Mutual coordination is usually more or less manifest, like waltzing together.
In modern game theory, the above payoff matrix is called a coordination game because (loosely) it is in each player's interest for the other to correctly predict his strategy; if one player chooses "Left," he hopes the other realizes this and chooses "Left" as well.
The important aspect of the above coordination game is that there are two equilibria, in the sense defined by John Nash, in which each player's strategy is a best-response, given the strategy of the other player.
However, unlike the famous Prisoners' Dilemma, in a coordination game the incentives do not compel the players to suffer in an undesirable (specifically, Pareto suboptimal) equilibrium.
Klein believes that this Schelling coordination corresponds to his (Klein's) second type of coordination, i.
It would seem that Klein's two meanings of coordination are both used in the typical discussion of Schelling coordination.
If Klein's taxonomy does not neatly confine Schelling's usage of the term coordination into a single box, Klein's treatment of Hayek is even more dubious.
W]e may say that, when Hayek, Polanyi, and Coase spoke of coordination in economic systems, the dedicated opponents [of] any conscious effort to arrange society as a whole meant, in fact, pleasing arrangement.
Recent research has yielded significant results on coordination models and languages.
Control-driven models, such as Abstract Behavior Types (ABT) , Law Governed Interaction (LGI) , ROAD , Reo , Orc , and CoLaS  isolate coordination by considering functional entities as black boxes.
There are control-driven models, such as ROAD, CoLaS, TuCSoN with Agent Coordination Contexts (ACC)  and Finesse , to name a few, that try to mitigate the scalability issues of open distributed systems by adopting role concepts.
The ARC  model differs from these models by separating inter-role coordination and intra-role coordination and distributing the coordination activities to coordinators and roles respectively.
Though the above actor-based models, like ARC, separate coordination from the functional core of a system based on concurrent actors, the focus of ARC is to address the dynamicity and scalability issues in coordinating large set of autonomous and asynchronous entities.
A set of coordination models has been proposed to address the coordination issues based on the Actor model [1, 2], such as Frolund's Synchronizer , Venkatasubramaniam's TLAM (Two-Level-Actor-Model) , and Varela's director .
ARC, however, is a three-layer system, with functional behavior confined to the lowest level, and coordination to the upper two levels.