Mind · Strategy
DiscoveryProtocol
Population-based discovery: Generate (parallel diverse solvers) + Reduce (composition-aware selection) behind hard boundaries
Definition
Mechanism
A protocol for population-based discovery. Distinct from the Discover primitive (query-and-return): Discovery Protocol is the Generate + Reduce architecture for searching a solution space. Decomposes discovery via ConceptualDecomposition into five dimensions that apply wherever a solution space must be searched: variance (generate candidates that are precise in different directions, each generator bound to its own declared cognitive mode), selection (judge which candidates are good), novelty (distinguish structural originality from surface variation, judged by Novelty), composition (merge compatible fragments into solutions no single candidate contains), and saturation (detect when further generation yields diminishing novelty). Two phases behind hard boundaries: generation by many parallel Solvers, then reduction by a solver whose faculty is evaluation and composition rather than generation. The reducing solver routes among modes: aggregation for ensembles, a tournament for adversarial selection, a portfolio for quality-diversity preservation, and Synthesis-based merging of compatible mechanisms. A taxonomist solver classifies outputs into a growing ontological graph for saturation detection. Applies to drug discovery, hypothesis generation, strategic planning, creative production.
Invariants
- Hard-boundary isolation: generators cannot see each other's outputs during generation; the reduction boundary is the first synthesis point.
- Cognitive-mode diversity: each generator is bound to a declared mode that no other generator shares; one model sampled at different temperatures does not make distinct modes.
- Non-compensatory reduction gate: novelty is a gate condition, not a tiebreaker — a structurally novel mechanism survives even if it scores lower on surface plausibility.
Risks
Failure modes
None stated.
Relationships
Builds on
Composes with
Finding what a concept is made of: locate it, carve it into the dimensions every case it claims needs, test each, search for what is missing, route real cases through it, say how the parts make the whole, and carve again below
Evaluates structural distinctness
Combining elements
Nothing in this library builds on this pattern yet.
