This one smokie And great idea. We do a $ 200 Chris Kringle.. I’ll put it on my wish list.. Or just buy one…hoho
That is an example of what I was working to confirm with Jarlve. He wanted me to make a message for him that randomly placed diagonal fragments. That is why the diagonal row of letters starts over at the top. It is randomly placed. Jarlve was working on trying to solve a message by breaking down the message into 55 individual fragments that are not necessarily in order. Because we can’t figure out how to un-transpose the 340.
Doranchak, wow you have really put a lot of work into this. I was wondering if perhaps the same five or six letter word, with potential context, appears EDIT: in multiple solves. Jarlve’s recent work with trying to solve the diagonal rows as individual fragments but together in the solver shows that despite a solve being almost totally gibberish, a few of the words survive the fragmentation process and can be found.
Doranchak, wow you have really put a lot of work into this. I was wondering if perhaps the same five or six letter word, with potential context, appears EDIT: in multiple solves. Jarlve’s recent work with trying to solve the diagonal rows as individual fragments but together in the solver shows that despite a solve being almost totally gibberish, a few of the words survive the fragmentation process and can be found.
That’s a very compelling idea. Here’s the breakdown of words found, with lengths ranging from 4 to 12:
https://docs.google.com/spreadsheets/d/1BJKLTvjoZ1tgmjhFz5w4BWkWg2B2dlGTREqp4CqNrfA/edit?usp=sharing
Click the tabs along the bottom to see different word lengths.
I very much want to expand this analysis by including word-level ngram statistics in my word search of the azdecrypt results. That way, more interesting and grammatically correct combinations of words can be found (i.e., find words that have good context together, as you’ve suggested).
@smokie, there may be some filler going from row 2 to 3 but I think that the row 2 to 3 section is just low on repeating fragments by chance but it may be worth exploring!
@Mr lowe, I ran your odds and evens separately 1000 restarts @ 1.000.000 iterations using 6-grams. Odds scored 21872. Evens scored 21685. Considering that these are 170 character portions I also ran a 170 character 408 which scored 23793. No solve.
@doranchak, wow nice work! That’s allot to take in. I like these swap operations you’ve considered. It’s a bit funny that 3×3 shows up again.
4) There seems to be some correlation between azdecrypt score and the number of bigram repeats.
Yes, I have noted it before. I think it’s because repeating fragments are a feature of language and that’s captured in the n-grams somehow. Or it’s because they have a slighty higher IoC on average.
5) I noticed a strong response in L=2 homophone cycles with the following transposition scheme: PeriodColumn(2) Swap(3, 156, 3, 3)
Yes, the cycles at the top of the list are of better quality but overall my measurement gives a slightly lower score for this string.
Jarlve, the highest azdecrypt score I found in my recent results was 20900. Do you think that is a significant bump from the average for 340-character cipher texts? I think the original 340 scores around 20300.
Hard to say, my advice is to conduct some kind of experiment that tries to answer that question. Yes, 20351.
@doranchak, wow nice work! That’s allot to take in. I like these swap operations you’ve considered. It’s a bit funny that 3×3 shows up again.
Can you remind me where 3×3 showed up before?
Yes, I have noted it before. I think it’s because repeating fragments are a feature of language and that’s captured in the n-grams somehow. Or it’s because they have a slighty higher IoC on average.
That makes sense.
5) I noticed a strong response in L=2 homophone cycles with the following transposition scheme: PeriodColumn(2) Swap(3, 156, 3, 3)
Yes, the cycles at the top of the list are of better quality but overall my measurement gives a slightly lower score for this string.
