Not sure what this means yet, but I tried some "olson" schemes and found a few that produce 43 bigram repeats, right at the spot Dan was talking about (row 10):
https://docs.google.com/spreadsheets/d/ … sp=sharing
An example of a procedure that produces 43 bigram repeats:
1) Split the 340 into two halves (rows 1-10, and rows 11-20).
2) Put them side by side, then rewrite the cipher into the same sized grid (17 cols 20 rows)
3) Rotate it 180 degrees
4) Rewrite it to column width 35
5) Rotate the result 90 degrees
6) Start at upper left corner, write out the cipher via diagonals
With so many permutations, it’s hard to say which sets of operations are isomorphic (equivalent) to others. For example, an Olson split might be occurring as a side effect of some other set of operations (or vice versa). Still, I think this is an interesting avenue to explore.
Here are the results of your permutations: https://drive.google.com/open?id=0B5r0r … WxXc2dveVE
Update: I noticed some of these ciphers have 64 symbols? My error?
Not sure what this means yet, but I tried some "olson" schemes and found a few that produce 43 bigram repeats, right at the spot Dan was talking about (row 10):
I’ll work on running these through my solver.
An example of a procedure that produces 43 bigram repeats:
Interesting. I’ll take a look at it tomorrow.
Update: I noticed some of these ciphers have 64 symbols? My error?
In the spreadsheet, an equals sign is automatically added when pasting strings that begin with + symbols. Otherwise, Google Sheets will interpret the incoming text as a formula. I didn’t think of a quick way to fix/prevent that, so I left it as is.
Here are the results of your permutations: https://drive.google.com/open?id=0B5r0r … WxXc2dveVE
OK, here’s the extract of every word my word search found among the candidate plaintexts:
http://pastebin.com/raw.php?i=cWTNuf5s
"it’s a bomb"
ketcandiealfour is
t in the air in s town b
rutal llasiconsbel
ontalamendmeiecna
thesthmccadescrea
tsareshinderbroat
hbbitoranesmondha
llodatseastattlee
aterbat its a bomb ir
an tech esubstalane
fiedacsreardrreer
issuitfroidessned
itripatheisonthcl
oseiarreselectron
sidnotprimeanwast
heaidofartoreofft
itstorsuintsbehei
nonasinotiethstha
tlemfetriechtunss
inschanforthassul
lunatedcounsensup
totheonetamanyous
esandaymssodorthe
roaburearttendile
tupsheersthanyoul
ldprasandsandoorm
siraybotsnorineou
tdappaldintortheh
eralalltthedatdea
rofaspopesteacomm
snualforprrsansit
eeillussedelestou
rtocaspaunotionme
nurandthmostleari
sthenthlyan a plan s
et in the police rndc
hiaitepainortisan
dcreascreassocyhu
mensmalcandeflito
usearnneieratiote
poadderssensonded
tortwascsingemasi
thisly did this wils
o held in ut again ere
useonlasanontarof
thes the long boy and
eediocetonh smith n
icmultancedfroemb
arrest in la ruesith
asbontastheatfilt
of lassie said rient
heedaneofulaginat
etsthatccoupied fr
om the thin days tsfr
ryour fiddle o and d w
hite emranomingara
wacrewsshayoo pull
s paul at eytoldinfr
ienetorerournwiso
nastacinguidthama
tooem sold charity a
theseredleaclemic
ancialsonedevebee
nry hero it for three
neills tap hope trir
toalonlastilantoo
nnarterracantloon
ehereaeminctlyhou
ndfmcadesa diane ma
rains ves heating ae
msitieshpledccast
aroostsregronbrat
ioistofthefper mur
der mccessedalitha
t al smith old a ben hi
d hard clssontonpol
itissoinotheyment
leathedidnysainma
ssscanelasseerthe
soillonsthiteffte
split the 340 into two halves. then test in two separate spirals both directions then opposite corners?
split the 340 into two halves. then test in two separate spirals both directions then opposite corners?
Already been done in one of my tests some while ago, double spirals. We are exploring schemes that relate to the bigram repeat peak @ period 15 (340 mirrored) and 19. 
Here are the results of your other permutations: https://drive.google.com/open?id=0B5r0r … F9OYTlBVzQ
If you have anything more don’t hesitate. Could do a few 10.000’s of these every day.
While it’s possible that certain readable fragments could appear I’m more interested in the score of some of these strings. And wonder, are these higher because we’re getting closer to the original scheme (typically so)? And what possibly could stand in the way of not getting better solves. You have to distort a message quite a bit to make it score like it was a bunch of random characters. A straightforward Olson scheme is certainly not enough.
Because of the high bigram count at period 29 it seems very plausible that the 340 had allot of bigrams at period 2 in the original message (just like the 408). Then it seems likely that correctly undoing the transposition would also results in higher than average period 2 bigram count.
