"Adam P. Goucher" <apgoucher@gmx.com> wroe:

Henry Baker <hbaker1@pipeline.com> wrote:

...

Why not use GIMPS as proof-of-work for a "coin" that is only minted ~1X per year?

The main reason is that proof-of-work solutions should be hard to find and easy to verify.

Mersenne primes are hard to verify.

I sympathize with the desire that the proof-of-work do something mathematically interesting. I propose finding dual palindromes. Preferably binary-ternary, as those seem to be the rarest. It's difficult to discover that 24014998963383302600955162866787153652444049 is a palindrome in both binary and (standard) ternary, but easy to confirm that it is one once someone makes the claim. Before anyone claims that that's not interesting, look up the tasks that Bitcoin mining solves. They're *designed* to be as ugly, boring, random, and unsymmetrical as possible, lest someone find an elegant shortcut and corner the market.