Commensurability

The deepest technical question in ethics is whether different morally relevant facts can be placed on a common scale. If yes, in some strong sense, then ethics reduces to arithmetic. Just find the option with the higher number, then choose it.

If no, in some strong sense, then comparison between options of different kinds becomes impossible, and "Better is the least-closing path" loses its meaning.

Modal Path Ethics has been operating in a middle position across the entire corpus without having named it. The framework rejects scalar collapse, accepts structured comparison, treats moral remainder as residue of structural multiplicity, and acknowledges genuinely hard cases as features of the moral field rather than failures of the framework. This article makes that position explicit.

It also engages with Ruth Chang, who has been working in adjacent territory for almost thirty years already and whose vocabulary pleasantly turns out to map cleanly onto Modal Path Ethics' structural moves now that I have caught up. The framework therefore owes her a public engagement.

Why This Matters a Lot.

Commensurability is the conceptual move that makes weighting do any work.

If the weighting variables (severity, irreversibility, breadth, centrality, asymmetry, distribution) collapse to a single scalar, then Modal Path Ethics just is utilitarianism in different clothes, and all the standard utilitarian problems return: the utility monster, the repugnant conclusion, the transplant surgeon, the trolley arithmetic objection.

If the weighting variables are genuinely incomparable to one another, then the framework can't actually compare options any more, and "Better" becomes empty.

The framework's whole architecture therefore depends on a position between these. Get this wrong and either the framework reduces to one of the existing positions or it loses its action-guiding capacity.

So, this one is the load-bearing question, while also being one of the hardest out.

The Two Failure Modes

Strong commensurability would be the first; the welfare-aggregation move. All morally relevant facts can be placed on a single scalar measure. Welfare sums across persons. Better choice equals higher number. The view has technical elegance and yields determinate answers in every case.

It also generates the disastrous problems the framework has spent its corpus arguing against.

The utility monster gets everything because concentrated welfare aggregates higher than distributed welfare. The repugnant conclusion follows because total welfare can always be increased by adding more lives barely worth living. The transplant surgeon should kill one to save five because the welfare numbers say so. Trolley arithmetic generalizes to cases where it stops being defensible.

The deeper problem isn't any individual counterexample.

It's that the summing operation loses information that actually matters. Severity gets averaged with breadth. Irreversibility disappears into a number alongside reversibility. Asymmetric burden distribution becomes invisible once everything is converted to a common scale. The structure of how harm actually contracts future-space gets compressed into a measurement that no longer carries the structure.

Hard incommensurability is the second failure mode; the position that morally relevant facts of different kinds genuinely cannot be compared at all.

Apples and oranges, full stop. No scale connects them.

This view honors the phenomenology of moral life (the felt sense that some choices are between options that don't reduce to one another) but also often ends in paralysis. If literally no comparison is possible across kinds, then there is no fact of the matter about which option is better, and the framework's normative capacity dissolves. We get descriptive structural analysis but no action guidance.

The framework clearly needs a position between these.

What Modal Path Ethics Actually Does Instead.

The weighting variables (severity, irreversibility, breadth, centrality, asymmetry, distribution) are not summands.

They are not pieces of a scalar measure that get added together to produce a number. They are structural features of contraction that get assessed independently along multiple dimensions, and the comparison between options is a partial-order operation rather than a total-order one.

Concretely, three things then happen:

1. When option A is worse than option B on every weighting variable that bears on the case, A is clearly worse. The comparison is determinate. This is what dominance reasoning looks like, and it covers a substantial fraction of real moral cases. The 1904 St. Louis Marathon analysis works this way. Sullivan's organization fails on severity, irreversibility, breadth, centrality, asymmetry, and distribution simultaneously, which is why the case is so unambiguous. The Therac-25 case works the same way. So does Joe Martin's case, and Sydney's, and most of the applied-case corpus in general. Dominance reasoning doesn't require a scalar; it requires the structural features to align.
2. When option A is worse on some variables and better on others, the comparison is genuinely harder, and the framework's honest answer is just that it is genuinely harder. Trade-offs between severity and breadth, between irreversibility and asymmetry, between centrality and distribution don't always resolve cleanly for convenience. The framework doesn't pretend they do. It admits hardness where hardness is real.
3. When the variables that distinguish two options are weak across the board, the cases are roughly equivalent on the structural analysis, and the framework can call them roughly equivalent without forcing a tiebreaker. Two slightly different paths through a contracted field, both least-closing in their respective topologies, might just be co-Better. This is fine. The framework doesn't need every case to have a unique winner, or invent some arbitrary tiebreaker.