Another question, if you take the smokie9 and 13 and unleash your permutation algorithm on it, would we find the solutions? If no, then a simple misalignment in some scheme could prevent us from getting it with the 340 also.
smokie9: +#(%5-DBW127!BUA; 3[LK,!P+'BU>&HFY3 D).EY!<$:@03'?2QI W3NTG_'3KR43)4ZJA .WUB[]&=;4G)^%P4T 7"!NJ9']Q+3ZWA2#3 *;K3$=>XYBU4,:^J 43#TP4->M5)2.7QTE %G%0Z(A>!D:.+BXSI 3),E34K^"TDWO33J Y03YG%>2I+]3#U4Q) 1U)ZJP](/KL)+BC9 6V_^W?BJW^8TZBKU$ 3X)%!BD>45WQT;O'B 7!63@*#-UAPGB,'F' !RU4:EK)>LC^X0V1Y >Y3ZQP(GGW"&-F33. %BMU32BHB<4U>A!UJ @K6+Q^BI=4T3D-!,% ;><P_]Q9.57UW33O smokie13: /6.(!H$0%5&Y73M)@ 2N]5?U.#DI>Z/<J- 4#LT!I^Z96%H[NA2Y /@W_)3]4$&U.CM'6 <N^HY3%@OUVQ,D^.W !JGC9]4'HY3I%VAZ@ Z.HU^^>/;&#>@AJW! Y:6)#<L7/?ZUS^1.% >[N_V]@+W.HUO$IZ7 ^/(%C5_#>)5F!@A A)QJ)IU4Y#I_T6NW2 L]]8SX]&M%'/-N]<% D=@TV(IZ%W4!3F.J5 V?]^>FU@#[/5CA>UM 2IUOLTB5DTE&>%@U M^%9N]!XI7HF<4/J ]I5KVYOT)NFP=UUAC G!S&]7^D"]Z%@^MAR 4^J[TNV%H-*@./?2W &0U](^AT4%I&!H@G
Thanks for running these. Just for reference, what would a message of length 340 score in the solver? How far off are we?
I am making some changes to my "permutations explorer" that will probably take me a while to finish. When I’m done I’ll see what it turns up for smokie9 and smokie13.
Sounds great doranchak!
My solver normalizes by length. With the Practical Cryptography n-grams the 408 solves at 23274. I’d say the solving range starts around 21000, and those would be low quality, partial solves.
An example of this (I recognized it as being p2 of my plaintext library) and it only scores 20770 but I’m guessing the higher order n-grams could create a larger difference.
177 Score: 20770 Ioc: 668 M: 195 C: 323 S: 63 mplyremarkahingso hasnoldreadbleint nkistomelymicethi onthewaytamuchout rabbitsaytosalthe ohdearofdearsspdn dbetoodatedhishal thoughtssomeeneme rwaresitocculante amerthastheorredt omamedonepleughts isbutitthetreatth ldeepmeequitmessa lidbutwhenthenasu stactuallysoprabb schousonitewakidi aitlocketindirstc easitinethenlooke rthelpdisnotmurri Original message: therewasnothingso veryremarkableint hatnordidalicethi nkitsoverymuchout ofthewaytohearthe rabbitsaytoitself ohdearohdearishal lbetoolatewhenshe thoughtitoverafte rwardsitoccurredt oherthatsheoughtt ohavewonderedatth isbutatthetimeita llseemedquitenatu ralbutwhentherabb itactuallytookawa tchoutofitswaistc oatpocketandlooke datitandthenhurri edonalicestartedt 7Q9E3,J1*J()B'2. ?T[OPQ$M<!A"N4)-5 O*)[R.7%Q9,@;%5?) P&GL3I!9G^V6#LCHG 1!"")G2T9GPW!QGL% Y?$KT1Pa$KTE[ZA& A"4_YCA!SU8?@2(J+ GLC6'(_W[.73KOFV= EI!1<Z@GC#;HQJ&_K ^V31_(T[R(%.113$S .V!748Y-<Q4H>?5[ @2"H5:55L3G]<TS5( NAFK,4<06)_,=W[! Q:A"6SI(4OSLUXT[H W_J;G6T+N9[CE!"" [;?CH[.&@SFI^*:8: ^:S/C;*=_:B$:]Z_# <T[)5:B<5(=-+YP*U E_(KQ8:ZXYGV6E1@
I’m now running through 1000 randomizations of the 340 and will get back with the results, at the moment the average score for a randomization is 19875 and the highest is 20416. So in a way the n-grams pick up on randomization.
Here are the results from 1000 randomizations: https://drive.google.com/open?id=0B5r0r … HphVHBiVUU
Highest: 20452, lowest: 19314, average: 19864. The base 340 scores 20351 here so it does fall in this range.
Here’s a table of n-gram scores for the 340. For each variation 1000 random restarts and 10.000.000 iterations per were considered so it is very thorough.
Average of the following: (Practical Cryptography 5-grams, Reddit 5-grams, Usenet 6-grams, Reddit 6-grams, Usenet 7-grams, Reddit 7-grams)
340 normal: 17513: (20351,20046,17702,17595,14877,14507)
340 mirrored: 17548: (20231,20218,17503,17696,14972,14671)
340 flipped: 17648: (20330,20208,17688,17874,15005,14784)
340 reversed: 17429: (20219,20007,17473,17520,14883,14475)
I have the feeling the code works with the system you both describe. It is nice to see people work together with their individual strenght in coding.
Thanks for running these. Just for reference, what would a message of length 340 score in the solver? How far off are we?
My question for both of you…: What kind of message/kind of form of info do you expect to find?
No idea. But if I’m forced to guess, I’d bet more of the same rambling stuff he liked to fill his letters and the 408 with. I don’t expect there will be anything substantial or groundbreaking in the message. If there even is a message.
