the smokie39, I believe that column 10 is a null or that you randomized its positions, at least the numbers seem to suggest it. Especially the top and bottom part of the column show higher randomness so I thought of flipping it but that does not improve the stats by much. You may have chosen its symbol makeup carefully to fit in with the frequencies of the cipher. I ask of you, is column 10 a null or not?
No. Knowing what I know, I can tell you that spending a great deal of time on smokie39 may not be fruitful. I have a basic system down, tried the system, and still cannot get a decent solve for smokie39 myself. There is a reason why, I think, and relates to when we were working together and you were trying to solve a message with individual segments. I didn’t intentionally make it so difficult. In fact, I only did a couple of things that I didn’t think would cause such problems. Stuff that we have talked about extensively. I would prefer that you ask me to show smokie39 instead of working on it, but that is your choice.
I’ve made some progress on the smokie41 and 42. Here is a – reasonable or workable – solve of the smokie41 scoring 21912.
I am glad. smokie41 should be easier than smokie42, and it is only slightly different. Unfortunately my system crashed and my spreadsheet is currently gone. I am going to looked for an autosaved version this morning.
Do you need another cipher? By the way, thanks for maintaining and organizing this thread and all your work on the Zodiac ciphers. I really don’t know where we would be without you.
I am fine for now. I may have some phrases / parts of sentences and am thinking of ways to detect abnormalities. Thanks for the compliment. I sincerely believe that we are on the general right track. But there are a lot of possibilities, variations of transposition, and homophonic substitution makes them difficult to detect. It is a murky science. Some things, however, are more difficult to detect with 1:1 substitution also. The correct period spike isn’t always the predominate one.
That makes me think… perhaps ask for some hints about smokie39 and you still may solve it. Whatever you want to do.
smokie41 used a 12 column x 28 row complete inscription rectangle. There is a message in row 10, written backwards, and transcription went LRTB and around the message in row 10. The principle is that a gibberish row does not make a message unsolve able. It creates a big misalignment, but a partial solve is quite possible.
Un transposed at period 28, then "repaired":
Nulls removed and un transposed at period 28:
It is Lord of the Flies.
Now, here is something very interesting. Something that I just recently realized.
The closer a gibberish row is to the top or bottom of a message, the easier the message is to solve because the uninterrupted rows are longer after untransposing. And the closer a gibberish row is to the middle of the message, the more difficult the message is to solve because the uninterrupted rows are shorter after untransposing.
Here is smokie41, except with the nulls on row 5. The uninterrupted rows are longer ( yellow and green sections go together ), allowing for a higher score.
Here is smokie41, except with the nulls on row 20. The uninterrupted row length is at a maximum, which would allow for the highest score possible.
With 63 symbols, a message may be solvable if the nulls are on row 1 or row 20, but maybe not if the nulls are on row 10 or 11. The location of a gibberish row affects how difficult a message is to solve.
Smokie39 is interesting, as it is a hard cipher and I’m in the unique position to ask questions. So I want to keep doing that. You answered that column 10 is a not a null. Is column 10 in any way different from the other columns, in other words, did you do something special with it or not?
I like it how you didn’t use a full row in smokie41. Nice visuals, they help me allot and it perfectly shows how the position of the added segment affects the difficulty. Could you make me another plaintext cipher such as the smokie41 with no further complications? It seems worthwhile to stand still on this for a while.
Column 10 has nothing special about it. It is the same as the other parts of the message that are part of the regular transcription.
I will make a message with a complete or nearly complete null row shortly.
Column 10 has nothing special about it. It is the same as the other parts of the message that are part of the regular transcription.
Okay thanks.
I’ve been working on a generic "substitution + transposition" solver for AZdecrypt with the aim of it being able to solve simple route transposition ciphers. It’s working quite well at the moment and in some aspects it is performing beyond expectations but there are some problems as well. It is not a magic wand and it needs to be taken very slowly. As part of the optimization process the transposition solver is running through some of the transposition ciphers I have collected including the 340 and smokie’s transposition ciphers. Feedback is very slow and better solves may slowly emerge over time.
Smokie, I have been using this tool to solve your latest ciphers. The smokie41 solve was a combination of 3 transposition operations without considering any nulls.
O.k. Here are smokie43A, 43B and 43C. All have complete inscription rectangles close to 340, but different, and all have a different plaintext. They have different null rows in different locations to see if some are more difficult than others.
