Phase 2
Phase 2: summary adequacy and dynamic metadata reclamation
Scientific protocol 2.2; backlog revision 2.2.1. The governing specification is
the frozen protocol. Implementation and evidence are tracked in
issue 84 and its published backlog inventory.
Scope
- Preserve the seven legacy strategy meanings and all existing replay formats.
- Define independent policy scope, usable authority, measured work, immutable
intent tapes and static projection equality. - Test the two matched-prefix families and clone controls in 12,600 runs.
- Prove or explicitly refute F2.1-F2.6 with checked assumptions; construct the
bounded fixed-point quotients and shortest distinguishing continuations. - Implement irreversible handle retirement and the contracted authority
skeleton, with the separate 1,536-trace descriptive stress campaign. - Compare the independent interpreter and the pinned seL4 correspondence subset.
- Freeze source/claim comparisons, analysis contracts, prospective manuscript
methods, reproducibility records and the Phase 2 evidence release.
Boundaries
Use complete information. Material carriers, persistence, restart, partial
observation and depth-dependent retention are forbidden. F2.7 and csaI's
material/recovery evidence are deferred to Phase 5. A static projection is
distinct from full replay state. Model work has no wall-clock interpretation.
Exit
Close only after P2-31 and all of its dependencies have accepted evidence.
Require preserved legacy behavior, valid semantic/replay checks, passing
negative controls, checked formal results or explicit refutations, the declared
finite and external validation, complete frozen campaign outputs, and
regenerable analyses and figures. Retain every unsupported or negative result.
Report the scientific outcome as supported, unsupported or
inconclusive, using the frozen inference and claim rules. A positive
divergence effect is not required for completion. A broken oracle, failing
control, missing required validation or unresolved required review cannot
satisfy the exit. Report method correctness, evidence completeness, manuscript
completeness and journal readiness separately.
Phase 2 does not establish whole-kernel refinement, native performance or the
Phase 5 recovery supplement. Journal readiness additionally requires the actual
contribution and independent scholarly review. A research report may state a
valid negative result or an unresolved publication contribution.