Modeling Change Without Breaking Promises

Promise theory defines a method by which static service bindings are made in a network, but little work has been done on handling the dynamic case in which bindings must change over time due to both contingencies and changes in policy. We define two new kinds of promises that provide temporal scope for a conditional promise. We show that simple temporally-scoped promises can describe common network behaviors such as leasing and failover, and allow an agent to completely control the sequence of sets of promises to which it commits with another agent, over time. This allows agents to adapt to changing conditions by making short-term bilateral agreements rather than the long-term unilateral agreements provided by previous promise constructions.