Smokie43A
1 2 3 4 5 6 7 8 9 7 10 11 12 13 5 1 11
7 8 12 1 7 14 6 11 13 7 3 6 4 9 5 8 15
10 6 2 12 16 17 7 5 10 15 11 7 4 12 6 4 5
17 15 18 10 16 1 1 3 11 9 9 14 10 1 3 3 7
6 12 12 8 18 18 1 18 19 15 12 1 3 11 7 7 10
3 10 6 20 20 6 1 20 3 14 3 11 3 11 1 1 7
3 11 1 6 4 9 3 1 10 20 8 11 14 8 8 1 5
1 14 1 6 11 18 4 6 6 11 13 13 3 7 20 12 14
5 6 20 7 5 8 12 14 12 14 19 6 3 5 6 16 14
4 10 6 2 11 8 20 12 5 14 5 6 2 1 5 14 10
10 6 6 8 15 7 1 18 6 20 15 8 4 14 14 1 3
10 1 2 11 1 11 19 14 12 14 5 4 10 16 3 14 8
14 13 1 18 2 11 20 6 4 8 6 15 7 8 5 16 14
16 3 12 4 1 5 1 16 11 20 7 1 6 14 3 6 5
10 14 9 10 6 6 5 18 21 6 6 14 8 11 5 1 8
5 6 2 4 3 6 1 8 4 3 10 1 7 12 7 6 5
20 10 5 11 6 14 1 1 4 7 15 14 3 11 1 1 6
14 15 2 6 9 14 2 2 20 16 14 1 21 11 11 5 3
14 4 18 3 12 2 3 6 5 22 17 15 23 6 5 22 3
6 1 8 7 4 14 14 24 13 1 4 4 14 1 5 8 8
Smokie43B
1 2 2 3 2 4 5 2 2 6 7 1 8 3 9 6 10
8 11 1 3 12 11 9 1 3 8 5 2 1 13 2 14 8
2 9 1 4 6 7 15 6 14 16 2 2 15 14 6 7 16
2 9 2 1 16 6 8 10 12 9 7 1 1 5 15 7 8
16 1 14 15 17 8 7 15 12 18 4 13 1 1 13 17 5
19 3 20 5 11 16 1 15 21 22 17 22 11 20 2 22 8
13 15 2 2 1 3 5 17 1 3 9 3 5 5 7 20 2
15 14 1 6 2 13 5 1 1 16 15 7 7 11 13 7 5
16 10 9 17 16 1 15 3 1 9 5 1 15 14 5 16 8
17 2 12 6 1 18 23 3 3 13 5 6 3 15 1 10 8
20 16 8 17 7 5 7 3 13 2 13 6 9 11 19 7 10
5 15 6 9 8 15 19 3 13 9 7 8 15 3 24 3 20
2 1 1 12 23 17 7 15 2 1 7 5 19 5 17 16 6
1 15 2 1 12 12 8 1 7 15 3 16 3 9 9 16 3
15 3 5 2 16 5 6 2 9 9 12 5 1 1 8 1 12
20 7 8 20 19 5 1 1 17 8 16 16 17 1 17 15 23
1 9 17 5 6 15 3 5 2 19 9 24 3 5 8 5 1
12 7 6 15 6 2 15 5 23 7 6 3 6 3 2 5 3
6 16 8 6 17 7 5 16 3 13 2 6 16 1 9 6 3
7 18 3 6 19 9 15 19 9 13 13 2 2 24 15 2 2
Smokie43C
1 2 3 4 2 5 3 6 7 8 4 4 9 7 10 5 1
11 12 1 7 13 1 14 5 15 1 1 4 2 9 16 3 16
17 18 7 5 1 10 4 19 18 18 18 17 2 20 14 10 3
1 15 12 1 11 5 18 14 4 14 10 10 15 10 10 10 9
18 17 18 7 9 5 19 21 15 10 12 1 18 17 19 1 17
16 17 4 2 1 5 14 5 15 17 5 14 17 9 1 2 1
19 12 1 4 1 10 7 19 12 4 18 14 14 15 8 1 18
12 7 2 1 10 12 18 12 1 4 17 17 10 1 8 2 10
20 15 12 20 4 1 14 9 14 1 16 18 12 14 12 5 19
22 8 16 5 14 3 5 12 14 4 4 7 2 18 15 6 2
23 6 6 13 15 14 8 9 1 16 15 9 17 9 5 21 7
4 4 18 17 8 6 2 12 12 1 14 20 1 3 10 5 18
12 5 6 1 1 14 1 8 17 12 14 17 17 18 8 18 1
17 18 10 10 15 1 12 12 8 18 1 14 1 18 14 9 2
9 10 4 1 9 17 17 1 2 3 3 12 10 2 2 14 2
2 16 20 3 16 1 20 20 1 1 12 21 5 5 5 17 10
8 20 10 8 12 9 1 4 18 14 10 19 12 9 14 10 7
18 1 17 5 9 17 5 1 8 4 5 9 5 17 1 9 18
5 17 7 10 3 1 22 4 1 22 3 8 8 1 10 4 1
8 20 15 18 15 5 17 5 7 1 10 5 1 17 9 5 12
There are more, longer words in context with each other so a partial solve should be easy to identify. I think you will like them.
