Earlier this year I wrote up an idea I called "multisig derivatives" that could be used to hedge against volatility. (I guess that would've been useful during Clampocalypse.) I've been thinking about it again, and now know how to present the idea in what I hope is a simpler way. It's basically a way of simulating buying and selling cryptocurrencies relative to other assets (e.g., fiat currencies). I wrote up a short explanation yesterday, using bitcoins and dollars:

https://medium.com/@trentrussell/contracts-simulating-crypto-fiat-exchanges-6f9421ea520c

I plan to implement this to work over clams, using clamSpeech to record the details of the contract.

One of the issues I'm not sure how to handle is identifying the parties. It's all intended to be pseudonymous and so the party is really identified by their pubkey. However, writing the full pubkey for 3 parties into each clamSpeech with a contract wastes space.

A way to deal with this would be for people to register "identities" (presumably short handles or pseudonyms, up to 14 characters long) associated with a pubkey. Then the contracts could reference the "identities" and look up the corresponding pubkey.

Here's how this could be done on-chain:

Suppose Alice wants to register the handle "Alice" and associate it with the pubkey Q which has address A. Assuming there's not an "Alice" already register, she could make a tx which spends with a first input from A and clamspeech "registerid:Alice" -- the pubkey Q would be visible since the first input spends from A, and its clear that the holder of the private key for Q signed the tx. This registration could be stored in a database (independent of Clams). Then in contracts when "Alice" is used, Q could be looked up.

It would probably also be important to allow Alice to change her pubkey registration. Changing from Q (A) to Q' (A') could be done by spending from A (input 1) and A' (input 2) in a tx with clamspeech "changepubkey".

Some of these identities could advertise themselves as escrow agents by registering certain terms under which they will act as an escrow agent.

Before going into any more details, I'm curious if anyone has comments. Are there objections to this? Or a clearer way to do this?

There is another possibility I've considered. Instead of storing the "identities" explicitly in the clamSpeech with the contract, the "identities" could be determined for each "contract tx" by looking at the inputs: (1) for party, (2) for counterparty and (3) for escrow.

I can see arguments for doing it either way. Thoughts?