Comparison Discipline

Summary

The comparison discipline governs Stage 4 of the six-stage pipeline. Stage 3 maps the feasible options; Stage 4 applies categorical, prior, and other entry rules. Given the Means-Bound, the Possession constraint ordinarily removes options that institute, expand, renew, or retain possession as their proposed means or end. Filters attach at the operative level: a proposed participant-scale rule prevents automatic attribution of a higher-level possession to every option of an actor inside it. A separate bounded-dismantling rule can admit a constrained path that itself carries possession already operating before the comparison. If no option remains, Stage 4 records no eligible option and stops. Otherwise it compares the eligible pre-classification claim-satisfaction profiles and produces one set-level output. The discipline is deliberately incomplete: a complete output need not be a complete ordering, and a pairwise ranking does not by itself establish a top option or top set.

The discipline adopts a defended counting rule for claims of the same order and proposes a relevance band that silences sufficiently lesser claims against a graver one. It supplies no interpersonal metric for the band and no exchange principle where one graver claim faces more numerous lesser but still relevant claims; instead it requires the gravity frame to be declared in advance and contestably, imports the ranking, and makes each unsatisfied claim's classification determinate relative to that declared frame. Counts remain counts of distinct holders whose claims would be failed; they do not create an aggregate claimant or imply that losses to one party are compensated by gains to another.

When Stage 4 fails to establish a top option or top set, the defended default selects the eligible option that minimizes irreversible foreclosure, with delay and continuation entered as options carrying effects of their own. A proposed non-reversal constraint preserves established strict pairwise relations where they can be jointly honored. If the default does not discriminate and time requires a choice, fair procedure selects for action without supplying comparative vindication.

Entry filters

Several rules govern the option set before ordinary comparison:

• A standard may not be enforced by putting anything into, or taking substance from within, a competent holder's body against a valid present refusal. Enforcement supplies the scope and limits of this rule.
• Provision precedes prevention: a present provision claim is not answered by substituting the prevention of possible people.
• A directed claim defeats a preference, institutional appetite, prestige, growth, or other claimless consideration. Provision Claims prevents public functions from being excluded or admitted by relabeling. This filter is the Stage 4 application of claim-grounded adequacy, not an independent comparison principle. Refusing that adequacy condition admits claimless considerations as candidate reasons but supplies neither a comparison rule nor comparative vindication.
• A valid performed exercise defeats any hypothetical, predicted, or constructed exercise offered in its place. Authority Across Time determines which exercise governs.

Where a consideration is asserted as a directed claim, it enters this filter as claimless only if its ground is absent on the current Stage 1 record. A live unresolved ground is not admitted as a claim, but neither is it defeated as claimless. Its admissible branches remain explicit until a governing uncertainty rule or an invariant result supports an unconditional output. An operational output under such a rule does not erase those branches or supply satisfaction, override, or wronging. Considerations offered expressly as impersonal aims remain claimless without first being redescribed as failed directed claims.

Given the Means-Bound, a competing claim cannot make another party's possession part of what its holder is owed. The ordinary constraint filter therefore removes every option that institutes, expands, renews, or retains possession as its proposed means or end. The scale-invariant reach of that exclusion remains conditional on the disputed pairwise-exhaustion premise.

At participant scale, the filter first applies Possession's operative-level rule to the actor's own course of conduct. A higher-level arrangement's current possession is treated as background rather than attributed to every participant option only where the complete Stage 3 map establishes all of the following:

• the arrangement's existence, scale, duration, and characteristic wrongful function remain invariant across the actor's feasible options for reasons independent of that actor's continued contribution;
• the actor lacks authority to direct those dimensions or to decide whether the possessive function continues;
• the actor's own conduct is not itself established to bear the three possession marks, perform, direct, commission, expand, or renew the characteristic wrongful function, or function as its material certification or reproduction; and
• the option map includes feasible conditional or coordinated courses and does not infer invariance merely from replaceability, low individual causal share, or the prediction that another participant would perform the same wrongful act.

