Stake-Geometry
← The library
Technical corpus, written for specialists. For the plain-language version of these ideas, start with the framework.
Applications

Service, Status, and Expression

Working draft This public copy has not yet had its final author pass; wording may still move.

A Worked Case: May a Business Refuse a Same-Sex Wedding on Religious Grounds?

What this document is

The foundational documents established where value comes from (aperturality), how it is weighted (stake-geometry), how privately mattering apertures produce shared values (network universality), and how that content sorts into tiers (tiered values). Applied Ethics connected the structure to the moral content traditions already carry; Relations Between Apertures gave the structural account of what a relation between two mattering-systems is; Political Philosophy did the parallel work for collective life. This piece runs all of that against a single contested case: a bakery whose owner objects to same-sex marriage on religious grounds and wants to refuse service to a gay couple for their wedding.

The case is worth a standalone treatment for two reasons. First, it is the cleanest available test of whether the framework can adjudicate a genuine conflict between apertures — not a case where one party is plainly collapsing another, but one where each side has a real stake and the question is which the state should weight. Second, the popular framing ("should they be allowed to refuse?") hides a structural ambiguity that the framework dissolves the moment it is named. The verdict is not one verdict. It is a clean one on part of the case and a deliberately unresolved one on the rest, and the discipline of saying which is which is the point.

Two methodological notes, because the case invites two specific errors. The first move below — locating the owner's conviction in the tier structure — is derived from the core documents directly, not borrowed from the incest analysis in moral_sentimentalism; that document runs a parallel test on different content, and is cross-referenced where the structures rhyme, but the anchoring here is Tiered Values, Stake-Geometry, and Network Universality. And because the word "tier" carries three different jobs in this project, the case opens by saying which one is in play.

A note on which tiers are in play

The project uses tier language in three places, and conflating them is exactly the kind of error this case is prone to. To keep them apart:

This document follows the standing convention: no bare "Tier N." Every use is qualified by what sits there — a Tier 1 norm, a Tier 2 principle, a Tier 2 good — and the emotion model's axis is named only as a processing level.

With that fixed, here is where the case's moving parts land before any argument:

First move: locating the conviction

Before the liberty question can be asked well, the framework wants to know what the religious conviction is in structural terms. The apparatus for this is in Tiered Values directly.

What separates a Tier 1 norm from a Tier 2 principle is not its sincerity, intensity, or seriousness. Tiered Values is explicit: "What makes a value Tier 1 rather than Tier 2 is not its intensity or seriousness. It is its structural anchoring." A Tier 2 principle is anchored in features any aperture must have to count as one; a Tier 1 norm is anchored in the specific composition and history of a particular network. The test the document supplies is the alien-entrant test: imagine a stake-bearer of wildly different biology and psychology entering the network — still an aperture, still capable of recognition. Does the value apply to them, and would they have reason to endorse it about themselves?

Run that test on "marriage should be between a man and a woman, and a same-sex union should not be solemnized." There is nothing in a genuine aperture as such for the norm to grip. It needs human sexual dimorphism, a particular reproductive arrangement, and a specific religious history to mean anything at all. Strip those and the norm has no content. By the document's own criterion, that is the signature of a Tier 1 norm dressed as more — the same diagnosis Tiered Values gives the dietary and ritual commandments, and the same one Network Universality gives the parts of scripture that had to be quietly reinterpreted as networks grew.

But the case is sharper than ordinary Tier 1 parochialism, because Tiered Values names a specific way a Tier 1 norm can be not merely local but wrong: when the network "has constituted itself by exclusion — by treating some apertures as outside the negotiation when they should not have been." The wrongness of Athenian slavery, on the framework's analysis, "is not that the Athenians failed to track a Platonic truth; it is that their agreement was structurally incomplete in a way they could have been made to see and were not." A norm that withholds a central form of recognition from a class of apertures on the basis of who they are is exclusion of exactly this kind. Network Universality describes its correction as the engine of moral progress: "Slavery becomes wrong when enslaved people are recognized as full apertures with full stake-geometries." The expansion of marriage to same-sex couples is, in the framework's terms, the same motion — a Tier 1 exclusion reformulated toward the Tier 2 principle that apertures symmetrically situated are to be recognized symmetrically.

