Library / Problem Structuring and Decision Support Principles Framework
Jump to passage
In this reading

Link to current text

Published source confirmed at last check

Source changed 2026-10-03 14:36:52 UTC · snapshot created 2026-10-03 14:38:14 UTC · last check 2026-10-03 14:55:10 UTC

Compare for this receiving use

Use PSD.9–PSD.12 for values, uncertainty, consequences and robustness. State the candidate domain and the subset being compared, the receiving question, characteristics and scales, protected conditions, evidence and uncertainty, and the comparison rule that the result actually uses. Adequate current comparison results can be reused without running all four patterns.

C.16 supplies characterization; C.11.CRC preserves configuration-relative contribution; A.19.CPM supplies a comparison mechanism when its conditions hold. Consume a G.5 selector result only when its candidate domain, eligibility, Method, evidence, result kind and closure are actually established. A ranked list is not proof of that application.

Explain what the comparison can and cannot conclude. Preserve a partial order or non-dominated directions when the basis does not warrant a total ranking. If one direction is supported, explain its advantage over the strongest remaining rival and the conditions that could reverse it. Keep important value disagreement visible instead of manufacturing a shared weight.

Human consent and transfer, organization continuity and AI safety remain unlike premises. A protected condition cannot be compensated away by a favorable total score. When its own justification is disputed, use PSD.9 §4.4 and C.11.DUA to compare the protected outcome, causal and measurement basis, threshold, additional protection and full burden. Carry the requirement recommendation and its actual amendment limit in the advice; current force still governs available action. Likewise, a comparison of internal development and one mixed arrangement does not rank an unexamined provider, another model version or a materially different probe.

When evidence is thin, first ask whether the missing distinction could change the answer and whether obtaining it is worth its burden and delay. A small request naming the missing result, a retained set or a blocker may be more useful than another elaborate model. A declared comparison is an input to advice, not the recipient’s choice.