Where these conditions hold, the ordinary filters apply to the claim-relevant effects and relations the actor controls. An option does not require bounded-dismantling admission merely because possession persists at a higher level under every option. Participation, controlled contribution, holding, initiation, imposition, and any claims created by the actor's role remain in the map; the rule transfers neither permission nor acquittal from the option-level judgment to the arrangement. A live evidential question about the actor's operative role remains in the Stage 3 description rather than being resolved by attributing the arrangement's status downward. Exit from Wrongful Structures applies the resulting map and requires renewed analysis whenever authority, coordinated alternatives, replacement effects, or the actor's function changes.

The proposed bounded-dismantling admission is a narrow eligibility rule for possession already operating before the compared exit paths begin, rather than possession instituted by an option under comparison. Prior creation or maintenance by the current participant remains separately assessable. A path carrying that possession may remain in the eligible set only where it:

• fixes an endpoint at which the three marks cease at the operative level carried by that path;
• does not expand the population assigned to the possessive function, its operating capacity, or the function itself;
• ties each period of continuation to an identified claim that faster cessation would leave unsatisfied, rather than to revenue, convenience, preference, or preservation of the old benefit; and
• remains open to acceleration as provision, claims, and feasibility change.

Admission is only to comparison and forced-set audit within a bounded exit analysis. It does not place possession inside another holder's claim, justify the possession to its victims, establish that the set is forced, or make the path selected or required. Exit from Wrongful Structures still requires immediate cessation where a clean path exists or every burden of cessation is adequately justified. Only where the admitted eligible set is nonempty and every member wrongs someone can Necessity govern a frontier; possession continuing along that frontier remains wronging.

If the applicable filters and this limited admission leave no eligible option, Stage 4 stops before comparison and Necessity does not apply.

Pairwise relations and Stage 4 outputs

A comparison rule can establish relations between particular options without settling the eligible set. X strictly ranks above Y where the governing rule positively establishes that pairwise relation. X and Y are in affirmative equality where the rule positively treats them as equal. Mutual failure to establish strict superiority does not establish equality.

If the applicable entry filters leave no eligible option, Stage 4 stops before comparison. This structural result is not a comparison output, a forced set, or a permission to reintroduce an ineligible option. Comparison applies only to a nonempty eligible set.

A complete Stage 4 output is a report over the entire eligible set, not necessarily a complete ordering. It records the positive relations established by the governing rule and whether those relations establish a top option or top set.

Stage 4 reports one complete output:

1. Uniquely best option. X is uniquely best only where the governing rule positively ranks X above every other eligible option.
2. Affirmative co-best set. Two or more options form an affirmative co-best set only where the co-best-set validity discipline certifies the complete positive top set.
3. Failure to establish a top option or top set. Where neither result is established, the output preserves any positive relations already established and records the source of the remaining failure. Evidential uncertainty remains distinct from missing comparison theory.

A pairwise result remains pairwise. If A ranks above B while the relation between A and eligible option C is not established, A is not uniquely best and {A} is not a certified top set. Stage 4 records failure to establish a top option or top set and identifies the A/C gap.

The proposed pairwise non-reversal constraint holds that, within the third output, a fallback may not select Y over X where X strictly ranks above Y and the established strict relations can be jointly honored. This preserves positive comparison results without supplying a general maximality rule. Where the strict relations cannot all be honored, their incompatibility remains recorded; this discipline supplies no cycle breaker, and an operational selection does not revise the relations.

Co-best-set validity discipline

A comparison may report a set of eligible options as affirmatively co-best only where all eight conditions hold:

1. Identified rule. The governing rule is stated and its support status remains attached to the output.
2. Complete rule-relative profiles. The Stage 3 profiles contain every established claim, satisfaction, gravity, procedure, exercise, reliance, feasibility, and other input that the governing rule legitimately uses. Stage 4 receives the grounded claim roster and any conditional maps required by unresolved Stage 1 records; it does not readjudicate claim admission.
3. Plural positive top-set closure. The reported set contains at least two eligible members. The rule positively establishes equality between every pair of members, ranks every member above every eligible nonmember, and includes every eligible option positively equal to a member. Mutual non-superiority is insufficient.
4. Set coherence. Equality is reflexive, symmetric, and transitive within the reported set. Incoherent pairwise equalities may be recorded without being represented as one co-best set.
5. Cycle exclusion. No unresolved strict cycle includes a member or eligible alternative whose relation is required by the top-set condition. This condition does not supply a general cycle theory or breaker.
6. Option individuation and representation invariance. Options are individuated by their complete claim-relevant consequences and relations. Splitting one fully specified option into duplicate labels does not alter the output. An added option that changes chances, reliance, procedure, feasibility, or another mapped claim incident is not a duplicate.
7. Relevant menu dependence. An addition or removal alters the set only through an identified change in feasibility, consequences, or mapped claim incidents. Bare relabeling and strategically inserted copies do not alter it.
8. Explicit evidential treatment. The output states how uncertainty was handled. The governing rule may incorporate evidence, risk, or probability and return the set at the threshold stated by that rule. Alternatively, the eligible set and complete output may remain invariant across specified input descriptions or ground-status branches consistent with the evidence under a stated admissibility rule. Robustness across every admissible description is a candidate evidential standard, not a universal requirement. If neither route establishes the set, the output fails to establish a top option or top set through the recorded uncertainty.

