Also.
I appreciate this might be just down to the high number of + symbols, but obviously 8 of the 12 short varying lengths begin with a + symbol. And 7 of the 12 end with a +.
I wondered if this was a hint that the + symbol was an instruction to shift alphabets?
Well, the genuine lengths most of them would be longer but for the high frequency symbols, and a few lengths would be longer but for the regional bias symbols. I mean, the high frequency symbols do not define the lengths, it is the symbols within the lengths that define the lengths. The high frequency symbols only serve to keep them as short as they are. In the back of my mind lately I have been thinking about patterns encoding maybe or nulls. In 2015 we looked at cycles with just odds, and just evens, and broke up the message into different parts, like three parts and four parts etc. One project to consider and I could do this myself, separate the message into different parts, but with all different combinations of two position patterns up to a certain length, then check for cycles. For example, AABBAABB or ABAABAABA or AAAABBBB etc. Because I am thinking about a simple cipher, a scytale, but encoding the columns with a pattern from two different directions. I wonder if that could also make the genuine lengths longer.
Here is a picture of my idea. A scytale cipher encode yellow perfect cycles LRTB, then encode blue perfect cycles RLTB. Possibly something like this could make the cycles, 25%-30% randomization, and 26 L>=17. It would make the cycles longer going LRTB, and depending on how the cycles align going RLTB, could maybe make 26 L>=17.
Here is a picture of my idea. A scytale cipher encode yellow perfect cycles LRTB, then encode blue perfect cycles RLTB. Possibly something like this could make the cycles, 25%-30% randomization, and 26 L>=17. It would make the cycles longer going LRTB, and depending on how the cycles align going RLTB, could maybe make 26 L>=17.
I created such a cipher though the ioc does not match so well:
Y85OD>'>ADL5*YO 8-K=::=>/-EK6VS W4H/;!]BPT307S, ^KUR])VH.0IY>I# K>J>7S[>%NHZ9PQ D/3+="0%<GM(ZU. YM(BCKLQ.'SA3OX N*"OBFJ5SAI@4O 89+)(TVWK";6.O? WO^?>-3=$]S2!<= :1##3I>CP(L@-X ,P94B&D>KM>DLV/ R%ZCO*YK'ES]_T] .^A>&6J3F4RK*B [EJSMP8.WS"G^I. <D,(>K)';(.,!N1 V]GP>5O[K.+$9/K .>#!+T=3Q2MFZ@C 1/8HUK_.K7X.K> KC&-K-3A$)Y]70? COP=VL@U2S.P4Y; *)]O0:'PKRABWR. 1 2 3 4 5 6 7 6 8 5 9 3 10 1 4 2 11 12 13 14 14 13 6 15 11 16 12 17 18 19 20 21 22 15 23 24 25 26 27 28 29 30 31 19 32 33 12 34 35 25 36 18 22 37 30 38 1 6 38 39 12 6 40 6 31 19 41 6 42 43 22 44 45 27 46 5 15 29 47 13 48 30 42 49 50 51 52 44 34 37 1 51 52 26 53 12 9 46 37 7 19 8 29 4 54 55 43 10 48 4 26 56 40 3 19 8 38 57 21 4 2 45 47 36 52 28 18 20 12 48 23 17 37 4 58 20 4 33 58 6 11 29 13 59 25 19 60 24 49 13 14 61 39 39 55 29 38 6 53 27 52 9 57 11 54 32 27 45 21 26 62 5 6 12 51 6 5 9 18 15 35 42 44 53 4 10 1 12 7 16 19 25 63 28 25 37 33 8 6 62 17 55 40 29 56 21 35 12 10 26 41 16 40 19 51 27 2 37 20 19 48 50 33 38 37 49 5 32 52 6 12 36 7 23 52 37 32 24 43 61 18 25 50 27 6 3 4 41 12 37 47 59 45 15 12 37 6 39 24 47 28 13 29 46 60 51 56 44 57 53 61 15 55 2 22 34 12 63 37 12 31 54 37 12 6 12 53 62 11 12 11 29 8 59 36 1 25 31 30 58 53 4 27 13 18 9 57 34 60 19 37 27 21 1 23 10 36 25 4 30 14 7 27 12 35 8 26 20 35 37
For this cipher there appears to be a right-shifted peak at L14=24 so that is a good sign:
But there are big problems:
– High number of unigram repeats counted by rows (38 per 17) because of mixing 2 encoding alphabets like that.
