Society · Protocols
PropheticQuorum
Aligning predictions before aligning votes
Definition
Mechanism
A two-stage consensus protocol. Phase 1 (reality Check): Agents receive a proposed action. Each runs a local SimulationTrace to predict the outcome State. They vote to confirm they share the SAME prediction. If predictions diverge, the protocol halts for model alignment. Phase 2 (Value judgment): Only after predictions match do agents vote on whether the predicted State is desirable. This ensures consensus is based on shared reality, not just shared language. It splits consensus into two phases: reality Check (via regime sensing) and Value judgment (via NormativeJudge).
Preconditions
- Determinism level of Simulation engine is high
Invariants
Risks
Failure modes
Model divergence (agents cannot agree on what will happen, so they never get to vote on the plan).
High compute cost (N agents running N simulations).
Relationships
Builds on
References
Infrastructure
Side-effect-free observational verification yielding Verified/Falsified/Unknown
Mind
Value-based state evaluation
Mind
A virtualized execution
Mind
Pre-execution mental model
Infrastructure
System condition at time T
Infrastructure
A measure of utility
Provide standard decision mechanism with integrity guarantees
Nothing in this library builds on this pattern yet.
Reference
Parameters
confidence_weightingBoolean, {true, false}- Weight votes by stated confidence
prediction_thresholdProbability, [0.6, 0.95]- Agreement level for valid prediction