Analyze and express SWAP iterated prisoner's dilemma game as a Mealy Finite State Machine

[oskarth]

Problem

Currently we have a basic hacky end-to-end PoC for SWAP accounting. We have specs that indicate how this can be analyzed as an iterated prisoner's dilemma. However, there are quite a lot of states and inputs that go into making decisions on what to do. Having a more formal model with rigor around this would make it easier to analyze, think about, spot flaws, bugs, and build on.

Suggested solution

We have a finite set of states, inputs and outputs w.r.t. the game being played for SWAP. List these more formally and make a basic FSM model through some means. Ideally it can be used directly in nim-waku, say, but it can also logically be ported over.

Details

Just a brain dump here in terms of states and inputs:

• A node can be in credit or debit
• It can receive message and send cheque
• It can redeem cheques or save for later
• Cheque can be invalid
• The relationship can be new or old
• A node can have various tolerance for delayed settlement
• A node may have promised some form of uptime
• A node may have detected missing messages through other means
• Historical records may be used as part of input (though with simple tit-for-tat memory model can be quite basic)

Start simple though, and leave more sophisticated/probabilistic models for later.

[oskarth]

I have some ideas for how to do this, but would need to dedicate more time to it. I think it'd be very useful to do though.