Tim sends us, "A way of encoding binary numbers into playing cards that I thought up. It usually allows many more bits than there are cards. The method can also store binary encoded letters of the English alphabet at less than 2 cards per letter on average, and has a theoretical ability to do less than 1 card per letter."
Tim isn't sure if his method of data-compression is novel or not, and neither am I. If you know of related work, please add it in the comments.
The method treats cards as representing a 1 or 0. Its ability to store more data than just 52 bits comes from the way that cards which can have their position deduced by examining the rest of the pack can be taken out and reused to encode more data.
The data in the cards can also be encrypted to the level of a one-time pad.
I don't know if the method is any use outside of being an interesting mathematical puzzle. It's fairly simple, but I haven't heard of the method anywhere else so I'd be interested to know if I'm the first person to think of it. If not I'd love to know who else has thought of it.