[Cryptography] Back to the initial generation of public keys

[Phillip Hallam-Baker]

So I have a refinement of the 'jelly-bean jar' random number generator
I proposed a while back. (Having eaten all the jelly beans).

The new setup has an 8x8 frame and a set of 256  six sided dies. These
are painted white on three sides and black on three sides.

As before the frame is shaken up and the resulting pattern recorded.
However this time the pattern is formed in a single layer using only
the coarse information of whether the die is white face or black face
up. And there are two separate public key generators, each of which
has a separate camera.

For the system to be accepted, the two generators must generate the
same key for every roll of the dies.


The device is operated in two modes, first with the end cap removed so
that the positions of the dies is visible and for 'production' runs
the endcap is closed.


The advantage of this approach is that every aspect of the operation
of the device is auditable. Especially if the code for the two
generators comes from separate sources.

To effect secure readout, the use of the type of printer used for bank
PIN codes could be used. As a further protection, the keys are split
into two with each generator printing a separate half. So if the
halves don't match the keys won't checksum correctly and will be
rejected when there is an attempt to use them.