Single-leader vs multi-leader vs leaderless conflict resolution

Any distributed social system will inevitably have conflicts that need resolution.

That is, when you have a bunch of different people and teams working on a shared goal, you will inevitably have disagreements that need some way of being resolved.

Single-leader

Disagreements go to a single leader to resolve

• Any disagreement goes to a single leader to resolve.
• Ensures that all decisions are coherent (assuming the single leader is coherent).
• Can have significant delays at-scale given the bottleneck on the single leader.

Multi-leader

Disagreements go to local leaders to resolve; local leaders coordinate to resolve inconsistencies

• Disagreements go to a local leader to resolve, who then coordinates with other leaders. Decisions are eventually coherent as the different leaders resolve inconsistencies.
• If the leaders themselves disagree, some method is required to resolve (which itself could be single-leader, multi-leader, or leaderless)
• There is less delay at-scale than with single-leader, but multiple-leader setup tends to be more complicated.

Leaderless

Disagreement are resolved peer-to-peer without leaders

• Disagreements are resolved peer-to-peer by referencing a shared set of decision principles, beliefs, etc. and a pre-agreed conflict resolution process.
• If leaderless consensus can’t be reached, some tie-breaking method may be required (which could be single-leader or multi-leader)