This is structural pattern recognition rather than arithmetic. This is closer to legal precedent reasoning, or to the way a clinician compares treatment options on multiple incommensurable axes, than to summing utilities. It yields clear comparisons in most actual cases because in most actual cases the weighting variables align rather than conflict. It also yields honest acknowledgment of difficulty when the variables genuinely trade off.

The framework is committed to this being a real feature of the moral field rather than an epistemic limitation. Some cases are hard not because we don't have enough information, but because the moral facts in those cases involve multiple structural features that don't reduce to one another. The hardness is a fact about the field, not just about us.

Again: The Weighting Variables Are Not Summands.

This is the central technical claim. Let me be more precise about what it means.

If the weighting variables were summands, severity could be exchanged for breadth at some rate. A severe contraction of one locus would be morally equivalent to a less severe contraction of N loci, for some N that the rate-of-exchange determines. The framework would need to specify that rate, defend it, and apply it consistently across cases.

It can't do any of this without becoming utilitarianism. Once you commit to a rate of exchange between severity and breadth, you've now committed to a scalar. The framework's distinctive moves all depend on refusing this commitment.

But, and this is what makes the position non-trivial, refusing scalar exchange doesn't mean refusing all comparison.

Severity and breadth can both be assessed in any given case. Two options can be compared on severity. They can be compared on breadth. When one option dominates on both, comparison is determinate. When they trade off, comparison is genuinely hard. The structural features remain real even when no rate of exchange between them is specified.

The corpus's working examples illustrate this. The Pascal Mugging article explicitly refuses to give a number that would let the mugger's claimed enormous future cash out against a small present cost. The Cluelessness article refuses to fold deep uncertainty into expected-value arithmetic. The Lifeboat case identifies multiple structural variables (centrality, asymmetry, repairability) that resist conversion to a common scale even as they yield clear comparison in the central case.

The framework treats the variables as multi-dimensional structure of a real moral field, not as components of a measurement. That commitment is what makes the rest work.

How Comparison Works in Practice.

The practical question is: when does the framework yield a determinate answer, and when does it admit hardness?

Dominance cases. When one option is worse on every relevant weighting variable, comparison is determinate. This covers substantially more cases than a reader skeptical of the framework's middle position might expect. Most real moral choices aren't between options that trade off on every variable. They're between options where one is clearly worse on most or all of them. Sullivan's organization of the 1904 Marathon dominates negatively on every variable that matters. So does the Therac-25 manufacturing decision to remove hardware safeguards. So does the institutional response to Joe Martin's communication. The framework doesn't need scalar arithmetic to call these cases what they are.

Partial dominance cases. When one option is worse on most variables and equivalent on the rest, the comparison is determinate by a slightly weaker form of the same reasoning. The framework doesn't need every variable to point the same way; it needs no variable to point the other way.

Trade-off cases. When variables genuinely conflict, when option A is worse on severity, option B is worse on breadth, and neither is broadly worse, then the case is just genuinely hard. The framework's honest answer is that it is genuinely hard. It does not force a verdict by smuggling in a rate of exchange of its preference. It identifies the trade-off as real, explores what each option preserves and closes, and acknowledges that reasonable agents in the field may differ.

Roughly equivalent cases. When the differences across variables are all small, the cases are roughly co-Better, and the framework can say so without insisting on a unique winner.

This is action-guiding without being action-forcing. In most cases, the framework yields clear answers. In hard cases, it acknowledges hardness. In equivalent cases, it accepts equivalence. That's the right behavior for a moral framework working in a structurally multidimensional field.

