[Cryptography] On improving the performance of the classical Playfair cipher

[mok-kong shen]

Am 08.04.2016 um 00:08 schrieb Tom Mitchell:
>
>
> On Wed, Apr 6, 2016 at 1:55 PM, mok-kong shen <mok-kong.shen at t-online.de
> <mailto:mok-kong.shen at t-online.de>> wrote:
>
>
>     The classical Playfair cipher and its variant "two square cipher"
>     (employing two Playfair matrices, used in WWII) are obviously too weak
>     to be practically used today.
>
> ...
>
>     Cities" from Project Gutenberg and took out the alphabetical characters
>
>
> The "obviously too weak" bit might be a bit harsh.
> It is a shared key system so would be fine for two person
> simple messaging where the traffic is limited and statistics
> are also limited by a small sample size.
> Improvements should be paper, terminal and JacaScript friendly.
>
> The alphabet limitations would bother some users (me).  Extended
> eight bit ASCII seems a necessary feature today but not to the
> point that smart VT100 type attacks and html escape problems
> are possible.
>
> The weakness should keep the big boys from bothering you
> even if bolted into an App or js.browser plugin.
>
> All systems have key management and usability issues.
> As a "scramble()" block of code to build a better looking mouse
> trap around it should be fine.
> As for attackers,  mice sure but not lions, tigers and bears.

I don't yet fully understand your points. The variant is intended
for manual use, though I had to test it with computer. 6 times
the work in comparison to the original Playfair isn't too much IMHO.
Certianly everything of utility in life has a cost (Principle of
No Free Lunch is ubiquitous). According to the test statistics, the
ciphertext is sufficiently random, so attacks should be correspondingly
hard.

M. K. Shen