We consider the problem of how to print a human-readable message, or
image, on a piece of paper, while simultaneously preventing the
participating printing agents or devices from learning its
contents. We examine the problem in two settings: with a trusted
dealer who knows the message, and in a distributed scenario that
allows two non-colluding printers to obliviously generate the secret
message and print it without ever learning it. We present a basic
protocol for printing arbitrary-length messages in the trusted dealer
model, as well as protocols for printing different types of messages
in the distributed model: a randomly selected element from a set, a
random permutation of elements in a set, and an optimization for
printing alphanumeric characters using 16-segment display logic.