Under CBV, all eight conditions jointly suffice to certify an affirmative co-best set under a separately warranted comparison rule. If any condition fails, CBV does not certify the set. This rule does not claim that no rival certification discipline is possible. CBV is one jointly proposed and jointly refusable output-validity discipline; it does not supply the comparison rule, establish holder-relative comparative vindication, create permission or requirement, or authorize a selector.

Relevance and silencing

An unsatisfied claim counts against a graver claim only when it is of comparable seriousness: its holder could intelligibly be told, pairwise, that their claim was of the same order as the graver one it is being weighed against. Claims below that relevance band are silenced against the graver claim. No number of them combines to defeat it.

The band is anchored pairwise, to the competing claim — not to whichever claim the selected option happened to satisfy. The distinction matters and the discipline takes a side on it. Anchoring to the outcome would be circular, since which claim prevails is what the comparison is for; and in a three-way conflict an outcome anchor would let the presence of a very grave third claim silence a claim that is of the same order as the one it actually competes with. Relevance is therefore assessed between each pair, and a claim silenced against one competitor is not thereby silenced against another.

Silencing is proposed. It preserves the intelligibility of the answer to the party with the grave claim. The Holder's Good supplies coarse gravity factors but no interpersonal metric that fixes the band's placement, and the framework does not supply one. What it requires instead is that the placement be declared: for a recurring conflict type, the gravity frame that sets which claims are of comparable order is adopted in advance, publicly, and contestably, before the case that will turn on it. Each unsatisfied claim's classification is then determinate relative to that declared frame. Two analysts who reach opposite classifications on the same facts have supplied different frames; the discipline's demand is that the frame be stated and defended, not that a metric decide it. A frame chosen to fit a favoured outcome in the instant case, or left undeclared so it can be adjusted after the facts are known, fails this discipline.

Counting claims of the same order

Where options leave claims of the same order unsatisfied, the framework's defended rule prefers the option that fails fewer distinct parties. This is counting, not summing. It asserts neither an aggregate holder nor cross-person compensation, and the answer to a losing party cannot cite a total as though the total were owed something.

The rule takes one side of a live dispute. A lottery or equal-chances view can preserve the separateness of persons while denying that a larger number should decide among same-order claims. Rejecting counting therefore reopens these cases; it does not establish aggregation.

Counting does not depend on the proposed silencing rule. Rejecting silencing removes the categorical treatment of sufficiently lesser claims but leaves this same-order rule available; it does not select unrestricted aggregation or any other replacement account of relevance.

The open cell

Where one graver claim faces more numerous claims that are lesser but remain inside the relevance band, this discipline supplies no exchange principle. Numbers may sometimes prevail, but no formula here converts their claims into the graver one. Any imported rule that resolves the case must remain expressible to each losing holder without invoking an aggregate claimant.

This is a gap in ranking, not in classification. The framework does not select among band-clearing claims that conflict; it imports that ranking, declared in advance under the discipline above. But whatever the imported rule selects, a band-clearing claim it leaves unsatisfied without a positively established comparative vindication is presumptively a wronging, owed violation-grade residue. So the framework declines to say which option to take here while still saying what is owed to whoever loses. Pre-commitment does not prove an imported rule correct; it makes the judgment visible, reviewable, and less available for opportunistic adjustment in a favoured case.

