3 ms·

I've been thinking about the arbitration problem, what do you think of the following possible implementation of a "3rd party free" arbitration? Alice desires t

I've been thinking about the arbitration problem, what do you think of the following possible implementation of a "3rd party free" arbitration? Alice desires to buy asset of V value from Bob with bitcoins. Alice sets up a k1xV value address and requires Bob to set up a k2xV one, and they give up control on those addresses for at most time T. Also, Alice and Bob set up a conditional transfer of this value V to a charity (or destroy it) if both don't agree on retrieving the wallets before time T. That way, if the merchant believes he will be punished (e.g. k2 = 2) he shouldn't engage in fraud. Also the merchant should hold the same faith in customers (by setting e.g. k1 = 1 for new customers, k1 = .5 for old ones). Thoughts? Note: As a bonus, the merchant could make his transaction history transparent to try to prove he punishes every fraud, and customers can likewise demonstrate some "proof of transaction" showing either they conducted transactions without issue or punished fraudsters.

nadaviv 13y ago
Huh! I actually started working on a system that does something very similar, which ended up being morphed into Bitrated after a few people convinced me that it wouldn't work. This idea has been brought up in the Bitcoin community from time to time [1] [2], usually referred to as mutual assured destruction, risk deposit or nash equilibrium. I personally really like this idea and think that it could work well with rational players. The problem is that there is room for abuse - one party could tell the other party "okay, I'm not going to finish that deal with you. we can both either lose everything, or you pay me 70% of the risk deposit and get to keep 30% of it". In the long run, rational players should always refuse to such offers. If scammers would know that everyone always refuses, they'll quickly get into a loss and stop doing that. The problem is that I'm not sure how people would react in such situations - why not take the 30%? As to your idea of publishing history - that could work, if that system could somehow know who's a fraudster and who isn't. Its kinda hard to tell by just looking at the blockchain and seeing what happens with the coins, there could be multiple reasons for the funds getting split. [1] http://nashx.com/HowItWorks http://nashx.com/HowItWorks (requires trust in the website operator, as he holds the funds. at least it was like that last time I saw it) [2] https://gist.github.com/etotheipi/2305966 https://gist.github.com/etotheipi/2305966
- darkmighty 13y ago
  Interesting, let me add that you should let people convince you it won't work without proof, especially for something as reliant on behavior! I think a reputation system could be built using some kind of peer-to-peer trust system. For example, I'm aware you can prove a certain transaction is yours. Simply by collecting a series of transactions conducted with the merchant and with a series of merchant's trusted peers, the merchant could perhaps decide on the (k1,k2) he might propose; and similarly users may assemble a database of verifiable cases where the transaction backfired (since each user can prove the transaction and the merchant address). It just bugs me -- one of the great promises of crypto currencies is the elimination of need of trusted 3rd parties -- only to be unable to perform useful real world transactions without them. The key to enabling this would be devising a system for voluntary transparency, I believe.
- Anderkent 13y ago
  >In the long run, rational players should always refuse to such offers. If scammers would know that everyone always refuses, they'll quickly get into a loss and stop doing that. The problem is that I'm not sure how people would react in such situations - why not take the 30%? People have a very strong fairness instinct, and will probably refuse such blackmail. There's a famous experiment where some amount of money is given to one person, and he decides how to split it with a partner (50-50, 100-0, etc.). The partner decides whether to accept the deal or refuse it, in which case the money is lost. If I remember correctly most people refused offers under 20%, and that's without the anger that someone trying to blackmail you would cause.
DennisP 13y ago
How do they give up control on those addresses for at most time T? Does that require trust in a service, or is there a way to do it with script?
- nadaviv 13y ago
  You can [1] create a transaction that destroys those coins (spends them on fees, sends them to a charity or really takes them out of circulation) at some time in the future (using nLockTime [2]) in such a way that it could be replaced with another transaction before that time (using a low sequence [3]). [1] In the future, its not all active yet [2] https://en.bitcoin.it/wiki/NLockTime https://en.bitcoin.it/wiki/NLockTime https://en.bitcoin.it/wiki/Protocol_specification#tx https://en.bitcoin.it/wiki/Protocol_specification#tx [3] http://bitcoin.stackexchange.com/a/2032 http://bitcoin.stackexchange.com/a/2032

nadaviv 13y ago

Huh! I actually started working on a system that does something very similar, which ended up being morphed into Bitrated after a few people convinced me that it wouldn't work. This idea has been brought up in the Bitcoin community from time to time [1] [2], usually referred to as mutual assured destruction, risk deposit or nash equilibrium. I personally really like this idea and think that it could work well with rational players. The problem is that there is room for abuse - one party could tell the other party "okay, I'm not going to finish that deal with you. we can both either lose everything, or you pay me 70% of the risk deposit and get to keep 30% of it". In the long run, rational players should always refuse to such offers. If scammers would know that everyone always refuses, they'll quickly get into a loss and stop doing that. The problem is that I'm not sure how people would react in such situations - why not take the 30%? As to your idea of publishing history - that could work, if that system could somehow know who's a fraudster and who isn't. Its kinda hard to tell by just looking at the blockchain and seeing what happens with the coins, there could be multiple reasons for the funds getting split. [1] http://nashx.com/HowItWorks http://nashx.com/HowItWorks (requires trust in the website operator, as he holds the funds. at least it was like that last time I saw it) [2] https://gist.github.com/etotheipi/2305966 https://gist.github.com/etotheipi/2305966

darkmighty 13y ago
Interesting, let me add that you should let people convince you it won't work without proof, especially for something as reliant on behavior! I think a reputation system could be built using some kind of peer-to-peer trust system. For example, I'm aware you can prove a certain transaction is yours. Simply by collecting a series of transactions conducted with the merchant and with a series of merchant's trusted peers, the merchant could perhaps decide on the (k1,k2) he might propose; and similarly users may assemble a database of verifiable cases where the transaction backfired (since each user can prove the transaction and the merchant address). It just bugs me -- one of the great promises of crypto currencies is the elimination of need of trusted 3rd parties -- only to be unable to perform useful real world transactions without them. The key to enabling this would be devising a system for voluntary transparency, I believe.
Anderkent 13y ago
>In the long run, rational players should always refuse to such offers. If scammers would know that everyone always refuses, they'll quickly get into a loss and stop doing that. The problem is that I'm not sure how people would react in such situations - why not take the 30%? People have a very strong fairness instinct, and will probably refuse such blackmail. There's a famous experiment where some amount of money is given to one person, and he decides how to split it with a partner (50-50, 100-0, etc.). The partner decides whether to accept the deal or refuse it, in which case the money is lost. If I remember correctly most people refused offers under 20%, and that's without the anger that someone trying to blackmail you would cause.

DennisP 13y ago

How do they give up control on those addresses for at most time T? Does that require trust in a service, or is there a way to do it with script?

nadaviv 13y ago
You can [1] create a transaction that destroys those coins (spends them on fees, sends them to a charity or really takes them out of circulation) at some time in the future (using nLockTime [2]) in such a way that it could be replaced with another transaction before that time (using a low sequence [3]). [1] In the future, its not all active yet [2] https://en.bitcoin.it/wiki/NLockTime https://en.bitcoin.it/wiki/NLockTime https://en.bitcoin.it/wiki/Protocol_specification#tx https://en.bitcoin.it/wiki/Protocol_specification#tx [3] http://bitcoin.stackexchange.com/a/2032 http://bitcoin.stackexchange.com/a/2032