Ah! Thanks for checking that. Was it this measurement? viewtopic.php?p=42120#p42120 I will run your measurement on all the current results, assuming I can reproduce the calculation properly. For some reason, my implementation gets 1599.5929 for Z340 (very close to yours) but 1378.7345 for Z408 (not very close to yours). Here’s my source: http://pastebin.com/raw.php?i=XLDp7tPB
UPDATE: I applied that measurement to the transpositions in https://docs.google.com/spreadsheets/d/ … sp=sharing. The best result was a reduction from 1599.5929 to 1398.8168 by applying the Swap(217, 85, 5, 4) operation, which looks like this:
Original 340: HER>pl^VPk|1LTG2d Np+B(#O%DWY.<*Kf) By:cM+UZGW()L#zHJ Spp7^l8*V3pO++RK2 _9M+ztjd|5FP+&4k/ p8R^FlO-*dCkF>2D( #5+Kq%;2UcXGV.zL| (G2Jfj#O+_NYz+@L9 d<M+b+ZR2FBcyA64K -zlUV+^J+Op7<FBy- U+R/5tE|DYBpbTMKO 2<clRJ|*5T4M.+&BF z69Sy#+N|5FBc(;8R lGFN^f524b.cV4t++ yBX1*:49CE>VUZ5-+ |c.3zBK(Op^.fMqG2 RcT+L16C<+FlWB|)L ++)WCzWcPOSHT/()p |FkdW<7tB_YOB*-Cc >MDHNpkSzZO8A|K;+ First box: (;8R 4t++ Z5-+ MqG2 B|)L Second box: p8R^ #5+K (G2J d<M+ -zlU Both boxes in grid: HER>pl^VPk|1LTG2d Np+B(#O%DWY.<*Kf) By:cM+UZGW()L#zHJ Spp7^l8*V3pO++RK2 _9M+ztjd|5FP+&4k/ ====FlO-*dCkF>2D( ====q%;2UcXGV.zL| ====fj#O+_NYz+@L9 ====b+ZR2FBcyA64K ====V+^J+Op7<FBy- U+R/5tE|DYBpbTMKO 2<clRJ|*5T4M.+&BF z69Sy#+N|5FBc____ lGFN^f524b.cV____ yBX1*:49CE>VU____ |c.3zBK(Op^.f____ RcT+L16C<+FlW____ ++)WCzWcPOSHT/()p |FkdW<7tB_YOB*-Cc >MDHNpkSzZO8A|K;+ After swap: HER>pl^VPk|1LTG2d Np+B(#O%DWY.<*Kf) By:cM+UZGW()L#zHJ Spp7^l8*V3pO++RK2 _9M+ztjd|5FP+&4k/ (;8RFlO-*dCkF>2D( 4t++q%;2UcXGV.zL| Z5-+fj#O+_NYz+@L9 MqG2b+ZR2FBcyA64K B|)LV+^J+Op7<FBy- U+R/5tE|DYBpbTMKO 2<clRJ|*5T4M.+&BF z69Sy#+N|5FBcp8R^ lGFN^f524b.cV#5+K yBX1*:49CE>VU(G2J |c.3zBK(Op^.fd<M+ RcT+L16C<+FlW-zlU ++)WCzWcPOSHT/()p |FkdW<7tB_YOB*-Cc >MDHNpkSzZO8A|K;+
The mirrored placement of the boxes is interesting to me, since they are adjacent to each edge of the cipher grid. And the first box ends at line 10, consistent with the "cipher is in two parts (lines 1-10 and lines 11-20)" idea.
Now I’m running all possible rectangular swaps for the original Z340 and this is interesting:
Original 340: HER>pl^VPk|1LTG2d Np+B(#O%DWY.<*Kf) By:cM+UZGW()L#zHJ Spp7^l8*V3pO++RK2 _9M+ztjd|5FP+&4k/ p8R^FlO-*dCkF>2D( #5+Kq%;2UcXGV.zL| (G2Jfj#O+_NYz+@L9 d<M+b+ZR2FBcyA64K -zlUV+^J+Op7<FBy- U+R/5tE|DYBpbTMKO 2<clRJ|*5T4M.+&BF z69Sy#+N|5FBc(;8R lGFN^f524b.cV4t++ yBX1*:49CE>VUZ5-+ |c.3zBK(Op^.fMqG2 RcT+L16C<+FlWB|)L ++)WCzWcPOSHT/()p |FkdW<7tB_YOB*-Cc >MDHNpkSzZO8A|K;+ First box: p+B(# y:cM+ pp7^l 9M+zt 8R^Fl 5+Kq% G2Jfj <M+b+ zlUV+ Second box: cyA64 7<FBy pbTMK M.+&B Bc(;8 cV4t+ VUZ5- .fMqG lWB|) Both boxes in grid: HER>pl^VPk|1LTG2d N_____O%DWY.<*Kf) B_____UZGW()L#zHJ S_____8*V3pO++RK2 ______jd|5FP+&4k/ p_____O-*dCkF>2D( #_____;2UcXGV.zL| (_____#O+_NYz+@L9 d_____ZR2FB=====K -_____^J+Op=====- U+R/5tE|DYB=====O 2<clRJ|*5T4=====F z69Sy#+N|5F=====R lGFN^f524b.=====+ yBX1*:49CE>=====+ |c.3zBK(Op^=====2 RcT+L16C<+F=====L ++)WCzWcPOSHT/()p |FkdW<7tB_YOB*-Cc >MDHNpkSzZO8A|K;+ After swap: HER>pl^VPk|1LTG2d NcyA64O%DWY.<*Kf) B7<FByUZGW()L#zHJ SpbTMK8*V3pO++RK2 _M.+&Bjd|5FP+&4k/ pBc(;8O-*dCkF>2D( #cV4t+;2UcXGV.zL| (VUZ5-#O+_NYz+@L9 d.fMqGZR2FBp+B(#K -lWB|)^J+Opy:cM+- U+R/5tE|DYBpp7^lO 2<clRJ|*5T49M+ztF z69Sy#+N|5F8R^FlR lGFN^f524b.5+Kq%+ yBX1*:49CE>G2Jfj+ |c.3zBK(Op^<M+b+2 RcT+L16C<+FzlUV+L ++)WCzWcPOSHT/()p |FkdW<7tB_YOB*-Cc >MDHNpkSzZO8A|K;+
My version of your measurement drops all the way 1215.593 for this one. And it has that nice "lines up with sides and the split between lines 1-10 and 11-20" quality.