Moral Remainder as Evidence for Structural Multiplicity.

If strong commensurability were correct (i.e. if all moral content reduced to a scalar) there would be no moral remainder to speak of. The chosen option would have captured the moral content fully. Nothing would be left over to mourn.

But moral remainder is a very real phenomenon.

Agents who make the right choice in tragic dilemmas still feel that something was lost. They feel that they owe something back. They recognize that the unchosen alternative had moral content that the chosen one could not preserve. Bernard Williams developed this concept formally; the experience predates his analysis by all of human history. Anyone who has stood at a hospital bedside, or made a call between two obligations that couldn't both be honored, knows this.

The existence of moral remainder is data. It tells us that the moral field contains structural content that survives any single comparison operation. The unchosen option had something the chosen one didn't, and that something wasn't just a smaller utility number; because if it had been, the choice would have been the wrong one. It was something the chosen option could not preserve regardless of how the comparison resolved.

That is the structural multiplicity of the moral field showing up phenomenologically.

Frameworks that deny moral remainder, like some versions of utilitarianism, have to dismiss this data as confused to survive. They have no choice but to treat the felt sense of loss as a misfiring of moral psychology, something to be educated away by its clearer thinking about expected utilities.

This dismissal is not credible.

The felt sense of moral remainder is not confusion. It is direct contact with the structural multiplicity of moral facts.

Frameworks that accept moral remainder need a structure to hold it. Modal Path Ethics has that structure. The weighting variables describe multiple dimensions of contraction, and a choice that minimizes contraction on net can still leave some dimensions un-preserved. The remainder is the un-preserved content. The variables genuinely don't reduce to one another, and the remainder is the experiential evidence that they don't.

This makes moral remainder part of the evidential case for the framework's middle position. It's not just some phenomenological observation that needs to be accommodated. It's data that supports the partial-order analysis against scalar reduction.

Ruth Chang and Parity.

Ruth Chang has been working on commensurability and incomparability since the late 1990s. Her core move is to argue that the standard trichotomy of comparison (better, worse, equal) is insufficient to capture how options actually relate in cases of hard choice. She introduces us to a fourth value relation: parity. Two items are on a par when they are evaluatively comparable but neither is better than the other and they are not equal in any precise sense.

Her famous example involves a choice between two careers, for example, banking and philosophy. They aren't equally good. They aren't precisely tied. Yet neither is clearly better. They are different in ways that matter, and the difference doesn't resolve into a verdict. Chang's argument is that this isn't because the options are incomparable. It's because they're on a par. They are comparable but not orderable in the standard ways.

She develops this out cleanly with the small improvement argument. If A and B are equally good, then A+ (a slightly improved version of A) should be better than B. But in cases like banking-vs-philosophy, slightly improving the philosophy career doesn't necessarily make it better than the banking career at all. The improvement doesn't move the comparison the way it should if the original options were equal. This shows they weren't equal. They were on a par.

Chang has also engaged the repugnant conclusion directly. Her 2022 essay in the Parfit memorial volume Ethics and Existence uses parity to show how population ethics can avoid the repugnant conclusion without committing to either total or average utilitarianism. The argument shows the breadth of what parity can do, not just for personal hard choices but also for the deepest puzzles in population ethics.

The conceptual alignment with Modal Path Ethics is, on review, substantial.

Chang's parity describes the structural feature this framework's weighting analysis identifies as the trade-off case. When weighting variables conflict (option A is worse on severity but better on breadth) the options are on a par in Chang's sense. They are comparable (we can identify what each preserves and closes) but not orderable in the standard better/worse/equal sense.

Chang's vocabulary names precisely what Modal Path Ethics has been describing structurally. I now wish I had consumed it fully before writing the book.

The framework owes her a real engagement now, and the corpus would benefit from absorbing her vocabulary in the trade-off cases where the language of "on a par" is far more precise than my language of "genuinely hard."

What Modal Path Ethics Adds to Chang, and What It Borrows.