Thanks smokie,
All three solve very well, even without considering nulls. I would like to step up the difficulty, perhaps multiple segments or increase symbol count. Considering increased symbol count I think it would be nice to have an intermediate between plaintext and 63 symbols, that would come out at around 45 symbols then. Do what you think is right.
Here are the segment regions the transposition solver found when considering nulls and the codebox after shows the solves and the scores.
smokie43a: from 308 to 321 smokie43b: from 87 to 99 smokie43c: from 172 to 183 smokie43a: Score: 23728.80 Ioc: 0.06403 Nulls(Remove,F:308,T:321,P:1) Period(UTP,P:24) THEYHADAHOUSEOFCR YSTALPILLARSONTHE PLANETMARSBYTHEED GEOFANEMPTYSEAAND EVERYMORNINGYOUCO ULDSEEMRSKEATINGT HEGOLDENFRUITSTHA TGREWFROMTHECRYST ALWALLSORCLEANING THEHOUSEWITHHANDF ULSOFMAGNETICDUST WHICHTAKINGALLDIR TWITHTBLEWAWAYONT HHOTWINDAFTERNONS WHENTHEFOSILSEAWA SWARMADMOTIONLESS AYTHEWINETREESSOO DSTIFFINTHEARDAND THELITTEDISTANTMA RTI smokie43b: Score: 22880.61 Ioc: 0.06491 Nulls(Remove,F:87,T:99,P:1) Period(UTP,P:14) ETOWNWASALLENCLOS EDANDNOONEDRIFTED OUTTHEIRDOORSYOUC OULOSEEMRKHIMSELF INHISROOMREADINGF ROMAMETALBOOKWITH RAISEDHIEROGLYPHS OVERWHICHHEBRUSHE DHISHANASONEMIGHT PLAYAHARPANDFOMTH EBOOKASHISFINGERS STOKEDAVOICESANGA SOFTANCINTVOICEWH ICHTOLDTALESOFHEN THESEAWASREDSTEAM ONTESHOREANDANCIE NTMENHADCRRIEDCLO UDSOFMETALINSECSA NDELECTRICSPIDERS INTO smokie43c: Score: 23427.08 Ioc: 0.06727 Nulls(Remove,F:172,T:183,P:1) Period(UTP,P:20) EMRANDMRSTHADLIKE DBYTHEDEADSEAFORT WENTYYEARSANDTHEI RANCESTORSHADLIKE DINTHESAMEHOUSEWH ICHTURNEDANDFOLLO WEDTHESUNFLOWERLI TEFORTENCENTURIES MRANDMRSTWERENOTL DTHEYHADTHEFAIRRO WNISHSPINOFTHERUE MARTIONTHEYELOWCO INEYISTHESOFMUSIC ALKPICESONCTHEYHA DLMTEDPAININGPICT URESWITHCEMICALFI RESWIMMIGINTHECAN ALSINTHSEASONSWHE NTHEWIETREESFILLE DTHEM
K I was already thinking about 45 symbols too. I will make some tonight or maybe tomorrow.
I goofed up on smokie42. It was supposed to have one mostly gibberish column, but it actually has one gibberish column and the last row is not transposed, essentially making it gibberish. So it is not slightly more difficult than smokie41. Here is the plaintext, transposed but not encoded. I imagine that Zodiac probably did something like this making the 340 transposition detectable, but not easily solved.
O K C R S A P E D D E N S T W E H
U T O S C R E P A R U G H L I S E
L O M Y O D D E N I N T E Y T A M
N D H F E M F H R T L D H D H H Y
W S F I O T F I E W F W E R I I N
H A T O M R U O V R I R A R O T A
D I Y A R T T N R E E T O S S U M
W E S F R C H A H T Y W H M P I E
G I J F R E E E B E H A A A A O I
D H T A A O D T N L E E R S T N S
L E T H C C M F H D Y D S D L E G
O I P H T K E T O E O O I T S O E
N D S I E H C A H R M G N N O A O
O D E H G L E R F I W T L A G L F
N P R T E S O I O E S A O I L H F
E D O I H D P O R U W C R S K L R
I F S F L E O A P B C I L D P E E
F S O T C P T N S B R H N U I E Y
U O D R O R E E T S R I E C E N A
N U I W P E N N H I U S D H T T O
