B.5.2:16 - Authoring and Review Guidance
B.5.2:16.1 - For abductive-publication authors
For an abductive publication, make the prompt, seriously considered rivals, plausibility grounds and outcome recoverable. If a prime hypothesis is warranted, explain why it is preferred, its allowed use and a meaningful possible discriminator. Otherwise state the abort, defer or split outcome and its grounds under §14.2. Return to prompt-shaping or lexical repair when the question, scope or intended meanings remain unclear.
B.5.2:16.2 - For hypothesis reviewers
Hypothesis reviewers should not ask only whether the chosen hypothesis looks plausible. They should also ask:
- whether the prompt was typed in an admissible way,
- whether at least one real rival was preserved,
- whether the filters named at selection time actually discriminate among candidates,
- whether the selected hypothesis has interpretable implications and a meaningful possible discriminator, with actual availability kept separate from that explanatory contribution,
- and whether any scope inflation occurred during selection.
A polished hypothesis with no visible rivals is usually less trustworthy than a rougher hypothesis whose rival space is explicit.
B.5.2:16.3 - For integrators and assurance leads
Integrators receive a qualified conjecture, not early assurance conferred by an L0 label. Preserve its prompt, rivals, supports, filter rationale and fragilities. Use B.3.3 only for a receiving assurance question, C.11 for a feasible-action or evidence-acquisition choice, and C.28 for causal support when needed. An unavailable probe can leave the explanation unresolved while an independently warranted bounded service action remains useful; do not turn the conjecture into a mandatory work item.