Here's my thoughts on how a decentralized reputation system might work. This is a response to casascius's lament (
There needs to be a new bitcoin address format...) that there are no good trust systems for Bitcoin. Also a lot of thinking from
retep's work on how to create fidelity bonds. I haven't seen any proposals like this yet but please feel free to point me to any that I missed.
I'll refer to this concept as 'Repcoin' because it sounds cool and because has not been used yet afaict (see below.)
The Repcoin System1) Repcoins are managed on an alt-chain so they can be traded similar to Namecoins, and to have the value float with respect to BTC.
2) An individual generates a public/private keypair, known as the Reputation account. These would reuse the Bitcoin address format.
3) For example, say one acquires Repcoins by donating a special value (see 4) to BTC miners or possibly a charitable fund address using BTC (straight out of
retep's work on fidelity bonds). A marker in the Repcoin blockchain would identify the creation of the Repcoin, it's destination Reputation account public key, and the source BTC transaction which satisfies the rules.
4) Other Repcoin users can verify that the creation of these coins matches the software rules and they can determine the value of each Reputation account by scanning the Bitcoin blockchain for creation and the Repcoin chain for accounting.
Some rules:
5) Donation amounts are based on a constant purchasing power. Work out an agreeable algorithm that pegs the repcoin cost with respect to a BTC:fiat ratio that floats. This might be an exchange basket. Maybe use fixed point instead of integer coin values.
6) A certain small amount of time-based deflation (easier when it is pegged to USD) is built-in so that older coins have more trust value? maybe not, idk. The early adopter effect would be a good jump start to the system.
Usage:
To use this, the sender would sign a transaction or their receiving address using their reputation keypair. It would turn into a sort of digital identity.
You could decide on the amount you are willing to trust them based on their purchasing power they used to acquire the reputation (retep's fidelity bond.)
Some problems:
7) Bad actors that have spent a while dumping money building up a reputation starting to scam people: This is a very difficult problem, I would think there would be a method for users to burn the reputation of other users by burning their own repcoins in a special SCAMMER transaction - say -2 repcoins for every 1 repcoin burned. Since this is also in the public and signed with the other person's account it becomes a reasonable way of seeing an overall picture of trust. Since real BTC is involved and the reputation is on the line hopefully it wouldn't be abused too much. There could possibly be another message that would nullify the hit to the other user in the case of resolution but the burned repcoin would still be gone.
8 ) Yet another chain to keep track of, though this one is mostly just an accounting table.
Future:
If using charitable organizations, they would need to get their addresses in the software without software patches. It could be bootstrapped with a few orgs at the beginning, then after a certain # of blocks a new election for the top n charitable organizations would be determined, by vote of people by share of their reputation. Users who are acquiring reputation could use any of those n. Organizations would hit up big repcoin owners to get votes for funding, kinda like how the world works today.
See also:
Whuffie, or read at least
the prologue to see how this might turn out in practice.
Uses of Repcoin: nothing on this forum, searching google yields a
very interesting paper (pdf) with this name that attempts to solve this problem but the implementation is still too centralized; and openassembly.org (site down) has decided on this name for a currency on some random page.
Quote from the end of
the 'Reputation Systems for Anonymous Networks paper' (pdf), in the 'Future Directions' section talking about how to decentralize it:
Note that this would require two significant changes: using a digital cash scheme that does not require a central bank, and devising some other mechanism for inflation resistance.
This is what it was like in 2008 I guess!