Results · corrections · retirements
Methods, provenance, taxonomies and research interfaces are infrastructure. This page records the smaller set of outcomes where prior art, formalization, evidence, criticism or implementation has constrained a claim.
Result rule
A result is something the programme is less free to believe after the inquiry than before it.
Positive findings count. So do failed novelty claims, null results, narrowed definitions, abandoned mechanisms and evidence that a simpler existing explanation is sufficient.
Items here may still be internal working results rather than peer-reviewed scientific findings. Their epistemic status is stated explicitly.
The point is not to accumulate defeats. It is to make it visible when the machinery actually bites.
Absorptive capacity, knowledge transfer, knowledge brokerage, innovation intermediaries, transactive memory and idea-evaluation research already cover the generic problem of moving knowledge toward actors able to use or evaluate it.
What survived: a narrower claim-sensitive protocol that makes the authority deficit, requested epistemic function, contribution provenance and resulting warrant explicit.
AI-assisted idea screening, selective prediction, learning to defer, active learning, editorial triage and value-of-information already establish the generic mechanism of protecting expensive human evaluation with cheaper screening.
What remains: test whether a provenance- and authority-aware routing policy can increase useful epistemic state change per expert-hour while controlling false negatives, status bias and presentation gaming.
The current formal-null review finds mature antecedents in production economics, complementarity, bottlenecks, innovation theory, production networks and rebound effects. The programme therefore does not presently claim a new law or primitive merely because AI makes candidate knowledge cheaper.
What survived: an empirical question about how binding constraints, shadow values and rents move when a previously costly capability becomes cheap.
The programme's innovation-search work now treats the adjacent possible and effective adjacent possible as established antecedents rather than territorial novelty claims.
What survived: the practical question of how machine cognition changes search breadth, recognition, validation, assimilation and realization.
Software can test bounded claims about representation, analyst performance, provenance interfaces, cognitive burden and discovery workflows. It cannot by itself establish macroeconomic claims about displaced scarcity, institutional allocation or economy-wide abundance.
What survived: Nemosyne as an experimental instrument for the claims its interventions can actually identify.
What does not belong here
A provenance schema, Memory Palace representation, handoff register, Active Frontier, possibility portfolio or website architecture can be useful infrastructure. Its existence is not evidence that the underlying research claims are true.
Those systems earn their keep by making inquiry cheaper, clearer or more falsifiable. If they consume more attention than they save, they should be simplified or removed.
Retirement discipline
When a proposed Agalmic concept collapses cleanly into established theory, the default response is to use the established theory. The question may remain interesting even when the local label does not.
A failed novelty claim is evidence that prior-art search worked.
Future retired claims and negative results should be added here rather than quietly disappearing from the public record.