And the object of the conviction confirms the direction. A committed partnership between two adult apertures is, in the framework's foundational vocabulary, stake-elaboration. Aperture Ethics lists love among the dilations: "a second aperture becomes visible through the first; reality deepens." Stake-Geometry says love "deepens existing stakes and adds new ones (the beloved's stake-geometry becomes partially internalized)." There is no stake-collapse anywhere in the union itself for a prohibition to be protecting against. The norm forbids an elaboration and withholds recognition; on the framework's own categories it is doing the negative thing, not preventing one.

So the framework's metaethics judges the conviction mistaken, and not marginally. Network Universality puts the point at its bluntest: once recognition is available, "treating the other's stakes as nothing starts to look less like a defensible philosophical position and more like a willful refusal to see what is there."

One thing this first move does not establish, and the rest of the case turns on the distinction. That a conviction is a defective Tier 1 norm does not make the holder's stake in not being coerced against it unreal. A false belief is still a real feature of a stake-geometry, and the framework prizes the aperture's authority over its own stakes and warns repeatedly against the state imposing geometric simplification (Stake-Geometry names totalitarianism as exactly "the state demands that citizens' stake-geometries align with a single permitted structure"; Political Philosophy makes anti-collapse rights against the state a Tier 2 commitment). The wrongness of the conviction lowers the weight the refusal can claim. It does not zero it out. Holding those two facts together — the norm is wrong, the stake in not being compelled is real — is what keeps the rest of the analysis honest.

The question is relational, and it splits

"Should the business be allowed to refuse?" is not "is the conviction correct." It is a question about coercion between apertures and about which geometry the state weights when their stakes conflict — which is the subject matter of Relations Between Apertures and Political Philosophy. Posed there, the single word "service" comes apart into two structurally different acts, and the framework gives them different verdicts.

The relevant primitives are the ones Relations Between Apertures names. Recognition — whether each party registers the other as a stake-bearer — is the axis the couple's complaint lives on. Effect on geometries — what the relation does to each side — is where the morality of the transaction is most legible. And the power modifier matters: a public-facing business in a market has a capacity asymmetry over an individual customer, so the moral weight of its conduct scales accordingly. With those in hand:

Case A: status-based denial of a standard good

The owner refuses to sell the couple a thing the shop sells to anyone — an off-the-shelf cake, ordinary goods — because of who they are.

This is the clean verdict, and it is not permitted. Structurally it is a recognition failure: a refusal to register the couple as full participants in the network of commerce, which is the constitution-by-exclusion wrong from Tiered Values enacted in miniature. The non-discrimination principle that forbids it is a Tier 2 principle — recognize other apertures, and treat symmetrically situated apertures symmetrically — given institutional armor by the Tier 1 institution of public accommodation, the "entrenched Tier 2 content" pattern from Political Philosophy.

The conflict here is lopsided. The owner's stake is thin to vanishing: selling a standard good to a couple compresses no dimension of their geometry, because the objection attaches to the customers' identity, not to any act the owner must perform. The couple's stake is a real recognition-collapse — being told, in the ordinary transactions of public life, that they are not full members of the network others belong to without question. And the slow structural channel the framework's no-victim analyses rely on applies with force: a regime that licenses status-based exclusion erodes the recognition-fidelity of the whole network over time, even where any single couple can shop elsewhere. The substitutability of the cake does not touch the wrong, because the wrong is in the exclusion, not the cake.

Case B: compelled creation of a bespoke expressive artifact

The owner will sell the couple anything in the shop but refuses to design and create a custom cake — scripted, decorated, and built to celebrate this specific wedding — because doing so is experienced as participation in, and endorsement of, the event.

This is the genuinely hard case, and the framework does not deliver a knockout. Now both sides hold a real stake.

On the couple's side, the recognition-harm from Case A is still present but partly transformed. The custom artifact is substitutable in one sense — another baker exists — but the dignitary component is not: being singled out as the one wedding this maker will not make is a recognition harm that the availability of substitutes does not dissolve. Relations Between Apertures would note that the relation is low in substitutability on the couple's side precisely because it has been made about them.

