Binary decision systems tend to force possibilities into two buckets: active or discarded. That is efficient only when the world is static. In changing environments, a possibility that is poor now may become valuable after cost, capability, timing, evidence, dependencies, or surrounding state changes.
Graded Dormancy preserves that distinction. A branch can leave active competition without being erased from the recognized possibility space.
Dormancy is a spectrum
Not every inactive possibility deserves the same treatment. A branch whose blockers are stable can sit deeply dormant. One whose governing conditions are approaching a threshold can remain lightly dormant. A branch can move toward active consideration as evidence accumulates that its constraints are weakening.
The grade therefore represents how much current attention and compute the possibility deserves, not a judgment about its permanent worth.
Why deletion loses information
When a system permanently deletes a branch because it currently scores poorly, it also deletes the option value of future change. The system must later rediscover the possibility from scratch, if it rediscovers it at all.
Dormancy preserves a compressed memory that the branch exists, why it is inactive, and what kinds of state changes could matter to it. That allows resource allocation to remain selective without collapsing the future search space.
The connection to Conditional Failure Memory
Conditional Failure Memory records why an attempt failed under a particular state. Graded Dormancy governs what happens to that possibility afterward. The failure can reduce its priority while the blocking conditions remain true, but the branch remains available for reconsideration.
This separates learning from permanent prohibition. The system remembers the lesson without pretending the conditions that produced it are eternal.
Causal reactivation
Dormancy becomes operational when paired with Causal Reactivation. Relevant state changes can increase a dormant branch’s grade and eventually return it to active evaluation. Irrelevant changes should not wake it.
The useful question is therefore not merely whether something changed. It is whether a condition connected to this branch changed enough to alter its viability.
Why this matters
Agents, research systems, strategy engines, and human decision processes all face the same resource problem: there are more conceivable possibilities than can receive full attention at once. Graded Dormancy creates a middle state between expensive active search and destructive forgetting.
That middle state matters because possibility has temporal structure. Some branches are not answers yet. Preserving them makes it possible for a system to become more capable when its world changes, rather than remaining trapped by conclusions formed under an older state.