The Lockbox Protocol: How Cooperation Begins Without Trust
I have been thinking a lot about what systems are needed in a world where traditional institutions have been destroyed. The main reason I went down this path is that economics always assumes a hidden power that is able to enforce rules. For example, in traditional bilateral trading games someone has an item and another person would like to have it. The economist then designs a game or new mechanism that enables trade to happen. But what if the buyer simply threatens violence and takes the item? The response is to just say that’s illegal. In other words, they assume there are laws and institutions that enforce these ownership laws.
In what follows, I am going to describe a specific protocol. I am not claiming that it solves all problems but it’s one version of an initial solution.
Alice needs a shelter built. Bob is a builder. The job is worth $90 and takes three days.
If Alice pays upfront, Bob can take the money and disappear. If Bob works first, Alice can refuse to pay. Both know this, so neither moves.
Picture a box set in concrete with two locks. It opens only when both keys turn at once.
The key step is to turn a one shot interaction into a multi shot one. Alice and Bob split the job into three stages worth $30 each.
Each morning Alice puts $30 of payment into the box and Bob puts in $30 of his own money as collateral. The box is sealed. At the end of the day, if Alice is satisfied, both keys turn. Bob takes his payment and his collateral back and they start the next stage. If Alice is not satisfied, the project ends. Bob loses the day's collateral and receives no payment. The remaining stages never happen.
In a one-shot deal you decide whether to cooperate on no information at all. Staging produces information as you go. After day one Alice has seen how Bob works. Bob has seen whether Alice opens the box when work is delivered. The stages also change what each party is protecting. Bob delivers on day two to keep his collateral and to keep access to day three's payment.
For the savvy you should detect a potential problem here. In finitely repeated games with a known last round, cooperation can unravel backwards. If day three is the end, Bob's reason to deliver weakens, and if Alice anticipates that she may not enter day three at all, which weakens day two.
For any of this to work there has to be some uncertainty as to whether Alice and Bob will interact again after this transaction. If there is some chance they will play again, then this unraveling is less likely. Another indirect way you can create multiple future interactions is to realize that Alice and Bob need not interact directly and that information from this transaction could spillover on to other interactions.
A platform can make this mechanism explicit rather than leaving it to word of mouth. Bob's opened lockboxes are visible. So is Alice's record of opening them. A potential counterparty can see this before deciding whether to enter a box with either of them. That is the difference between a clever arrangement between two people and something a market can run on.
The lockbox record is the transaction. It opened or it did not, and the box does not care what anyone thought about the experience. The proof is in the pudding.
So the question for a platform is how to stage a transaction be staged, where can exposure be capped, and where can adherence be made visible to the next counterparty in order to generate cooperative outcomes.
Ansible Architecture designs incentive systems and interaction protocols for platforms and marketplaces.
