Zodiackillersite.com

I came up with the idea that a simple algorithim (using plaintext) might be able to mimic homophonic substitution. I think I have come up with a rough example but I'm not really sure how good it is.

What do you think?
Would you believe this is homophonic substitution or not, why?
Can we fashion better approximations?
Would any of these be able to provide a better fit for some of the observed 340 tendencies? (diagonal repetitions, clustering)

Code: Select all

8DMX]h1=IR`g7<K[g
[]bef;NS\09L_n;PS
JPe3<PYl9HZ_e:H\d
Tbm6BNWelCLX\cd16
n<PX]c2DV[nBDILMb
di67ENa5=BO^1EIJX
NSe4I\]k4ABN]cd0<
?NYbn:M\inBJSah/8
=BUbg;CHUd7K_g9BN
IR`glDTYk0>AFOcl?
3IN\^c7KPb6>?MTYm
09GNg6K]/>AL_n4:Q
I]efm6HTh057<Oc34
5IX^g;DWk34H_gl:C
6?DMdm9EGL^ce4FT]
Zjk=>BKNSTbfg3?S[
OX`a7<GP\hm1HQ]ik
_b1>CPi<HI_d7@W`l
g5DX_h>C\k@Mf45BG
8=@AVinGVkBKWc7Ib


I've noticed some difference in IoC between the top and bottom 10 rows (untrue for the 340), so that might be a weakness as it currently stands.

Actually it is the 340 which stands out for not having much difference in IoC between top and bottom parts. I do not have a cipher in my possesion which difference is so low.

The cipher provided earlier are the first 340 characters of the 408 and the algorithm itself is as follows:

P = Position number in the plaintext.
A = Alphabet number of letter P.
T = Total of A.
S = Number of symbols.

Then per plaintext character do the following:

T = T + A
Symbol number = (T + (P MOD 17) MOD S)

If you don't know what MOD means, it is the Modulo function which is a binary operator like "addition, multiplication, etc" and you can find it on most scientific calculators including the Windows one. Basicly it returns the integer rest value of division. For instance 6 MOD 5 = 1, 11 MOD 5 = 1. The number 17 I came up with myself and seems to work fine.

you do fascinating work. i don't have much to contribute to your efforts but i do read all of your posts. very interesting stuff.

Jarlve wrote:Actually it is the 340 which stands out for not having much difference in IoC between top and bottom parts. I do not have a cipher in my possesion which difference is so low.

I was curious about this so I ran my own test:

Z340 top half IoC: 0.01802993386703794
Z340 bottom half IoC: 0.018169161155586495
Difference: 1.392272885485553E-4

Z408 top half IoC: 0.017386264850767892
Z408 bottom half IoC: 0.017434560030908916
Difference: 4.829518014102391E-5

So the difference is even smaller for Z408. Do your calculations match mine? If not, how are you computing IoC for top and bottom parts?

Our numbers match. Back then I think I compared the top and bottom halves of the 408 as a 20 row ciphers. I'm not sure if it is anything, it may be another curious quality of the Zodiac ciphers.

Jarlve wrote:Our numbers match. Back then I think I compared the top and bottom halves of the 408 as a 20 row ciphers. I'm not sure if it is anything, it may be another curious quality of the Zodiac ciphers.

On average, wouldn't the IoC of both halves be expected to match, since the symbols would be well distributed throughout the entire cipher text? I was thinking of doing a shuffle test but it seems reasonable to assume an IoC spread would generally be unexpected.

doranchak wrote:

Jarlve wrote:Our numbers match. Back then I think I compared the top and bottom halves of the 408 as a 20 row ciphers. I'm not sure if it is anything, it may be another curious quality of the Zodiac ciphers.

On average, wouldn't the IoC of both halves be expected to match, since the symbols would be well distributed throughout the entire cipher text? I was thinking of doing a shuffle test but it seems reasonable to assume an IoC spread would generally be unexpected.

Yes, I suppose there's nothing more to it.