Interesting. A while back I wrote Crypt::Solitaire which is another implementation. I'm interested to run our versions side-by-side.

Btw, Paul Crowley did some nice work in pointing out a random number bias in Solitaire. He attempted to write a new card cipher (Mirdek) that wasn't encumbered by the limitations in Solitaire, but by his own admission, his version suffers from weaknesses as well. Here's the URL of his stuff if you're interested: http://www.cluefactory.org.uk/paul/crypto/mirdek/