So, first off:
A really good way to not get sucked into the paradigms of a given tool set is to start with requirements. Rather than retype the work of others, I would suggest you take at look, FOR EXAMPLE ( just like BOINC is an EXAMPLE, you dingbat ) at this refined set of proposed requirements posted elsewhere on this forum:
https://bitcointalk.org/index.php?topic=54222.msg647423#msg647423I looked at this proposal, and it's still missing the secret sauce. It has the idea of using "trustworthiness" on the network to determine nodes' weights in getting to sign off on a transaction having occurred or not occurred, but doesn't explain how "trustworthiness" is determined.
Now, certainly, if there was some way to establish such a thing, algorithmically, in a way that prevents both sockpuppet shenanigans and the long (but big) con, then you'd be done! You wouldn't need proof-of-work anymore! The problem would be solved!
But... that's a hard problem. That's one of the
fundamental hard problems in this field, from what I understand. I've not yet heard of any algorithm to do it. And the proposal given doesn't say one word as to how it would be accomplished.
So let's put that aside for the moment. Let's take a step back and look at the
requirements for a currency. To me, they seem to be:
- Multiple accounts, each with a persisted balance that cannot change except in atomic transfers
- The ability for the holder of an account, and only the holder of that account, to atomically transfer funds to another account
- Privacy of the account balance with respect to any individual account holder. (Possibly. Not sure if you care about this, although many folks around here do.)
Are there any proposals that address these requirements, of which you're aware? I should note that since reliance on a "trusted authority" would allow such a party to violate invariant 2, the solution cannot rely on one.