@cwgoes and I will contribute two sections for refining the analysis of the game:
Role Separation
Starting from the notes here we should check wether this role separation for solving is complete and refine the analysis of their composition with each other and internally.
Given no further refinement of the basic roles solvers are composed of is necessary, the broad problem fields to be treated will be:
- Selectors will deal mostly with time sensitive things like intent flow influenced by network traffic and other lower layers: They choose which intents to include in the batch.
Examples:- One node that has a wall clock and creates a batch for every minute that passes.
- Three nodes that pick intents, compute a union and run consensus over that.
- Searchers will be concerned with computationally intensive matching.
- Pickers will require most mechanism design work, since there will be questions on how to choose utility functions to pick amongst the possible solutions for state updates.
- Provers can be used to outsource computation of proofs.
@AHart as far as I understand right now, this role separation should not affect the formalization of the solving itself.
The solver decomposition might require some refinement, in the sense of modeling distributed solving, but Selectors and Pickers should “just” happen before and after solving and are more mechanism design problems.
Market Structure Taxonomy
Given the above roles, refine solver composition (section 4 of the intent machines ART) and analyze which market structures can potentially be induced.
Then, work out a mapping from the example applications (from earlier sections) to entries of the taxonomy.