In a similar vein (pie in the sky) to my previous post on bitcoin CDSes, and stemming from some discussion with BTC_Bear, vragnaroda and imsaguy on IRC, I wanted to gather some thoughts on credit ratings in bitcoin land.
As BTC_Bear said in another thread, interest rates around here are often very high, and that's often because there's almost no information to build a reasonable expectation of the likelihood of default. Having a good reputation on WOT and the forum are a good indicator, but fraud isn't the only reason for default. Some people are irresponsible, or make bad predictions, or might just be unlucky. And I think it'd be nice to have some independent credit reporting facilities on borrowers.
I've already gone on about how I think the approach people often use in loan requests (posting a deposit address up front) is flawed and could cause some unpleasantness if multiple people claimed a loan. But taking things a step further, I think it would be cool if we could design a standard format for making loan requests, and it would be built on top of a GPG identity (which might also be associated with a WOT identity).
One example of this might be to have you paste information into a form like
BTC requested:
Loan duration:
Type of loan: (whether it has periodic "coupon" payments or just pays back a lump sum at the end)
Special features: (anything unusual)
Reason for loan:
Other loans or similar contracts currently open:
And crucially, a link to a credit report. For now, this could just be a text file in a standard format in a standard location (writable by people who are willing to sign their contributions), containing a sequence of signed statements from previous lenders about how the loan worked. Specifically, this would involve a double-signed statement at the point the loan is issued and another double-signed statement when the loan is fully repaid (or defaulted on). The gpg signature scheme would include transaction IDs and should be fairly secure and fair to both parties. If the repayment statement is only singly signed there might have been a disagreement, but that's fine and should go on the report. Other factors listed should include things like punctuality and how communicative the borrower was, and whether other special conditions of a loan were respected.
Entities could also voluntarily disclose other information about themselves, like real identity, or ties to well known online pseudonyms, which could also be signed by third parties who are willing to vouch for them. How meaningful those vouching signatures are depends on how much you trust the person signing them (this could get cumbersome)
The centrally available credit reports would be visible to everyone, and there would be one for every GPG identity. As these reports would often get quite long (and the meaningfulness of signatures depends on the trust of the signer), it could be a good place for third-party credit evaluators, who would develop their own criteria and look over longer credit reports and synthesize them into a few key points, and then would sign those syntheses (possibly something as succinct as a score, but who knows). To discourage people from discarding bad credit reports by creating new identities, the length and age of the report would also be factored positively into the overall rating. One open question is compensation for the evaluators so as to not create too many perverse incentives, but I don't think there's too much we can do to avoid that other than to double check their reports.
Ideally the reports would not be a text file, and would just be a database of signed statements by one entity for another entity that could be searched by both. This would allow people to evaluate lenders, too. Ideally, it'd also be decentralized (much more like bitcoin) but we want something lightweight or nobody will use it, and it could already become fairly complicated.
Anyway, just more random thoughts and I'm half asleep, so I apologize if I've written anything particularly stupid
But I think that a better infrastructure would lead to better rates for longer term loans than we're able to give these days. Lending would still be risky (exchange rate risk!), but the risk would be a bit easier to evaluate. Maybe one day we could have a real lending-specific website that dealt with all the complicated issues above
