Reputation and trust systems

Approach

One approach is to create systems that have a wide range of ability to set functions, and then try figure out what parameters/functions actually do well.

Ideas

Criteria

List of specific ideas here.

Vlads Reputation System.

On-blockchain vs off-blockchain

Reputation systems dont necessarily require consensus system, and Ethereum contract execution carries a cost. For this reason it might be useful to have an off-blockchain version. For instance assume reputation system consists of pubkey-signed statements regarding opinions of other entities. In that case you basically find those signed statements in the system, this means you depend on what you find. If you find a statement by Y 'i love X', you cannot necessarily exclude that there is on that says 'i hate X' aswel.

So off-blockchain reputation systems like that have to figure out how to ensure people can get the statements relevant to them.

Hybrid

Hybrid versions are possible too. For instance, to participate you put some asset or ethers at stake in a contract. Then you make those signed statements. The contract can try enforce rules about statements, like 'cant add to repute more than 1 person per day', by punishing taking away of the stake in the case two statements are submitted that imply that you violated the rules.

Testing these

It might be good to have some way to simulate which methods might work, or have criteriums/metrics that can be applied. It is preferable if these can be generally be applied to compare approaches apples to apples.

Trustlet.org has suggested data sets.

(future expansion?)