The problem with one time pads is that they are a bit hard to produce and distribute. One of the hardest parts of the process is to make a pseudo random stream of 0's and 1's. I've had an idea about how to do this.Why not simulate a homogonizing device, such as say,a food mixer. Start off with a 3d cell model of a circular container, much as you would with a CFD model. Lets say take a million cells. Lets set eachj of the cells either on or off. The initial method is relevant but lets'forget that for a moment. Next we introduce the 'paddles of the mixer. These paddles operate on the bits in the bowl in the same way as they would in a real mixer: we'll simulate viscosity and a distance/velocity algorithm to affect each of the cells during each iteration of the mixer. The 'whicking head' can be moved around by the user to form a unique patter or the heads could be induced into 'chaotic' motion. We'll have some kind of distribution check to show when the process is complete. I've a clever plan for this based on voume shading, but ignore than for now. Once the mix is complete, you have (hopefully) a stream of bits which is as random as you could possibly hope for. these can be read out in any order and used for the pads. A sensible precaution would be a trim to remove data from the 'edges' as this may be less likely to move.