340 wildcard experiment

[smokie treats]

Wow, I am impressed. I guess I don't have to try to figure out how to solve a cipher like that anymore. Perhaps you will solve it soon.

[Mr lowe]

deep ttalbut isnorn
s the showing leisfi
ed may beconsish i do
nt that will to their
day time ntor but jus
t wear toning urpri s
hould ware arol list
so r the holds picks a
ney cancer reading i
nitely any other edn
eve sometime t anymo
re atention my later
in and histore asawe
tors a formal all mur
der i immage plecons
tallies so tally dor
e and singeartnetis
going in a bond nssit
trunnehmedtoeinga
pyid stuni cow i thaf

Lots n lots of words and mini sentences.. Especially in the top two thirds .. How do you judge a good partial "solve"
Not sure if it's just garble or needs further investigation..

Not sure if breaking it down helps.. If not let me know.
Cheers

Multiplicity: 0.3117647 Characters: 340 Symbols: 106 Bigrams: 18

[Jarlve]

Seventh iteration.

Code: Select all

340:

AZdecrypt 0.993 (Reddit corpus 6-grams)

Manipulation operation(s): LengthSymbol(2,1,1,[l])

Score: 18964.26 Ioc: 0.06583377 Entropy: 4.08104 Chi-square: 18.84333 Characters: 340 Letters: 23

keepthalbutisnorn
stheshowongleitri
edmaybeconsishido
ntthatwilltotheir
daytimentorbutjus
twearsoningurpros
houldwarearollist
sortheholdsgivesa
neycancerreadinfi
nitelyandotheredn
evesometogetanymo
reasectionmylater
inandhistoreasawe
sorsaformalallmur
deriimmageplecons
talliessotallydor
eandsingepronetis
thinginabonanssit
trunnehmedgoeinga
pyoustunicowithaf

Multiplicity: 0.3294118 Characters: 340 Symbols: 112 Bigrams: 17

ˆER>p‹^VPk|1LTG2d
Np!B(#O%DWY.<*s…)
By:cM"UZGW()L#z‰}
Spp7^Œ8*V3pO$'Ru2
_9M,ztjd|5FP0&€k/
p8R^FO-*dCkF>2D(
#5=vq%;2UcXGV.zL|
(G2~†j#O?_NYzI@L9
d<MQb[ZR2FBcyA6w
-zŽUV\^]Op7<FBy-
U`R/5tE|DYBpbTMxO
2<cRJ|*5T‚M.a&BF
z69Sy#eN|5FBc(;8R
GFN^‡52ƒb.cV„tgh
yBX1*:49CE>VUZ5-i
|c.3zB{(Op^.fMqG2
RcTmL16C<nFlWB|)L
or)WCzWcPOSŠT/()p
|FkdW<7tB_YOB*-Cc
>MDHNpkSzZO8A|K;+


[Mr lowe]

keep thal but is norn
s the show on gle i tri
ed may becons ishi do
nt that will to their
day time ntor but jus
t wear son ingurpro s
hould ware a roll ist
sorthe holds gives a
ney cancer f read in fi
nitely and other edn
eve some to get any mo
re a section my later
in and histore as a we
sorsa formal all mur
der iimmage plecons
tallies so tally dor
e and sing epronetis
thing in a bonanssit
trunnehmed goeing a
p you stun ico with af

Multiplicity: 0.3294118 Characters: 340 Symbols: 112 Bigrams: 17

[doranchak]

Lots n lots of words and mini sentences.. Especially in the top two thirds .. How do you judge a good partial "solve"
Not sure if it's just garble or needs further investigation..

It's hard to answer that. In most cases, you can just tell when a partial solve is right via visual inspection, because the sentences and phrases require very few edits to finalize. They will have more "high order structure" to them, such as properly applied grammatical rules. Most solvers maximize the resemblance of "low order structure" (ngrams and words) to real language. So recognizable words and small phrases will often appear in a false solution. But if someone ran the false solution through a grammatical parser or some other measurement of high order structure, it would score very low.

Solvers can try to correct for this by supporting larger ngram statistics, as Jarlve has done by generating 6-gram and 7-gram statistics. But it gets challenging as you get higher, because of the storage and memory requirements imposed by such large sets of data. But, I think there's another tradeoff with larger ngrams, since it may be harder to hillclimb with such large "building blocks". For example, on your way to reaching a valid, large ngram (such as a 7- or 8-gram), you will encounter valid, small ngrams (such as a 4- or 5-gram). It may help the solver to look for the small ngrams (smaller structure) while looking for the large ones (larger structure), so the "hill" it's climbing might be a bit smoother.

[Mr lowe]

thanks Doranchak.. I'll keep breaking them down. Looking for a needle in a haystack..trying to find a word or phrase that Z would use, something unique to him. Also visually trying to see if a half spelt word on a particular line will be finished off on a different line (misalignment) or that it reads backwards or spiral.
Cheers

[Jarlve]

Eighth and final iteration.

