An interesting aspect of this work is the notion of intention/promise and commitment of actions, which I somehow see a relationship to distributed transactions.

I was familiar with the work of (Winograd & Flores, 1988) on “action conversation” as a basis for task management in a collaborative environment, but in my opinion it does not have the same formal basis as protocols in distributed computing.

I will add a more basic problem: consensus, where actors must agree on the same decision when there are failures, but it is not possible in an asynchronous system (FLP impossibility result).

I hear long time ago that there are four basic problems of coordination:
1. Reach a common time notion in a systems
2. Execute some actions simultaneously
3. Exclude mutually the execution of actions
4. Order the execution of a sequence of actions following some rules (scheduling)

Thanks for the question. I think "coordination" in distributed systems is always about establishing some kind of agreement between several processes or actors to achieve a common objetive. Coordination has to do with control of distributed actions and communication for reaching agreement.