.
Doranchak, thanks for posting the words that all of your transpositions found. There were a lot of them. I wonder if any high count context words map to the same ciphertext fragments across different un-transpositions. This entire transposition project is huge to say the least.
My project for the week is to try to disprove the transposition idea with the tools and resources that I have. I am going to find which of Jarlve’s 100 plaintext messages has the highest count of period 19 plaintext matches and then make a message with that with the highest number of symbols and period 19 matches possible.
I may also work on coloring my period 19 matches on my spreadsheet.
EDIT: I have another idea that may have to wait until next year. The heatmap finds locations where adding a possible skipped symbol or deleting a possible extra symbol will increase the count and score for period 19 matches. What about doing that and also taking into consideration increase of count and score for period 38 matches, which untransposed would be period 2 matches. Here is the top part of the list for Zodiac letter period 2 bigrams by count. The English language has them as it does period 1 bigrams, because words like THE are high in count and the period 2 bigrams would be T.E. Maybe that would be a way to make the heatmap more accurate. Maybe also take into consideration period 3 or more? Could that be a more accurate way to detect skipped or extra symbols?
T.E 296
E.T 145
A.E 134
E.O 134
E.A 118
O.T 117
E.I 116
E.E 111
O.E 110
I.E 105
I.G 100
T.I 94
I.T 88
I.H 82
I.I 76
A.T 75
T.A 72
S.E 72
N.T 72
E.H 71
S.A 69
T.H 68
A.L 67
O.I 67
T.O 66
I.L 65
H.N 65
O.L 64
E.N 63
N.I 63
N.H 62
Y.U 62
S.H 61
O.S 61
S.O 60
A.I 60
N.O 59
E.S 58
N.E 58
I.A 58
H.S 56
L.N 52
H.M 51
H.T 51
L.E 50
.
Smokie, those ideas sound interesting and I look forward to hearing more about your progress along those lines. I am wondering if I should add a new operator to my set of transposition operators that would add/remove a symbol or two or three, similar to your approach. Then maybe the search would explore different combinations of misalignments that might lead to improved scores.
I am still very interested in the azdecrypt score bump to 20900 (and bigram bump to 48) I observed for the operations "PeriodColumn(2) Period(18) Swap(202, 104, 8, 1)". The swap operation exchanges boxes of height 8 and width 1. Perhaps these kinds of exchanges of small regions might be another way of "fixing" misalignments to the transposition scheme.
I really want to make some animations of the transposition schemes, too, to make it easier to visualize them. So many curious paths!
The heatmap spreadsheet is pretty massive and probably wouldn’t be able to do both periods 19 and 38 at the same time. But I made it so that I can change the period in one cell. I think that maybe I can modify a bit and calculate different periods separately and add the results. Then make a new heatmap with those values and see if it works better.
EDIT: Maybe compare the results. Or sort by the score for period 19 and then for period 38. Try different approaches.
Can you remind me where 3×3 showed up before?
Ah! Thanks for checking that. Was it this measurement?
No it was my 2-cycle measurement which includes all 2 symbol cycles. The code is a bit messy but I may share it after an update.
For some reason, my implementation gets 1599.5929 for Z340 (very close to yours) but 1378.7345 for Z408 (not very close to yours).
I normalized by cipher length. score=(score/cipher_length)*340.
My version of your measurement drops all the way 1215.593 for this one. And it has that nice "lines up with sides and the split between lines 1-10 and 11-20" quality.
I did allot of work in this direction through the start of the year and noticed that for each measurement system I could find such pockets/hot swaps in the 340 (probably just because of the higher randomness in the cipher). I want to say that these measurements certainly aren’t the all in one answers to our problems with the 340. And I believe they become less useful when the set of operations considered becomes larger.