What this framework adds to Chang's work is a structural grounding. Chang's parity is a fourth value relation that emerges from analysis of value comparison, defended on the strength of the small improvement argument and the phenomenology of hard choice.

This view is conceptually rigorous, but doesn't ground itself in a particular ontology. It's compatible with various positions about what values are and what makes things valuable.

Modal Path Ethics provides one such ontology. The reason some options are on a par rather than orderable is that the moral field has multiple structural dimensions of contraction, and these dimensions don't reduce to one another. Severity and breadth are real features of how reachable future-space closes, not summands of an underlying utility. The trade-off cases that Chang identifies as parity cases are exactly the cases where the structural variables conflict. Chang's phenomenological observation gets a metaphysical grounding in this framework.

What Modal Path Ethics could borrow from Chang is the small improvement diagnostic. This argument provides a clean test for whether two options are actually equal or on a par. In ambiguous cases, ask: would a small improvement in one option make it clearly better? If yes, they were equal and the improvement broke the tie. If no, they were on a par and the improvement isn't enough to move them across the structural multiplicity that separates them. This specific diagnostic could sharpen Modal Path Ethics' analysis of borderline cases very cleanly; distinguishing genuine equivalence from parity in particular cases.

There is also the broader claim Chang makes: that "comparativism" is foundational to rational choice, regardless of which moral theory one holds. She argues that whatever your background ethics may be (consequentialist, deontological, virtue-theoretic, contractualist), you should always first be a comparativist about practical reason.

Modal Path Ethics' structural-comparison move is consonant with this. The framework treats Better-as-least-closing as a genuine comparative judgment that depends on real structural facts. Chang's broader project legitimizes this kind of move philosophically.

The relationship between this framework and Chang is ultimately one of mutual reinforcement. Her parity is the value-theoretic correlate of Modal Path Ethics' trade-off cases, and this framework's structural analysis is the metaphysical grounding for why parity is a real feature of value rather than a measurement artifact. Each makes the other more defensible by existing. Both are working on the same problem from different directions.

Distinguishing From Other Positions.

Vs. utilitarianism. Modal Path Ethics rejects scalar collapse. The weighting variables don't sum. The repugnant conclusion doesn't follow because we don't aggregate welfare across persons in the first place. The utility monster doesn't follow because asymmetric burden distribution is morally relevant on its own terms, not as a quantity to be summed against the monster's gains. The transplant surgeon case doesn't reduce to arithmetic because the structural feature being violated (the trust-field of medical care) isn't on the same scale as the survival numbers. The framework keeps the action-guiding structure of consequentialism without the scalar-aggregation move that generates its many problems.

Vs. pluralism (Ross-style prima facie duties). This is the closest neighbor in spirit. Ross argued that there are multiple independent moral duties (fidelity, reparation, gratitude, justice, beneficence, non-maleficence, self-improvement) that don't reduce to one another. Modal Path Ethics accepts the multiplicity but grounds it differently. The weighting variables are derived from the structural ontology of contraction. They correspond to features of how reachable future-space gets contracted, not to a list of intuited duties. This grounding answers the standard pluralism objection ("why these duties and not others?") with a structural rather than intuitive answer. The plurality is real because the structure of moral fields really is multidimensional.

Vs. lexical priority (Rawls's first-principle-over-second). Modal Path Ethics has approximate lexical features. Severity often dominates breadth. Irreversibility often dominates reversibility. Centrality often dominates peripherality. But there's no absolute lexical ordering. Any of the variables can be outweighed by sufficient pressure on the others. The framework's lexical tendencies are statistical and structural, not absolute and rule-based.

Vs. Chang's parity. Strong conceptual alignment with structural addition, as discussed above. Thank you, Chang.

Vs. virtue ethics. Modal Path Ethics accepts substantially more comparison than most virtue-theoretic positions, while remaining more structurally grounded than virtue ethics typically is. Where virtue ethics often treats comparison as a matter of practical wisdom resistant to systematic articulation, Modal Path Ethics articulates the structural variables that practical wisdom is implicitly tracking.

