Topic: New wiki page on Bitcoin credit (Read 2789 times)

This sounds a lot like a Collateralized debt obligation, wherein loans are aggregated into a pool and split up into tranches based on assessed risk. In this case, lenders who bid a lower rate of return will get the low-risk tranches and are first in line to collect when the pool matures (less risk, less return). Lenders who bid a higher interest rate are last in line to collect, but have the most to gain if none of the borrowers default on their loans (high risk, high return).

CDOs have a lot of negative publicity right now because of the subprime mortgage crisis in the US, so I think anyone willing to undertake such an endeavor in the Bitcoin world would be wise to learn from the mistakes of others.

I think the web of trust is a key element here.  The pot could also enforce a minimum level of trust, or there could be parallel bidding for that in the same fashion as interest rates.

I think Wei Dai's idea suggests a practical approach for Bitcoin lending.  Assuming a widely used reputation system (like #bitcoin-otc's web of trust is becoming), negative ratings can take the place of leg breaking.  With large enough participation in a pot, there could be some statistical assurances about default rates.  Perhaps a trusted website could operate a service something like the following.

• Borrowers indicate how much they want to borrow, how much collateral they have and how much interest they're willing to pay.  The site generates a BTC address to which the borrower sends the collateral.
• Lenders indicate how much they want to lend and what interest rate is acceptable.  The site generates a BTC address to which lenders send their Bitcoins
• The site issues signed receipts for these BTC deposits
• The site accepts borrower and lender requests for a week prior to the loans being issued
• At the end of the week, the market clearing interest rate is calculated based on the offers
• Borrowers paying more than market clearing rate participate in the pot.  Lenders accepting less than market clearing rate participate in the pot.  Others offers don't participate and their coins are returned
• When the loan term is up, the borrowers return the BTC with interest
• The pot operator adds positive reputation to the honest borrowers and negative reputation to the dishonest borrowers
• The pot operator distributes the pot value to lenders based on priority.  The lender accepting the lowest interest rate gets paid his principal and the interest he requested.  This continues through each lender until the one requesting the highest interest rate has been paid
• The pot operator publishes the Bitcoin addresses that participated to allow auditing

This approach could scale to large numbers of participants.  The interest rate is set by the market.  The pay out protocol gives the highest risk to those lenders expecting the highest reward and provides motivation to offer the lowest interest rate one is willing to accept.

There are many possible variations on this theme.  There could be a minimum reputation threshold before a borrower could participate.  There could be a minimum collateral limit.  There could be separate pots with different limits and loan durations to accomodate the risk profile of different lenders.

Conveniently, the whole thing could operate as a hidden service since it deals only with Bitcoins and a distributed reputation system.

That's an interesting system, but it seems rather academic (that said, I once thought the same of BitCoin yet here we are).

Leg-breaking is I'm sure very satisfying, but ultimately you're still out of pocket. A key part of lending is assessing the credit-worthiness of the borrower to try and minimize the amount of leg-breaking required. With complex lending you may have to consider all details of their business scheme to decide "is it good or is it wack?". That isn't possible under Wei Das design.

The problem it solves is also obscure. It is a complicated way of making or receiving anonymous loans between non-anonymous participants. But that's exactly the same model of privacy already used in the real world. If I take out a loan from my bank, only they and I know about it. I didn't need any Pot to disguise that. The built-in privacy of the financial system (or BitCoin) is good enough. The scheme isn't useful for some kind of cypherpunk crypto-anarchy either, as all participants need to be strongly identified not only between loaner and loanee like today, but to everyone!

Finally, the system distributes responsibility, which is the opposite of what the legal system requires to go after somebody. In a Pot system, when somebody doesn't repay it's clear who is at fault, but not who should take the trouble to go after them.

Daniels page appears to be a subtle advert for Ripple, not that there's anything wrong with that. Ripple is kind of the opposite of anonymous lending in that people are all too painfully aware of who is in default ... because it's their own friends who may transitively owe them on the behalf of other people!

BitCoin doesn't seem to have much to say on the topic of credit. Whilst you can do some clever things with the scripting system, credit is ultimately about people and the trust between them - not something that can be easily encoded into a protocol.

Leg breaking. 

what "serious consequences"?

Back in the 90's on the cypherpunks mailing list, Wei Dai (of Crypto++ and b-money fame) invented a system of anonymous lending. Of course the classic problem with anonymous loans is that you don't know whose legs to break if they don't pay it back. Wei solved this problem, with some conditions.

The participants in the protocol are borrowers, lenders, and bystanders (who don't want to lend or borrow right now, but might wish to do so on future runs of the protocol). The key is that all participants are fully identified, to the extent that they are vulnerable to that all-important leg-breaking. Nevertheless, the protocol conceals the role of each participant, and the amount they are borrowing or lending, if any.

There needs to be a system of anonymous payments, and a simple trusted machine called the Pot. (In practice, the Pot would be simulated by the participants, using a cryptographic multi-party computation.)

In a preliminary phase, participants would anonymously negotiate and agree on the amount each person would borrow or lend, such that the total amount borrowed equalled the total amount lent. The actual protocol then has four phases, two for borrowing/lending, and two for repayment.

Phase 1 is private. Each participant anonymously puts money into the Pot, and gets in exchange a signed receipt for the amount. Lenders would put the most money in, bystanders less, and borrowers little or none. You'll see why in a moment.

Phase 2 is public. The Pot now has a certain amount of money in it; this is divided equally and distributed publicly to all participants.

The borrowing/lending is now done. People who put more money in the Pot than the per-person distribution in phase 2 are net lenders; people who put in less are net borrowers; and people who put in the same amount are the bystanders, with no net change. No one knows whether anyone else is a borrower or lender.

When it is time to repay the loan, two more phases are run, the mirror images of the first two.

Phase 3 is public. Each participant publicly and verifiably puts back into the Pot the exact amount taken out in phase 2 (or, that amount plus interest). This is leg-breaking time! No excuses, everybody's got to pay up.

Phase 4 is private. Each participant anonymously presents his receipt from phase 1 to the Pot, and gets back that amount (or, that amount plus interest). This unwinds the earlier transactions and all the funds are back where they started.

All that is visible publicly is that everyone splits up the pot evenly in phase 2, and then returns that same amount in phase 3. But behind the scenes, transfers of value are taking place. And failure to repay a loan means failure to carry out phase 3, which be publicly visible and will lead to serious consequences.

Just created a simple page on Bitcoin credit:

https://en.bitcoin.it/wiki/Credit

Improvements welcome!