On the owner's side there is a stake that is genuinely independent of whether the conviction is correct. Compelling any aperture to generate expressive content against a deep commitment is a constriction the framework takes seriously on its own terms — the aperture's authority over its own stakes, and the standing warning against state-imposed geometric simplification. This is the load-bearing point, and it is easy to miss because the first move established the conviction is wrong. But the compelled-expression concern does not run through the conviction's correctness. It runs through the framework's refusal to let the state conscript an aperture into producing meanings it does not hold — the same structural worry that makes propaganda and aperture pollution wrongs in Aperture Ethics and Political Philosophy, now pointed at the citizen rather than by the state. Applied Ethics gives the adjacent principle in its treatment of the Platinum Rule: respect the aperture's own authority over its stakes except where the geometry has clearly collapsed — and a mistaken religious commitment, however parochial, is not a collapsed geometry.

So Case B is a conflict between a real recognition-and-dignity stake and a real compelled-expression stake. And this is exactly the terrain Political Philosophy flags as the framework's limit. The framework "can rule fascism out and rule a high floor in" but "cannot, on its own, adjudicate whose stakes prevail" in a contested conflict; aggregation is the named unsolved problem, and it bites hardest here. The framework can specify the trade-off precisely and say which considerations are load-bearing. It cannot mechanically output a winner.

What it can defend is the narrowest expansion-stable resolution: compel access, do not compel endorsement. The state may require that the couple be sold whatever the shop sells anyone — this discharges the Tier 2 recognition principle and forecloses Case A. It need not require that an artisan produce a specific celebratory expressive work — this protects the aperture against compelled content-generation. The line tracks the distinction the real jurisprudence circles (the Masterpiece Cakeshop and 303 Creative line of cases), and the framework grounds it rather than merely asserting it: access protects against recognition-failure; the expression carve-out protects against forced geometric simplification.

Where the framework lands

The resolution is not symmetric, and two asymmetries keep it from collapsing into a general license to discriminate.

First, the carve-out lives only in the bespoke-expressive zone. The instant a refusal becomes status-based — "I will not serve gay customers," or a refusal of standard goods — it reverts to Case A and the framework forbids it outright. The expression carve-out is not a tier; it is a narrow protection for the act of creating meaning, and it does no work the moment the refusal is about identity rather than expression.

Second, the thumb is on the scale toward access even within Case B. The conviction grounding the refusal is, by the framework's own first move, a defective Tier 1 norm of the exclusionary type; the slow structural channel counts the erosion of network recognition as a real cost; and the power asymmetry of a public-facing business weights its conduct more heavily. None of this overrides the compelled-expression stake, but all of it tilts the close case toward construing the protected expressive zone narrowly rather than generously.

The honest output, then: the business may not refuse to sell a same-sex couple what it sells anyone, and the framework treats that as firm. Whether it may decline to create a bespoke artifact it experiences as expressive endorsement is a genuine conflict the framework can frame but not mechanically resolve, with its narrowest defensible answer being to compel non-discriminatory access while declining to compel expressive endorsement. Saying the case does not resolve cleanly is not the framework flinching; it is the same honesty the foundational documents practice wherever a real conflict between apertures meets the unsolved aggregation problem.

The real-world cases and how they were decided

The case is not hypothetical. Two recent United States Supreme Court decisions sit almost exactly on top of the two sub-cases above, and reading them through the framework is the sharpest available test of whether the framework's split is doing real work.

Masterpiece Cakeshop v. Colorado Civil Rights Commission (2018). Jack Phillips, a Colorado baker, declined to create a custom wedding cake for a same-sex couple on religious grounds while remaining willing to sell them other baked goods. The couple filed under the Colorado Anti-Discrimination Act, and the state commission ruled against Phillips. The Supreme Court reversed, 7–2, but on a narrow ground that left the central conflict untouched: it held that the Colorado Civil Rights Commission had failed to act neutrally toward Phillips's religion — some commissioners had voiced open hostility to his beliefs — and that the Free Exercise Clause requires the state to give religious views neutral and respectful consideration. The Court explicitly declined to resolve whether free expression or free exercise can override a public-accommodations law. A concurrence by Justice Kagan, joined by Justice Breyer, stated the two-sided structure most clearly: religious objections do not generally entitle a business to deny equal access to goods and services under a neutral public-accommodations law, but the state, in enforcing that law, may not treat the objector's religion with contempt.