Code: Select all

340:

AZdecrypt 0.993 (Reddit corpus 6-grams)

Manipulation operation(s): LengthSymbol(2,1,1,[<])

Score: 19016.81 Ioc: 0.06412458 Entropy: 4.112194 Chi-square: 17.44045 Characters: 340 Letters: 23

keepthalbutisnorn
stheshowonglyitri
edmaybeconsishido
ntthatwilltotheir
daytimentorbutjus
twearsoningurpros
houldwerearollist
sortheholdsgivesa
ndycancerreadinfi
nitelyandotheredn
ecesometogetanyfo
rmakestionmylater
inandhistoreasewe
forsaformalallmur
deriimmageplecons
talliessotallydor
eandsingforanetis
thinginabonanssit
trunnahmedgoeinga
pyoustunicowithev

Multiplicity: 0.3441176 Characters: 340 Symbols: 117 Bigrams: 17

ˆER>p‹^VPk|1LTG2d
Np!B(#O%DWY.‘*s…)
By:cM"UZGW()L#z‰}
Spp7^Œ8*V3pO$'Ru2
_9M,ztjd|5FP0&€k/
p8R^FO-*dCkF>2D(
#5=vq%;2UcXGV.zL|
(G2~†j#O?_NYzI@L9
d’MQb[ZR2FBcyA6w
-zŽUV\^]Op7“FBy-
U`R/5tE|DYBpbTMxO
2”cRJ|*5T‚M.a&BF
z69Sy#eN|5FBc(;8R
GFN^‡52ƒb.cV„tgh
yBX1*:49CE>VUZ5-i
|c.3zB{(Op^.fMqG2
RcTmL16C•nFlWB|)L
or)WCzWcPOSŠT/()p
|FkdW<7tB_YOB*-Cc
>MDHNpkSzZO8A|K;+


[Mr lowe]

keep thal but is norn
s the show on gly i tri
ed may becons ish i do
nt that will to their
day time ntor but jus
t wear soning urpro s
hould were a roll ist
sor the holds give sa
ndy cancer read infi
nitely and other edn
ece some to get any fo
r makes tion my later
in and histore asewe
for sa form al all mur
der iimmage plecons
tallie s sotally dor
e and sing for anetis
thing in a bonanssit
trunnahmed goeing a
p you stun ico with ev

[Jarlve]

Smokie33 vs 340 started. Same settings as previous test except that now 2 symbols are expanded per iteration.

In the meanwhile I added something to AZdecrypt that detects wildcards. It seems to work like a charm. Here is a solve for smokie's cipher in under 3 minutes with the new intelligence. I have yet to try it seriously on the 340 or smokie33.

Code: Select all

AZdecrypt 0.993 (Reddit corpus 6-grams)

Manipulation operation(s): LengthSymbol(2,4,4,2,[E,-,8,Q])

Score: 22879.79 Ioc: 0.05777373 Entropy: 4.229834 Chi-square: 32.57436 Characters: 340 Letters: 23

purplehaveallinmy
brainlatelythings
dontseemthesameac
tinfunnybutidonok
nowwhycausemewhil
eikisssheckyourpl
ehaveallamousddon
tknowifitcomingup
ordownasihappyori
nsiderswhateverit
istsatgirlputaspe
llancehesprehespr
eohnoohnoahpurple
haveallinthesecdo
ntknowifitsdayorn
ightyouvegotheblo
wingblowisgoodind
isitforosrowoojus
ttheendoftimehelp
meyeahpurplshavem

Multiplicity: 0.3382353 Characters: 340 Symbols: 115 Bigrams: 5/1821 Sequential: 6978/5136

041D,%(!9=:AN).r
"1J@‹,T3KA‚G?MO'2
`/X3FU\sGL_R!t%:#
3W.&HBOƒ"4G]<CX[+
.a6IV„Z;H2=uK6()N
U@+MFRS?\Z+…bŒ1c,
_LJ9%TN!dYHS$<^O
3eX/IWf]v#Cwg.>40
P1$Y6B:x@VJD0†^1M
OhWŽ=1‡IiTGK5U1]3
)FGS!3'@1,DHG:R0\
AXy_L%SDz=VKS0{
UC(.PY?ˆ[JLD4‘0N_
VT9j!,AMB|k=‰KZ<l
O’+X[6W&]G2$:ŠC1.
)>(38PH5m'YG}\"N^
I@B>",/6MS'~nWX<
]F)3“CQPS1YI^[*4R
3G?_%.$/&3@€oLKAD
-Up\JV0HqDNS(TE%7


[doranchak]

In the meanwhile I added something to AZdecrypt that detects wildcards. It seems to work like a charm. Here is a solve for smokie's cipher in under 3 minutes with the new intelligence.

That's very impressive! Can you elaborate on how you perform the detection?