VMs: Mario and Cesar

[Jacques Guy <jguy@xxxxxxxxxxxxxxxx>]

... I can't find it. That message from... who was it?
the message about the Papal cipher, a 5x5 grid, and 
the key: MARIO.

Anyway, I implemented it, and applied it to my "Dear
Mom and Dad" plaintext. However, some explanations
first.

The cipher text has only 10 "glyphs" (0, 1, 2, 3, 4,
5, 6, 7, 8, and 9). So I had to find a way to turn
it into 20, to make it more Voynich-like. An obvious
way is to assign two letters to each digit, e.g.
0 => (a,b), toss a coin, and write 'a' if falls tails
turn up, 'b' if heads up. But that would increase the
entropy, and make it less Voynich-like. So I decided
to rely on the previous character: if odd, take the
first alternative, if even the second.

The alphabet of the plaintext consisting of 31 letters
I reduced it to 25 by using digraphs-- to fit the 
5x5 grid of course. 

And I enciphered it. Here is the beginning (I called
it MARIO.TXT):

angngoeobsamckaobtbqaoengnfqakbtfqak
cmakmcmakntgmdsbsocnalpgoalphrfpfrhrfp

Oh, no, I didn't use "mario" as the key, so don't
waste your time trying to crack it with that crib.

Now I needed a test case. 

So I took the same plaintext, and reduced it to
20 different letters, again using digraphs.
And I encoded it using a simple-substitution
cipher (unnecessary, but if I hadn't you would
have no trouble figuring out the language, which
I want to keep my little secret for the time 
being). Here is the beginning (I called it
CESAR.TXT):

upwppeflrwfrcpkxwkxjxejxurlgpwt
kxwkxjxejxflxeplgsrzppcugpetfpp

Then I computed the entropy:

     CESAR    MARIO
h0   4.32193  4.32193
h1   3.93827  4.16993
h2   2.78085  2.28272
h3   2.20274  1.78937

I was surprised at the low values of h2 and h3
for MARIO. So I enciphered it again, this time
keeping it to a 10-letter alphabet. Here is
the beginning (MARIO10.TXT):

aibijejbiahcfajbjbgajeibiagfbjagfchafhcha
fijbhdibijciagfbjagfchafhchafcfafhbihdief

and the entropy:

h0 3.32193
h1 3.16959
h2 2.34368
h3 1.83350

Just what I had suspected: my trick for expanded the
alphabet from 10 to 20 letters would reduce the 
entropy.

Still, I must say I am surprised. My even/odd trick
might go a long way towards explaining away those
pervading alternances between Voynich letters,
<i> and <c> for instance. Note that you can write
directly in that cipher almost as fluently as in any
made-up alphabet.

Finally, note that is is awfully easy to crack.
Here, I enciphered "hellohellohellohello" into
20-letter "Mario":

brfthqcltgmaoclqobrfthqcltgmaoclqo
        
It would be even easier to crack without the 
"10-to-20" letters trick.







______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list