303 Creative LLC v. Elenis (2023). Lorie Smith, a website designer, sought to offer custom wedding websites while declining to create them for same-sex weddings, and challenged Colorado's law pre-emptively, before any actual refusal. The Supreme Court ruled for her, 6–3, on Free Speech rather than Free Exercise grounds: the First Amendment's bar on compelled speech prevents the state from forcing a designer to create custom expressive content conveying a message she rejects. The majority stressed that Smith would serve LGBTQ customers other products and objected only to producing a particular message, casting the refusal as message-based rather than status-based, and affirmed that states may generally enforce public-accommodations laws — including protections for gay people — except where enforcement compels protected expression. The dissent, by Justice Sotomayor (joined by Justices Kagan and Jackson), rejected the message/status distinction, warned that the ruling was the first to grant a public business a constitutional right to refuse a protected class, and argued the logic would let a business define the expressive quality of its goods so as to exclude protected groups — "passport photos for white people," in the dissent's example.

Did the Court get it right, on the framework's terms?

The first thing the framework notices is that both cases landed in its Case B zone, not its Case A zone. Each involved a custom, plausibly expressive artifact — a designed wedding cake, a designed wedding website — not the refusal of a standard good. So the question on which the framework is firmest, and delivers a flat not permitted — status-based denial of ordinary goods — was never the one that reached judgment. Both real disputes sit in the framework's genuinely contested region, where it declines to name a clean winner. That alone reframes the question: the Court was adjudicating the hard case, not the easy one.

On Masterpiece, the framework's verdict is that the Court identified a real but narrow wrong and stepped around the one that matters. The neutrality holding is something the framework can affirm on its own terms: a state adjudicator showing contempt for an aperture's deep commitments is a recognition-failure committed by the state against the citizen, a pollution of the process by the very body charged with weighing both sides. The recognition principle is bidirectional, and it binds the state's posture toward the religious objector even where the framework judges that objector's conviction mistaken. But the holding answered a question — was the state neutral? — orthogonal to the framework's central one — whose stake prevails? The Court reached the prevailing party by a route that leaves the conflict unresolved. That is congruent with the framework's own admission that Case B does not resolve cleanly, but it is not the framework's analysis. The opinion closest to that analysis is Kagan's concurrence, which holds both halves together: protect access (the Tier 2 recognition and symmetry principles) and forbid the state from treating the objector's aperture with contempt (aperture fairness in the process). That two-part structure is the framework's, in constitutional dress.

On 303 Creative, the assessment is genuinely split, and the split is the framework's signature.

On outcome, the ruling is broadly congruent with the framework's narrowest defensible resolution. "Compel access, do not compel endorsement" is close to what the majority held: public-accommodations law applies and protects gay people, with a carve-out only for compelled expressive creation. The compelled-speech doctrine — the state may not force an aperture to voice a message it rejects — is structurally the framework's authority-over-one's-own-aperture principle, the bar on conscripting an aperture into producing meanings it does not hold. The framework cannot call the result wrong, because the framework itself names no mandatory different result in Case B.

On reasoning, the framework faults the Court on two counts, and on both it sides with the dissent's structural instincts without adopting the dissent's conclusion.

First, the majority treated a two-aperture conflict as a one-aperture speech problem. The couple's dignitary recognition-harm never enters as a genuine counterweight; it is compressed into "the state's interest in non-discrimination" and handled at the scrutiny stage rather than recognized as a competing stake. The framework's whole claim about Case B is that it is a conflict, and an analysis that makes only one side constitutionally visible is incomplete by construction. The observation, widely made about the decision, that majority and dissent were not even answering the same question is, in the framework's vocabulary, the diagnosis that each opinion saw only one aperture. The framework's distinctive move is to hold both in view at once — which is what lets it agree with the majority's outcome and the dissent's anxiety in the same breath.