About the Column period 2 bigram bump thing. It’s stuff like this that I find very interesting. I’ve managed to recreate this observation in a test cipher very easily (2nd attempt) without actually using the operation. The scheme: for each step move left by 2 and move up by 1 and wrap around the cipher. So it may be of some worth in trying to deduct why the operation increases bigrams in the first place, probably because the operation causes it to behave more like a regular period scheme. Which is what we use the measure bigrams. If the 340 is not using a regular period scheme then trying to shoehorn it in one may not be the solution.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 12 26 27 11 28 28 29 30 31 11 14 8 32 33 34 35 30 36 4 37 18 34 19 38 39 5 6 9 15 7 16 40 41 25 22 32 5 23 1 42 27 13 43 21 44 28 45 46 3 47 6 21 17 39 12 4 26 33 42 35 38 10 48 24 41 8 14 43 17 48 9 15 30 28 19 20 35 49 22 29 5 50 36 1 44 8 49 40 3 10 31 19 13 30 39 50 20 2 34 42 26 2 6 9 48 47 45 38 15 7 49 22 46 41 16 5 18 24 28 6 26 44 35 21 45 23 43 27 12 40 21 14 3 31 50 29 30 10 36 24 25 1 17 49 44 9 50 7 38 4 49 47 8 31 3 13 19 35 18 30 50 20 15 29 41 23 26 22 27 32 38 49 36 39 5 4 41 33 18 1 6 13 19 10 25 16 23 47 42 32 26 12 20 33 8 43 27 9 38 29 40 4 14 16 46 45 25 18 36 17 39 15 24 42 23 22 32 50 11 12 48 46 1 27 43 13 40 44 4 8 33 3 18 41 7 5 10 6 20 29 34 14 25 9 17 39 36 1 16 15 35 19 7 48 2 22 31 42 47 43 13 5 45 24 8 44 6 26 3 11 34 11 17 38 20 2 32 12 9 2 15 31 22 23 5 29 39 36 14 2 1 46 7 49 48 7 6 40 35 42 43 13 34 50 20 33 16 9 27 48 29 24 25 41 19 :S?Y*4B1#AG)R=/ZW MU$,3'L+)9NG66@-F G=1K>VQ-&YXMVUC!* 4#/BZD+3K*':JNR8 ,T6.5?I4,W!)Y9>JQ CAOLD1=8WO#/-6U$Q [3@*"&:T1[?AFUR- !"$SVJ9S4#OI.C/B[ 35DZ*ML649TQ,.'8N ),=?F"@-A&L+:W[T #"BCY[I1F?RUQM-"$ /@D'93NKC[&!*YD>M :4RUA+Z'IJK9)$>18 N#C@Y=Z5.+M&W!/L J'3K"G)O5:N8RTY1 >?MDB*A4$@V=+#W!& :Z/QUBOS3FJI8R*.L 1T49?GVGWC$SK)#S/ F3'*@!&=S:5B[OB4 QJ8RV"$>Z#NO@L+DU
My project for the week is to try to disprove the transposition idea with the tools and resources that I have.
That will be fun.
Ah! Thanks for checking that. Was it this measurement?
No it was my 2-cycle measurement which includes all 2 symbol cycles. The code is a bit messy but I may share it after an update.
OK, thanks. I tried another flavor of 2-symbol cycle scoring, based on an exponential summation (higher "runs" of cycles are given greater rewards). I updated my transposition spreadsheet with these scores. A result is that this transposition produces the best increase compared to the original 340 (increase from 248 to 579):
LinearSelection(106, 61) PeriodRow(3) Swap(1, 139, 7, 1) SwapLinear(83, 20, 47) H+R>pl^VPk|1LTG2d SJp8R|H.3zBK(Op^. fMqG2|ckdW<7tB_YO B*-CcNp+B(#O%DWY- z+9Sy#+N|5FBc(;7^ l8*V3pO++RK2#/+Kq %;2UcXGV.zL|-NlUV +^J+Op7<FBy.<*Kf) _9MEztjd|5FP+&4k/ (G2pfj#O+_NYz+@L9 U+R55tE|DYBpbTMKO lGFz^f524b.cV4t++ RcT6L16C<+FlWB|)L >MDcNpkSzZO8A|K;+ By:FM+UZGW()L#zHJ p8R^FlO-*dCkF>2D( d<M+b+ZR2FBcyA64K 2<clRJ|*5T4M.+&BF yBX1*:49CE>VUZ5-+ ++)WCzWcPOSHT/()p
Can you run your measurement on that cipher? How does it compare to Z340?
