[Cryptography] The ultimate random source

[Jerry Leichter]

On Feb 16, 2014, at 10:40 PM, Bill Stewart <bill.stewart at pobox.com> wrote:
>> >> I actually started with rather more elaborate steampunk devices that had dice with different colored faces. These would be machine read somehow.
> 
> For entropy generation, dice have the limitation that physics is going to typically force them into alignment, at least with the flat surface on the bottom and possibly horizontally with each other, as opposed to the much more random positioning of the candies which a camera can take advantage of.
> 
> Also, Phill's method means you get to eat the candy when you're done, at least if they're chocolate, though it's somewhat counterbalanced by the failure mode that somebody eats the candy before you've got all the random numbers you need.
If you really want to go the route of physical "randomness", get a bunch of glitter, put it in a fluid, and use something to stir it.  Put a strong light on it from above, point a camera, and record only the timing and positions of the flashes of light when the light source is reflected into the lens.

You can certainly get chaotic behavior by correct choices of the parameters of the moving fluid.  One would have to do the math to be sure, but I suspect that there is an element of true randomness from Brownian motion.  (If there isn't with standard glitter, you could use smaller reflective particles and a microscope.)

An interesting (and probably very difficult) question is how many camera angles you can use simultaneously without there being too much correlation between them.

Not that I feel the need to go and do the computations, because I also don't feel the need to actually build such a thing.  But if you're driven to it - by all means!  Beyond providing random numbers, this could be really pretty to watch.
                                                        -- Jerry


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4813 bytes
Desc: not available
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20140217/7a72b567/attachment.bin>