Second, the majority left the boundary of "expressive" undefined, conceding only that hard questions would arise about which businesses qualify. That undefined boundary is precisely the framework's own first Known Weakness, and precisely where its thumb on the scale toward inclusion and its slow-structural-channel concern are meant to do their work. The dissent's "passport photos" worry is, translated, the Case A concern: that an unpoliced expressive carve-out lets status-exclusion re-enter wearing expression's clothing and erodes the network's recognition fabric over time. The framework shares that worry. Its answer is that the carve-out is defensible only if the expressive zone is read narrowly and policed actively — and the majority supplied the carve-out without the policing.

There is also a grounding difference worth flagging, because it bears on the slippery slope. The Court protected the designer through a near-categorical free-speech right that, once triggered, functions close to a trump. The framework reaches a similar place by a more qualified route: its no-compelled-endorsement protection is grounded in authority over one's own aperture, is defeasible, is weighed against the couple's recognition-harm, and carries a standing tilt toward inclusion. A weighed protection is structurally less prone to the dissent's runaway worry than a categorical one, because it does not hand the expressive objector an automatic win the moment expression is in play. The framework would predict that the categorical route is the more dangerous of the two over time — one more way it agrees with the dissent's fear while declining the dissent's remedy of upholding compulsion.

The honest bottom line: the framework cannot say the Court reached the wrong result in the only sub-case actually litigated, because the framework's own indeterminacy in Case B names no winner to have been denied. It can say the Court reached a defensible result by reasoning it regards as structurally incomplete — one-sided where the framework insists on two sides, and unpoliced exactly where the framework locates the danger. And it can say that the verdict on which the framework is least equivocal, against status-based denial of ordinary goods, was never the question that came to judgment. Both cases were the hard case. The framework's firmest verdict remains, at the Supreme Court level, untested.

What the framework can and cannot offer a court

A natural question follows from the comparison above: if the Court had had this framework when it first faced these facts, would it have reached a more comprehensive result? The answer marks the boundary of what an ethics like this one can do for a legal institution, and the boundary is worth stating because it generalizes past this case.

What the framework offers a court is diagnostic, not dispositive. Its first move forces both apertures' stakes into a single structure at once, tier-locates the conviction, and separates the access question (Case A) from the expression question (Case B) before any weighing begins. That is exactly the discipline whose absence produced the one-sidedness of the later ruling, where one stake was treated as a right and the other dissolved into "the state's interest." A court reasoning through the framework would have written a more comprehensive opinion: both stakes visible, the conflict mapped rather than half-seen.

Three limits keep that from becoming a more comprehensive verdict.

First, the framework names Case B as unresolved, and even with every stake on the table the weighing is stipulated rather than computed (the measurement weakness from Stake-Geometry). A court so equipped would reach a better-mapped indeterminacy, not a determinate answer it had otherwise missed. The actual narrow ruling already lands where the framework lands — merits unresolved — and the gain is candor: an explicit narrow settlement in place of a sideways procedural off-ramp. That is a better opinion, not a different judgment.

Second, the one concrete gain to the output is a limiting principle. The framework carries, built in, the instruction that an expressive carve-out is legitimate only if the expressive zone is read narrowly and policed actively, with a standing tilt toward inclusion. A ruling written through it would set that policing as the rule for future cases — the precise thing the eventual expression-based holding left undefined.

Third, the framework's first move is one a court is forbidden to make. Tier-locating a religious conviction as a defective, exclusionary Tier 1 norm is a judgment that the belief is mistaken, and courts are barred from ruling on the validity of religious beliefs. Only the framework's secular scaffolding can be imported — equal recognition, anti-exclusion, authority over one's own aperture — translated into the vocabulary a court may use; the tier-verdict on the religion stays in the reasoning behind a holding, never in the holding itself. Underneath this sits a structural mismatch: a court must return a binary judgment and a generalizable rule, while the framework is content to say "genuinely contested, here is the narrowest settlement." The framework's candor about open conflict is something the judicial role can adopt only in part.

The net contribution, then, is to a court's understanding of a case more than to its bottom line. Comprehensiveness of analysis is something the framework can supply on demand. Comprehensiveness of resolution is not, because in the contested zone the conflict is genuinely open, and no framework manufactures a determinate answer where the stakes themselves do not yield one.

