Distributed reputation system

[Bitcoiner]

I propose an idea for a distributed reputation system for Bitcoin. It is based on a "marking" of the public transaction history. Here is how it would work:

Every Bitcoin transaction can be voted either positive, neutral, or negative. Only addresses directly involved in said transaction can tag said transaction. Comments can be attached to the vote.

An address' reputation can be the sum of (positive votes x BTC amount) - (negative votes x BTC amount). Each address owns their own votes.

Since the transaction history is public to all, transaction amounts can be easily verified. All that is needed is transaction tagging. In order to avoid coupling this functionality into Bitcoin, it should ideally be part of a standalone package and call Bitcoin externally (or at the very least, should probably not be part of the Bitcoin core API). That is why Bitcoin needs to be standardized into an API so that third-party apps such as this can be developed; I understand that JSON is a step in this direction, but it has security risks and cannot be public by default. Perhaps what is needed is a tiered approach so that public data can be publicly exposed by default, whereas the full JSON lib needs to be unlocked.

This is just an idea that popped into my head, and I thought that I should write it down. What do you guys think? Please write down your own ideas here, as well!

[Anonymous]

I propose an idea for a distributed reputation system for Bitcoin. It is based on a "marking" of the public transaction history. Here is how it would work:

Every Bitcoin transaction can be voted either positive, neutral, or negative. Only addresses directly involved in said transaction can tag said transaction. Comments can be attached to the vote.

An address' reputation can be the sum of (positive votes x BTC amount) - (negative votes x BTC amount). Each address owns their own votes.

Since the transaction history is public to all, transaction amounts can be easily verified. All that is needed is transaction tagging. In order to avoid coupling this functionality into Bitcoin, it should ideally be part of a standalone package and call Bitcoin externally (or at the very least, should probably not be part of the Bitcoin core API). That is why Bitcoin needs to be standardized into an API so that third-party apps such as this can be developed; I understand that JSON is a step in this direction, but it has security risks and cannot be public by default. Perhaps what is needed is a tiered approach so that public data can be publicly exposed by default, whereas the full JSON lib needs to be unlocked.

This is just an idea that popped into my head, and I thought that I should write it down. What do you guys think? Please write down your own ideas here, as well!

It seems easily verifiable to tag transactions like that.

[Bitcoiner]

Doh... Just thought of an easy way this can be hijacked. Someone could just create a new address, send all their bitcoins to it, and tag it with a good result...

That sucks. It's going to be hard to get anonymous btc loans or other services without knowledge of rep. It's not that important for bc right now; perhaps someone has already thought about this and figured it out. Let's keep the discussion flowing, guys!

[theymos]

You could pay the network (whoever solves your block) a fee to make ratings. That should limit abuse, though I don't know if anyone would actually pay the fee.

[Anonymous]

I think for the time being we can mold existing reputation systems to bitcoin.I was also looking lately at http://www.mywot.com/ web of trust which has an api that could be used to build a bitcoin specific browser tool.Maybe some kind of firefox plugin.

[imanikin]

What about P2PRep protocol? Is that what you were seeking? Is there a working implementation of it in something?

http://p2pfoundation.net/Reputation#P2PRep
http://www2002.org/presentations/samarati.pdf

[Mike Hearn]

The problem with trying to define a trader reputation like that is it needs to be connected to something difficult to forge or change. Passports serve as a "good enough" credential like that in the real world. EV SSL certs try to do the same for businesses in the digital world. Unfortunately I don't think there's a form of digital certificate that has the same rigorous connections to a real world identity for personal users in the same way EV certs do for business.

[Nefario]

I think is is non-trivial to create a p2p reputation system.

A persona may participate in multiple communities. These communities may not in anyway be related to each other. Such that the reputation in one community does not usefully transfer to another community.

An example, a person who has built an excellent reputation as a good body builder in the body builder community of bbuilders.com. Their reputation would be little use on...... mises.org with regards to economics.

Personas can be reputable and responsible in one area but this does not mean it will transfer to another area.

Most communities are centralised, for online communities they are centralised around specific websites. The bitcoin community is centralised around bitcoin.org and it's forums.

It is to the advantage of the communities to develope a centralised reputation system. Ebay has one. And it does not only need to be based around reputation, you could buy reputations (with bitcoin) and losing it would cost you your btc deposit(this is poorly worded).

I assume you are specifically talking about a trading reputation system, this needs to be tied into the trading system it is used with (such as an exchage or bidding website).

I have been pondering the details of a trading related reputation system to allow trades to occur using bitcoin that will minimise loss to scams and cheating. I have not finished pondering.

[Daniel]

Decentralized credit systems like Rain Droplet (see http://bitcointalk.org/index.php?topic=2550.0) can also function as p2p reputation tools. Currently there is planning to develop the technology...