– Visible randomization by columns (see image below).
Yeah but what about using the same encoding alphabet for both yellow and blue? Did you use the same key for both or different keys?
Previous cipher are 2 different keys. Reusing the same key does not serve much purpose.
Here is one that use the same key for both directions. It looks better but I have my reservations.
VU[*@Q"!-E%^HC3 C5ZZ"*.AM5[V9C- ]O?QD'J.,=MC_LJ 4]UA&C!,[?1C/F: O-"SR#$<R??(C0$ EDS8I?NBI^T+C3F V586+P&N@Q-[M2 C&@?2CY[9R!,IZ3 C1A==CO6W85)Z*G ]24[&SM9F%-#DY "G:[M1I0RWI[85 .0CN7E+,STH"'CA X8C&/VYQT[R:$%P O?!-9C3S9C^AF$* #>"KJC5].-IB46+ ;"WZN5M&86HD&L CQ^0[3RZ=,A[CVO FS:/=2I$U(T'C^0 )]CG>+[N[6__Z!@ ,&755M:9[&VD_CU 0*.,%ICK1RO0^C5 H#'2"QCF0B!O]A+ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 14 16 17 17 7 4 18 19 20 16 3 1 21 14 9 22 23 24 6 25 26 27 18 28 29 20 14 30 31 27 32 22 2 19 33 14 8 28 3 24 34 14 35 36 37 23 9 7 38 39 40 41 42 39 24 24 43 14 44 41 10 25 38 45 46 24 47 48 46 12 49 50 14 15 36 1 16 45 51 50 52 33 47 5 6 9 3 20 53 54 14 33 5 24 53 14 55 3 21 39 8 28 46 17 15 14 34 19 29 29 14 23 51 56 45 16 57 17 4 58 22 53 32 3 33 38 20 21 36 11 9 40 25 54 55 7 58 37 3 20 34 46 44 39 56 46 3 45 16 54 18 44 14 47 59 10 50 28 38 49 13 7 26 14 19 60 45 14 33 35 1 55 6 49 3 39 37 41 11 52 23 24 8 9 21 14 15 38 21 14 12 19 36 41 4 40 61 7 62 27 14 16 22 18 9 46 48 32 51 50 63 7 56 17 47 16 20 33 54 45 51 13 25 33 31 14 6 12 44 3 15 39 17 29 28 19 3 14 1 23 36 38 37 35 29 53 46 41 2 43 49 26 14 12 44 57 22 14 58 61 50 3 47 3 51 30 30 17 8 5 28 33 59 16 16 20 37 21 3 33 1 25 30 14 2 44 4 18 28 11 46 14 62 34 39 23 44 12 14 16 13 40 26 53 7 6 14 36 44 48 8 23 22 19 50
Peak at L12=25. Acceptable shape?
– 25 unigram repeats by rows of 17. Acceptable I guess.
– Grid encoding randomness still revealing columns more or less (image below).
Same image for the 340:
O.k., I tried it too and it doesn’t work to make long genuine lengths. Thanks for looking at it.
I created such a cipher though the ioc does not match so well:
For this cipher there appears to be a right-shifted peak at L14=24 so that is a good sign:
Also, when comparing this to the z340, the number of real lengths doesn’t peak around 14,15 and 16, as below (real lengths in orange).
compared to z340
Again, this seems to be the defining feature of z340 to aim to reproduce, not just the right shifted peak.