Known weaknesses and open problems

The access/expression line is not sharp at the margin. The verdict leans on a distinction between a standard good and a bespoke expressive artifact, but the boundary is contestable: a decorated cake sits somewhere between a commodity and a commissioned artwork, and the framework does not yet supply a principled test for when creation becomes expressive enough to trigger the carve-out. The risk is that a generous reading of "expressive" lets Case A smuggle itself in wearing Case B's clothing — which is exactly the move the second asymmetry is meant to block, but blocks by disposition rather than by rule.

The verdict in Case B inherits the aggregation problem rather than solving it. The framework's inability to adjudicate the conflict is principled, not incidental, but it means the resolution offered is a defensible settlement, not a derivation. A reader who wants ethics to deliver a determinate answer will read "compel access, not endorsement" as the framework declining the hard part. This is the same exposure moral_sentimentalism records in its first weakness — a verdict many will experience as a bullet rather than a result — surfacing here on the liberty side rather than the harm side.

The compelled-expression weight partly shelters a conviction the framework judges wrong, and the tension is real. The first move says the norm is a defective Tier 1 exclusion; Case B then grants its holder a protection. The framework's reconciliation — that the protection runs through authority-over-one's-own-aperture and the bar on compelled meaning, not through the conviction's content — is coherent, but it sits in visible tension with the framework's commitment not to entrench exclusion. A critic could argue that an expressive carve-out, applied across enough cases, reconstitutes by accumulation the very status-exclusion Case A forbids. The framework's answer is the slow-structural-channel thumb on the scale, but a thumb on the scale is not a settled weight.

The dignitary harm is asserted as non-substitutable without a measure. The claim that being singled out is a recognition harm untouched by the availability of other vendors is doing real work in Case B, but the framework's known difficulty measuring stake-geometry (the measurement weakness in Stake-Geometry) means the magnitude of that harm — and therefore how heavily it should weigh against the compelled-expression stake — is stipulated rather than established.

Summary

The case looks like one question and is two. The framework's first move, grounded in Tiered Values, Stake-Geometry, and Network Universality rather than in any prior worked example, locates the religious conviction as a Tier 1 norm of the exclusionary kind: it fails the alien-entrant test for a Tier 2 principle, it withholds recognition from a class of apertures, and the thing it objects to is stake-elaboration rather than any collapse. The conviction is wrong on the framework's own metaethics. But its holder's stake in not being coerced against it is real, because the framework honors an aperture's authority over its own stakes independent of whether the stakes are well-founded.

Posed as the relational and political question it actually is, "service" splits. Refusing a standard good on the basis of identity is a recognition failure and constitution-by-exclusion: not permitted, firmly. Refusing to create a bespoke expressive artifact is a genuine conflict between the couple's non-substitutable dignitary recognition-harm and the owner's compelled-expression constriction — a conflict the framework can specify but, given the unsolved aggregation problem, cannot mechanically resolve. Its narrowest stable answer is to compel access and decline to compel endorsement, with the protected expressive zone read narrowly and the thumb on the scale toward inclusion. The framework's contribution is not a clean win for either side. It is the apparatus that says precisely where the easy part ends and the hard part begins, and why the hard part is hard.

Tested against the two Supreme Court cases that sit on these facts — Masterpiece Cakeshop (2018) and 303 Creative (2023) — the framework reaches three findings. Both were litigated in its contested Case B zone rather than its firm Case A zone. 303 Creative's result broadly matches the framework's narrowest resolution, while its one-sided, free-speech-categorical reasoning misses the conflict structure the framework insists on and leaves unpoliced exactly the expressive boundary the framework treats as the danger point — so the framework lands with the majority's outcome and the dissent's anxiety at once. And the verdict on which the framework is most confident, against status-based denial of ordinary goods, was never the question that reached judgment.

Behind those findings sits a limit on the framework's role itself: it is an instrument for making a court's moral reasoning more complete and more candid, not a decision procedure a court could adopt. Its sharpest verdict on the conviction is one a court is barred from issuing, and its comfort with leaving a conflict open is one the judicial role can share only in part. The framework improves the understanding of the case; it does not, and cannot, manufacture a resolution the stakes themselves withhold.

This document records an applied analysis developed through dialogue. It is exploratory, not peer-reviewed, and is offered as a contribution to thinking about how an aperture-based ethics handles conflicts between apertures in public and commercial life.