Unresolved comparisons

An unresolved comparison is the complete Stage 4 output that fails to establish a top option or top set. It may preserve positive pairwise relations while recording the evidential, criterial, or rule-level source of the set-level failure. The defended default is the eligible option that minimizes irreversible foreclosure across affected parties, subject to the proposed non-reversal constraint where established strict relations can be jointly honored. Delay and continuation must appear in the map as options with their own effects; delay consumes time, while the status quo may entrench the burden under dispute.

If the foreclosure default does not discriminate and time requires a choice, selection passes to fair procedure under Standing Answerability. The institution must account for burdens imposed while a matter is pending. Uncertainty about claim gravity remains in the result, and Risk governs probability within established claims and the conditional candidacy branches supplied by Genuine Reason. Unresolved factual, recognized-ground-boundary, and candidate-new-ground records remain indexed to Stage 1; they are not converted into probability-weighted claims. An actor may not resolve either kind of uncertainty by adopting whichever description best serves its preferred option. Neither the default nor fair procedure supplies comparative vindication or establishes a ground. Where surviving strict relations cannot be jointly honored, their interaction with the fallback remains unresolved rather than being erased by the selection.

A comparison battery

Case: Compare a technician being electrocuted by equipment whose shutdown ends a broadcast delighting millions; rescuers able to save five people or one from the same peril; and a health budget able to prevent one death or two hundred cases of paraplegia.

Verdict: The broadcast is stopped because the spectators' claims fall below the declared relevance band; being silenced, their loss is a justified override, not a wronging. The five are selected under the defended counting rule, while the equal-chances rival remains coherent. The death-versus-paraplegia case occupies the open cell: the framework supplies no exchange principle to rank the one death against the two hundred paraplegias, so that ranking is imported under a declared frame. It classifies nonetheless — whichever group the imported rule leaves unsatisfied without a positively established vindication is presumptively wronged and owed violation-grade residue.

Machinery: Silencing excludes weak provision claims at any multiplier without constructing a total, and under the declared frame the gravity gap is itself the answer to the silenced holder. Counting applies only among claims of the same order and counts distinct holders rather than summed welfare. In the open cell the imported rule must remain expressible to the losing holder without an aggregate claimant; the framework supplies the classification, not the rule.

Cost: The final case is where a complete comparison method would supply an exchange principle, and this discipline does not; it imports the ranking under a declared, contestable frame. What the framework adds is not the ranking but the classification of whoever loses.

Limits

• No interpersonal metric fixes the relevance band; the discipline requires the placement to be declared in advance and contestably, and makes classification determinate relative to that declared frame rather than delivering the frame.
• The discipline does not say who may declare the frame. This is not only the missing metric: silencing is the route by which an unsatisfied claim is positively classified a justified override, and declaring a frame that places foreseeable losers below the band reduces what the declaring party owes. The framework applies a beneficiary bar at other decisive inputs — a party whose costs fall if the holder's good is described one way may not certify that description — and does not apply one here. The question of a declarer condition is therefore open, and the corpus stands in a live debate about it: accountability-for-reasonableness answers the interested-decider problem by process rather than by disqualification, and declines the advance explicitness this discipline requires.
• The graver-versus-more-numerous open cell lacks an exchange principle; the ranking is imported, while the classification of whoever loses is not.
• Counting same-order claims is defended rather than derived; the equal-chances rival remains coherent.
• Foreclosure-minimization is a defended default, not a consequence of uncertainty itself. Status-quo deference is its named rival.
• The proposed non-reversal constraint preserves established strict pairwise relations where they can be jointly honored. The discipline supplies no general maximality rule or cycle breaker where they cannot.
• Participant-scale option scope is proposed. Its invariance, authority, operative-role, and coordinated-option conditions require evidence; replaceability alone cannot satisfy them.
• Bounded-dismantling admission is proposed. Refusing it excludes every path that itself still bears current possession at the operative level under comparison; participant options governed by the separate scope rule remain available where they do not themselves carry that possession.
• The discipline constrains judgments. It does not supply the empirical knowledge needed to describe options or forecast their effects.
• Stage 4 does not establish grounds. Conditional branch comparison can preserve a live Stage 1 question, but neither a comparison output nor its fallback may turn that question into a claim or a clean absence.

Related pages

• Adequate Justification
• Risk
• Necessity
• Open Questions