Vs. contractualism. Scanlon's "what we owe to each other" handles intersubjective comparison through justifiability to all affected. This works for some applications but struggles with non-subjects (pre-life harm, ecosystems, future generations whose existence depends on the choice). Modal Path Ethics' structural grounding handles those cases because it doesn't require justifiability to subjects who exist.

Practical Implications.

The framework's commensurability position has consequences for how moral reasoning should actually proceed.

In dominance cases, the framework yields clear answers without requiring agents to compute scalar comparisons. Most real moral choices are dominance cases. Most agents in most situations don't face genuinely tragic dilemmas, they face situations where one option is worse on most or all dimensions, and the work is identifying which option that is rather than performing arithmetic between competing utilities. The framework's structural analysis is well-suited to this work.

In trade-off cases, the framework acknowledges hardness rather than forcing a verdict. This is morally important. Frameworks that force verdicts in genuinely hard cases produce false confidence and burn moral remainder unacknowledged. The framework's willingness to say "this is genuinely hard, and here is why" preserves the agent's contact with the structural reality of the choice.

For agents in trade-off cases, the framework offers tools without offering algorithms. Identify the structural variables. Assess each option on each variable. Look for dominance. If none, identify the trade-off precisely. Consider what each option preserves and what each closes. Recognize that reasonable agents may resolve the trade-off differently. Choose, knowing that moral remainder is real and acknowledging it. This is more than utilitarianism offers (which forces a verdict and dismisses remainder) and more than hard incommensurability offers (which abandons the agent to arbitrary choice). It is structured guidance through cases that resist mechanical resolution.

For institutions, the framework offers a vocabulary for honest disagreement. The Smogon Tier Council's reasoning over the sleep ban, which the RBY UU 2020s article documents extensively, is a working example of structured comparison without scalar collapse. Council members assessed multiple structural variables, weighed them differently, voted, implemented, observed consequences, and revised. They didn't pretend their reasoning was scalar. They didn't pretend the decision was forced. They did the work of structured comparison and were willing to be wrong. This is what institutional moral reasoning under Modal Path Ethics' commensurability position looks like in practice.

What This Isn't, Though.

It is not relativism. The structural features are real. Where they yield comparison, the comparison is real. Where they don't, the field genuinely contains multiple incomparable structural dimensions. None of this depends on the agent's preferences or the community's conventions. The structure is a feature of the moral field, accessible to careful structural analysis, defensible to other agents through structural grammar.

It is not skepticism about moral judgment. The framework yields determinate judgments in most cases. The cases where it admits hardness are cases that any honest framework should admit are hard. The framework's hardness-acknowledgment is calibrated to the actual structure of the moral field, not to a generalized doubt about moral knowledge.

It is not wishy-washy.

The framework is more demanding than utilitarianism in many cases. It refuses arithmetic shortcuts that utilitarianism would accept, it identifies harms (asymmetric burden, structural distortion, false repair) that utilitarianism systematically misses, it grounds its moral claims in structural features of extance rather than in subjective preferences. The middle position on commensurability is not a softer position in any way. It is really just a more accurate one.

Closing.

The line Modal Path Ethics draws on commensurability is structural and specific.

It accepts comparison. It rejects scalar collapse. It accepts that the weighting variables describe multiple dimensions of contraction that don't reduce to one another. It accepts that some cases yield determinate comparison through dominance reasoning, and that other cases involve genuine trade-offs that don't fully resolve. It treats moral remainder as evidence that the moral field has structural multiplicity.

It accepts Chang's parity as the value-theoretic correlate of its trade-off cases and offer structural grounding for why parity is a real feature of value. It refuses to force verdicts in genuinely hard cases and refuses to dissolve verdicts in genuinely tractable cases.

The framework is action-guiding without being action-forcing. It is comparative without being arithmetical. It is honest about hardness without retreating into incomparability.

This is the center line position that makes the rest of Modal Path Ethics work. The corpus has been operating in it across every article. Comparison is real. Comparison has structure. Comparison is not arithmetic. That is the line.