Re: Which paper and pencil cipher to use ?

From: Bob Jenkins (
Date: 01/25/03

From: (Bob Jenkins)
Date: 25 Jan 2003 14:47:07 -0800

Paul Rubin <> wrote in message news:<>...
> (Simon Johnson) writes:
> > > 1) do RC4 with pencil and paper;
> >
> > Standard RC4 with its 256 8-bit words is unmanagable on paper.
> How do you know that? Have you tried? I might give it a try.
> 99-element RC4 should be pretty easy to do (divide the paper into a 9x11
> grid, write the S-box in the grid with the pencil, use two coins to
> track the i and j pointers, and use an eraser to swap things around).
> You can do 52-element RC4 with a deck of cards spread face up in front
> of you. Somebody actually tried this on one of the previous times
> this question came up in the ng. He said with a little practice he
> could do a few characters a minute, iirc. The idea of doing
> 52-element RC4 with a deck of cards was the inspiration for Solitaire.
> > If we use a much reduced internal state with say 16 4-bit words then
> > it has been shown to have problematic biases.
> Yes, 16 words is nowhere near enough. 52 may not be enough either.

I've got code on the web (,
gather.c) that can break very very long strings of 16-element RC4 (< 1
second) and 32-element RC4 (4 hours). I suspect breaking 64-element
RC4 is achievable, the techniques I already know of aren't maxed out,
but I'm not currently pursuing it.