Here are the top 30 2-symbol cycles appearing in the transposed ciphertext:
p| [p|] [p|] [p|] [p|] [p|] [p|] [p|] [p|] [p|] [p|] p 10 6.0862344E-31 0.95238096 5.796414E-31 |z | [|z] [|z] [|z] [|z] [|z] [|z] [|z] [|z] [|z] 9 1.0509019E-28 0.94736844 9.9559125E-29 Rl [Rl] [Rl] [Rl] [Rl] [Rl] [Rl] [Rl] R 7 6.2605362E-24 0.93333334 5.8431673E-24 zF z [zF] [zF] [zF] [zF] [zF] [zF] [zF] FFFz 7 1.7337753E-22 0.7368421 1.2775186E-22 O| || [O|] O [O|] [O|] [O|] [O|] [O|] [O|] [O|] O 7 3.6248936E-22 0.7 2.5374255E-22 ^. [^.] [^.] [^.] [^.] [^.] [^.] 6 9.121544E-22 1.0 9.121544E-22 2K [2K] [2K] [2K] [2K] [2K] [2K] 2 [2K] 2 6 2.6199707E-20 0.75 1.964978E-20 pz p [pz] p [pz] [pz] [pz] [pz] [pz] [pz] z [pz] p 6 3.9451578E-19 0.6 2.3670947E-19 *y [*y] [*y] [*y] [*y] [*y] * 5 1.1771107E-18 0.90909094 1.0701006E-18 .U . [.U] [.U] [.U] [.U] [.U] 5 1.1771107E-18 0.90909094 1.0701006E-18 GU G [GU] [GU] [GU] [GU] [GU] 5 1.1771107E-18 0.90909094 1.0701006E-18 VU V [VU] [VU] [VU] [VU] [VU] 5 1.1771107E-18 0.90909094 1.0701006E-18 () (( [()] [()] [()] [()] [()] 5 2.544199E-18 0.8333333 2.1201657E-18 Gl lG [Gl] [Gl] [Gl] [Gl] [Gl] l 5 6.3307834E-18 0.7692308 4.8698335E-18 lL [lL] [lL] [lL] [lL] [lL] Lll 5 6.3307834E-18 0.7692308 4.8698335E-18 OM [OM] OOO [OM] [OM] [OM] [OM] [OM] MO 5 8.1414365E-17 0.5882353 4.7890806E-17 BL LBBBB [BL] [BL] [BL] [BL] [BL] BBB 5 9.372892E-17 0.5555556 5.2071625E-17 Oz zOO [Oz] [Oz] [Oz] [Oz] [Oz] z [Oz] [Oz] O 5 2.860378E-16 0.5263158 1.5054622E-16 f9 [f9] [f9] [f9] [f9] 4 3.6698536E-16 1.0 3.6698536E-16 ft [ft] [ft] [ft] [ft] 4 3.6698536E-16 1.0 3.6698536E-16 D) [D)] [D)] [D)] [D)] ) 4 8.959601E-16 0.8888889 7.96409E-16 8y [8y] [8y] [8y] [8y] y 4 8.959601E-16 0.8888889 7.96409E-16 JC [JC] [JC] [JC] [JC] C 4 8.959601E-16 0.8888889 7.96409E-16 OB B [OB] [OB] B [OB] O [OB] [OB] [OB] [OB] [OB] BBO 5 1.2053613E-15 0.45454547 5.478915E-16 WD [WD] [WD] [WD] [WD] WW 4 1.857863E-15 0.8 1.4862904E-15 Gt G [Gt] [Gt] [Gt] [Gt] G 4 1.857863E-15 0.8 1.4862904E-15 pB p [pB] [pB] B [pB] Bp [pB] [pB] [pB] [pB] [pB] BBp 5 1.9412468E-15 0.4347826 8.440203E-16 U) U [U)] [U)] [U)] [U)] ) 4 2.1874027E-15 0.8 1.7499221E-15 k) k [k)] [k)] [k)] [k)] ) 4 2.1874027E-15 0.8 1.7499221E-15 dC d [dC] [dC] [dC] [dC] C 4 2.1874027E-15 0.8 1.7499221E-15 kC k [kC] [kC] [kC] [kC] C 4 2.1874027E-15 0.8 1.7499221E-15 NU N [NU] [NU] [NU] [NU] U 4 2.1874027E-15 0.8 1.7499221E-15 MZ MMM [MZ] [MZ] [MZ] [MZ] 4 3.441921E-15 0.72727275 2.5032153E-15 fM [fM] [fM] [fM] [fM] MMM 4 3.441921E-15 0.72727275 2.5032153E-15 #^ ^^# [#^] [#^] [#^] [#^] 4 4.5357996E-15 0.72727275 3.2987635E-15 N. ..N [N.] [N.] [N.] [N.] 4 4.5357996E-15 0.72727275 3.2987635E-15 NG GGN [NG] [NG] [NG] [NG] 4 4.5357996E-15 0.72727275 3.2987635E-15 N^ ^^N [N^] [N^] [N^] [N^] 4 4.5357996E-15 0.72727275 3.2987635E-15 #( ((# [#(] [#(] [#(] [#(] ( 4 8.403125E-15 0.6666667 5.6020836E-15 (N [(N] [(N] [(N] [(N] N((( 4 8.403125E-15 0.6666667 5.6020836E-15 K) KKK [K)] [K)] [K)] [K)] ) 4 8.403125E-15 0.6666667 5.6020836E-15 5T T555 [5T] [5T] [5T] [5T] 4 8.403125E-15 0.6666667 5.6020836E-15 .< . [.<] [.<] [.<] [.<] <<. 4 9.405435E-15 0.6666667 6.27029E-15 ^G [^G] [^G] [^G] [^G] G [^G] ^ 4 9.405435E-15 0.6666667 6.27029E-15 .l l. [.l] [.l] [.l] [.l] ll. 4 1.7424726E-14 0.61538464 1.07229086E-14 KG G [KG] K [KG] [KG] [KG] [KG] K 4 1.7424726E-14 0.61538464 1.07229086E-14 z) zz [z)] zz [z)] [z)] [z)] [z)] 4 2.2962479E-14 0.5714286 1.3121417E-14 zN [zN] [zN] [zN] [zN] z [zN] zzz 4 2.2962479E-14 0.5714286 1.3121417E-14 O# O [O#] [O#] [O#] [O#] OO [O#] OO 4 3.4998443E-14 0.53333336 1.8665836E-14 O) OOOO [O)] O [O)] [O)] [O)] [O)] 4 3.4998443E-14 0.53333336 1.8665836E-14 .z [.z] [.z] [.z] [.z] zz [.z] z [.z] 4 4.7615005E-14 0.53333336 2.5394671E-14 zW [zW] Wzzzz [zW] [zW] [zW] [zW] 4 4.7615005E-14 0.53333336 2.5394671E-14 ^z [^z] [^z] [^z] [^z] zz [^z] z [^z] 4 4.7615005E-14 0.53333336 2.5394671E-14 RM R [RM] [RM] [RM] [RM] M [RM] R [RM] 4 5.5070735E-14 0.53333336 2.937106E-14 BN BB [BN] [BN] [BN] [BN] B [BN] BBBB 4 7.2572794E-14 0.47058824 3.4151904E-14 FL L [FL] F [FL] [FL] [FL] [FL] FFFF 4 7.2572794E-14 0.5 3.6286397E-14 |L [|L] || [|L] | [|L] [|L] [|L] [|L] | 4 7.2572794E-14 0.5 3.6286397E-14 zM [zM] z [zM] z [zM] [zM] [zM] [zM] Mz 4 8.821267E-14 0.5 4.4106334E-14 OR RROOO [OR] O [OR] [OR] [OR] [OR] RO 4 2.293658E-13 0.44444445 1.01940354E-13 KB B [KB] BBBK [KB] [KB] [KB] [KB] B [KB] B 4 2.7879562E-13 0.42105263 1.1738763E-13 BM [BM] BBBB [BM] [BM] [BM] [BM] M [BM] BB 4 2.7879562E-13 0.42105263 1.1738763E-13
Capped to 340 chars:
340: 179
408: 271
Your 340: 142
By my measurement this seems much worse, given that these numbers are the percentual difference from the randomized average.
I believe your measurement may be too exponential. Also, the cycle "|p" is very strong but also somewhat unlikely because it’s almost twice as long as expected. Expected average length per unique symbol in a cycle is 5.39 for the 340 (340/63).
