It's a really clever idea with technical merit. More practical than the old lava lamp CRNG. It's an entropy source, and I think there are an infinite number of

It's a really clever idea with technical merit. More practical than the old lava lamp CRNG. It's an entropy source, and I think there are an infinite number of possible variations on what I'm calling "entropy rituals," which may or may not generate a sufficiently large field that mitigates brute force attacks. In its favour, good, exogenous entropy mitigates part of post-hoc decryption attacks and wrecks the economics of passive bulk interception and surveillance. It means that threat actor needs to a) use direct attacks on user endpoints, b) sabotage algorithm implementations to truncate keys or c) bias the entropy source with loaded dice (because everybody knows...) or longshot d) use a TAO-like operation to add dice with duplicate facets from other cubes to dice kits in transit to reduce the total field size, which would be easily discoverable, but chalked up to a "printing error."