Society · Coordination
OrganisationGame
A pure coordination game in which the best joint strategy keeps changing, so the key is how cheaply players can move to the next game
Definition
Mechanism
Classical game theory ranks cooperation problems by how aligned the players' payoffs are, and treats perfectly aligned (pure coordination) games as needing only good communication. An organisation game keeps the payoffs perfectly aligned but lets the game change over time: in a conversation, at first both talking is best, later one should listen, later ending it is best for both. Players now need to know the current value of information to each other, which they often lack even with identical incentives. The ease of moving from one game to the next depends on the Laterality of the change, the number of players who must switch strategy. Good positions are those from which the likely next games can be reached with low-laterality moves, e.g. leaving pauses so either party can end a conversation alone. Not following an instruction can then be computation, not selfishness, as in the overlapping work of ChimericalComputation.
Invariants
- Payoffs are shared equally in every outcome, so incentives are perfectly aligned
- The best joint strategy changes over time
- How easily the game changes depends on how many players must change strategy
Risks
Failure modes
Assuming that aligned incentives make coordination easy, ignoring changing games
Strategies that require every player to switch at once, e.g. no pauses in which anyone can end the exchange, producing zombie projects
Treating non-compliance with instructions only as a defect to punish
Relationships
Builds on
References
Natural work overlaps: doing more of one task always means doing a little less of another, and imperfect alignment between them packs more function into the same agents
The number of functional units that must coordinate for an action to happen; the more units, the less likely it happens
Used by
References
Pre-position part of a possible future configuration so that, when the game changes, most agents already fit the new optimum and only the remaining fraction must change strategy
Build in points where one party alone can stop or change course, so that ending an activity never needs everyone to switch at once