My list with weight 5 (order will differ with different weights), the Ws: value is the final score for each given cycle. 2-symbol cycles (homophone sequences) top 50 (Cpl weight ^5): ----------------- p|p|p|p|p|p|p|p|p|p|p (Ws: 20 C: 21 Cpl: 1 Pd1: 194 O/E: 11/10 Ngs: 180) RlRlRlRlRlRlRlR (Ws: 14 C: 15 Cpl: 1 Pd1: 138 O/E: 7/8 Ngs: 84 ) ||z|z|z|z|z|z|z|z|z (Ws: 13.52 C: 19 Cpl: 0.94 Pd1: 176 O/E: 12/7 Ngs: 128) ^.^.^.^.^.^. (Ws: 11 C: 12 Cpl: 1 Pd1: 111 O/E: 8/4 Ngs: 50 ) ||O|OO|O|O|O|O|O|O|O (Ws: 10.89 C: 20 Cpl: 0.89 Pd1: 185 O/E: 12/8 Ngs: 128) 2K2K2K2K2K2K22K2 (Ws: 10.62 C: 16 Cpl: 0.93 Pd1: 148 O/E: 8/8 Ngs: 84 ) *y*y*y*y*y* (Ws: 10 C: 11 Cpl: 1 Pd1: 101 O/E: 7/4 Ngs: 40 ) ppzppzpzpzpzpzpzzpzp (Ws: 8.04 C: 20 Cpl: 0.84 Pd1: 185 O/E: 11/9 Ngs: 112) GKGKKGKGKGKGK (Ws: 7.76 C: 13 Cpl: 0.91 Pd1: 120 O/E: 8/5 Ngs: 50 ) BOBOBBOBOOBOBOBOBOBBBO (Ws: 7.30 C: 22 Cpl: 0.80 Pd1: 203 O/E: 11/11 Ngs: 128) zOOOzOzOzOzOzzOzOzO (Ws: 7.23 C: 19 Cpl: 0.83 Pd1: 176 O/E: 12/7 Ngs: 98 ) f9f9f9f9 (Ws: 7 C: 8 Cpl: 1 Pd1: 74 O/E: 4/4 Ngs: 18 ) ftftftft (Ws: 7 C: 8 Cpl: 1 Pd1: 74 O/E: 4/4 Ngs: 18 ) ^G^G^G^GG^G^ (Ws: 6.83 C: 12 Cpl: 0.90 Pd1: 111 O/E: 10/2 Ngs: 40 ) (-(-(-((-(-( (Ws: 6.83 C: 12 Cpl: 0.90 Pd1: 111 O/E: 4/8 Ngs: 40 ) ppOpOpOpOOppOpOpOpOOp (Ws: 6.55 C: 21 Cpl: 0.8 Pd1: 194 O/E: 11/10 Ngs: 112) HPHPHPH (Ws: 6 C: 7 Cpl: 1 Pd1: 64 O/E: 3/4 Ngs: 12 ) H1H1H1H (Ws: 6 C: 7 Cpl: 1 Pd1: 64 O/E: 2/5 Ngs: 12 ) 8;8;8;8 (Ws: 6 C: 7 Cpl: 1 Pd1: 64 O/E: 6/1 Ngs: 12 ) f_f_f_f (Ws: 6 C: 7 Cpl: 1 Pd1: 64 O/E: 6/1 Ngs: 12 ) ^k^k^^k^k^k (Ws: 5.90 C: 11 Cpl: 0.9 Pd1: 101 O/E: 6/5 Ngs: 32 ) VVUVUVUVUVU (Ws: 5.90 C: 11 Cpl: 0.9 Pd1: 101 O/E: 5/6 Ngs: 32 ) LULULULLULU (Ws: 5.90 C: 11 Cpl: 0.9 Pd1: 101 O/E: 9/2 Ngs: 32 ) ..U.U.U.U.U (Ws: 5.90 C: 11 Cpl: 0.9 Pd1: 101 O/E: 6/5 Ngs: 32 ) <y<y<<y<y<y (Ws: 5.90 C: 11 Cpl: 0.9 Pd1: 101 O/E: 7/4 Ngs: 32 ) GGUGUGUGUGU (Ws: 5.90 C: 11 Cpl: 0.9 Pd1: 101 O/E: 8/3 Ngs: 32 ) MFFMFMFFMFMFFMFMF (Ws: 5.66 C: 17 Cpl: 0.81 Pd1: 157 O/E: 9/8 Ngs: 72 ) zzFzFzFzFzFzFzFFFFz (Ws: 5.12 C: 19 Cpl: 0.77 Pd1: 176 O/E: 9/10 Ngs: 84 ) 7;7;7; (Ws: 5 C: 6 Cpl: 1 Pd1: 55 O/E: 2/4 Ngs: 8 ) P/P/P/ (Ws: 5 C: 6 Cpl: 1 Pd1: 55 O/E: 4/2 Ngs: 8 ) _/_/_/ (Ws: 5 C: 6 Cpl: 1 Pd1: 55 O/E: 6/0 Ngs: 8 ) ;/;/;/ (Ws: 5 C: 6 Cpl: 1 Pd1: 55 O/E: 5/1 Ngs: 8 ) WDWDWDWDWW (Ws: 4.99 C: 10 Cpl: 0.88 Pd1: 92 O/E: 5/5 Ngs: 24 ) GfGGfGfGfG (Ws: 4.99 C: 10 Cpl: 0.88 Pd1: 92 O/E: 8/2 Ngs: 24 ) GGtGtGtGtG (Ws: 4.99 C: 10 Cpl: 0.88 Pd1: 92 O/E: 6/4 Ngs: 24 ) |L|||L||L|L|L|L| (Ws: 4.91 C: 16 Cpl: 0.8 Pd1: 148 O/E: 12/4 Ngs: 60 ) |G|G||G||G|G||G| (Ws: 4.91 C: 16 Cpl: 0.8 Pd1: 148 O/E: 11/5 Ngs: 60 ) zKzKKzKzzKzzKzKz (Ws: 4.91 C: 16 Cpl: 0.8 Pd1: 148 O/E: 9/7 Ngs: 60 ) zMzzMzzMzMzMzMMz (Ws: 4.91 C: 16 Cpl: 0.8 Pd1: 148 O/E: 12/4 Ngs: 60 ) lLlLlLlLlLLll (Ws: 4.82 C: 13 Cpl: 0.83 Pd1: 120 O/E: 9/4 Ngs: 40 ) lGGlGlGlGlGll (Ws: 4.82 C: 13 Cpl: 0.83 Pd1: 120 O/E: 8/5 Ngs: 40 ) RkRkRkRRkRkRR (Ws: 4.82 C: 13 Cpl: 0.83 Pd1: 120 O/E: 5/8 Ngs: 40 ) ppBpBBpBBppBpBpBpBpBBBp (Ws: 4.47 C: 23 Cpl: 0.72 Pd1: 213 O/E: 10/13 Ngs: 112) RRFRFFRFRFFRFFRFRF (Ws: 4.44 C: 18 Cpl: 0.76 Pd1: 166 O/E: 7/11 Ngs: 72 ) RRMRMRMRMMRMRRM (Ws: 4.19 C: 15 Cpl: 0.78 Pd1: 138 O/E: 10/5 Ngs: 50 ) GzGzGzzGzGzzGzz (Ws: 4.19 C: 15 Cpl: 0.78 Pd1: 138 O/E: 11/4 Ngs: 50 ) ^z^z^z^zzz^zz^z (Ws: 4.19 C: 15 Cpl: 0.78 Pd1: 138 O/E: 11/4 Ngs: 50 ) ^2^2^22^2^2^222 (Ws: 4.19 C: 15 Cpl: 0.78 Pd1: 138 O/E: 10/5 Ngs: 50 ) .z.z.z.zzz.zz.z (Ws: 4.19 C: 15 Cpl: 0.78 Pd1: 138 O/E: 9/6 Ngs: 50 ) JCJCJCJCC (Ws: 4.10 C: 9 Cpl: 0.87 Pd1: 83 O/E: 5/4 Ngs: 18 ) ----------------- Average Ws : 6.55 Average C : 13.22 Average Cpl: 0.88 Average Pd1: 122 Percent O/E: 144 Average Ngs: 52
Interesting! Would you be willing to run your measurement against a bigger batch of ciphers (the ones from my transposition spreadsheet)? Here they are:
http://pastebin.com/raw.php?i=jCW4CGNc
Let me know if you want me to process them into a more convenient format.
Eventually I would like to understand how you compute the measurement so I can implement it too.
Jarlve, the highest azdecrypt score I found in my recent results was 20900. Do you think that is a significant bump from the average for 340-character cipher texts? I think the original 340 scores around 20300.
Hard to say, my advice is to conduct some kind of experiment that tries to answer that question. Yes, 20351.
OK – I did an experiment along those lines. The cipher that scored 20900 has 48 bigrams and 6 trigrams. So I generated 100 random shuffles with the same number of bigrams and trigrams, then ran them through azdecrypt with the same setup (30 restarts and 1000000 iterations). Result:
min: 19990
max: 20815
average: 20369
On the one hand, it makes me think 20900 is significant. On the other hand, if the 1000+ transpositions I tested are all behaving as random text, 20900 is expected to appear as an outlier.
just an idea on partial solves.. as you all have many partial solves built up over time is it possible to put them all together in a file to see if any words or sentences align with each other. Maybe those results could help find the path.
