The Shape of the Disagreement: Why the Sex and Gender Debate Has the Structure It Has

The Shape of the Disagreement

Why the sex and gender debate has the structure it has, and where its real cruxes actually sit

Andraž Đurič

Version 2 · June 2026 · A diagnostic, not an argument for a side

The public argument over sex and gender is usually narrated as two incompatible accounts of reality, each accusing the other of denying the obvious. That framing is right about the heat and wrong about the structure. This piece does something narrower than take a side: it asks why the disagreement has the shape it has, where each of the two dominant models genuinely captures something, where each breaks, and which apparent collisions are real collisions versus people answering different questions and mistaking the answers for rivals.

The thesis is not that one model "uses more of reality." That is a category error, and an earlier version of this essay made it. The corrected thesis is that most of this debate is the predictable friction produced when a very good statistical regularity is mistaken for a law, when one word is asked to carry several distinct questions, when two different phenomena are merged, and when a disagreement about goals is dressed as a disagreement about facts. Locate the cruxes precisely and the noise resolves into a small number of real disagreements: one empirical and open, one definitional, and one about goals that does the loudest work and is not a factual question at all. (I also state, in §8, the one place where I think the empirical evidence leans, and why that lean does not decide the policy question. The reason it is at the end and not here is that the lean is only coherent once the machinery separating fact from goal is already built.)

Claim labels used throughout: definitional conceptual, settled by stipulation; empirical supported by evidence; contested empirical but disputed; normative a goal-relative claim, out of scope here.

Scope, and the one metaethical premise it rests on. This is a diagnosis of a disagreement, not a resolution of it. It recommends no policy and takes no position on how anyone should be treated or how any space should be organized. One premise makes the scope honest rather than evasive, and the reader is free to reject it: I treat normative claims as agent-relative and goal-dependent, not as free-standing facts about the world. On that view there is no separate moral magisterium floating above the physics. Once a goal is fixed, the facts constrain its policy implications tightly. What the facts cannot do is select the goal. That is why getting the science right does not end the policy fight, and it is why "out of scope" here means "this piece declines to pick the governing goal," not "values live in some domain facts cannot reach." A reader who holds that moral facts are mind-independent and non-natural will reject this premise; the diagnosis still mostly survives, but its treatment of the final crux is downstream of it.

1. Two models, and the questions they share and split on

Strip the slogans and two models are doing the work.

The binary model holds that the load-bearing fact about a person, for the purposes it cares about, is natal sex, understood as the body's organization around one of two gamete types. definitional It treats sex as fixed at development and most other sex-related variation as downstream of it.

The identity-inclusive model holds that the load-bearing fact depends on what you are explaining, and that for many purposes the relevant fact is gender identity: a stable, felt sense of one's own gender that usually but not always tracks natal sex. definitional It does not deny natal sex; it denies that natal sex is the only fact that matters across all contexts.

For some questions these models are not even rivals, and treating them as rivals manufactures heat. "What is this organism's reproductive type and what follows physiologically?" is the binary model's home question; the inclusive model has little quarrel with the answer. "What determines how this person understands and navigates their own gender?" is the inclusive model's home question; the binary model has little to say about it.

But it would be too charitable, and a softer relapse of the superset error, to claim the models only ever answer different questions. On at least one question they answer the same thing with rival answers: what should a category like "woman" pick out for legal and social purposes? normative The careful gender-critical position is not merely doing reproductive biology that happens to stop short of identity. It is advancing a sex-based-rights analysis: that for certain legal and political purposes the category should track natal sex, because the interests it exists to protect are organized around natal sex. The inclusive position answers the same question with identity or social role. That is a genuine collision, not cross-talk, and it is the collision this piece will eventually locate as normative rather than factual. Naming it now prevents the diagnosis from quietly recasting an opposed position as a merely incomplete one.

2. Why the binary feels like a law, and who is actually making the error

For the large majority of people, natal sex, gender identity, and social presentation all line up, and the binary predicts all three at once with very high accuracy. empirical A model that is right this often, this effortlessly, tends to get stored not as "a strong statistical regularity with rare exceptions" but as a law of nature. And violations of a law do not read as the tail of a distribution; they read as impossibilities, frauds, or errors.

This is the engine of much of the public reaction, and it is worth stating without contempt, because it is epistemically ordinary. empirical If a rule has held in every case you have personally met, treating a counterexample as suspect is normal inductive behavior, not malice. The friction is not mainly between good people and bad people; it is between a heuristic that performs so well it is taken for an exceptionless law and the real but uncommon cases a heuristic by definition does not cover.

Two things must be said immediately, or this diagnosis becomes the very partisanship it claims to avoid. First, the symmetric error sits on the other side: because the exceptions are real, it is tempting to argue the regularity is therefore not a regularity, that the binary is "a myth" dissolved by its exceptions. That overshoots. A rule with rare exceptions is still a rule; dusk does not abolish day and night. definitional Second, and more important, "mistaking the heuristic for a law" describes a common public error, not the considered gender-critical position. A careful gender-critical thinker holds both thoughts at once: the binary is a strong regularity, the exceptions are real, and natal sex should nonetheless govern certain categories. That is not a failure of statistical reasoning. It is a different weighting of interests, which is to say a normative position, and it cannot be diagnosed away as a cognitive mistake. Where the considered position lives, the disagreement is at Crux 2 below, not here.

3. The word "sex" is carrying at least three loads

A second source of cross-talk: "sex is binary" and "sex is not binary" can both be true, because they pick out different referents. definitional Disambiguating removes a large fraction of the apparent disagreement.

• Gametic sex. Defined by which of two gamete types the body is organized to produce. There is no third gamete, so for the typical case this is the best-covering and cleanest-defined of the three senses. empirical It is not, however, the clean ontological foundation its more confident defenders present, because "organized to produce" is partly teleological and frays at exactly the cases the debate is about. Complete androgen insensitivity (XY, testes, no functional gametes) already forces counterfactual reasoning about a pathway that did not complete. Swyer syndrome is worse: XY, gonadal dysgenesis, female-typical development, streak gonads producing no gametes at all. The pathway there did not stall, it diverged, and "organized to produce male gametes" is close to empty. Gametic sex is the right anchor for the modal case and a poor foundation for the disputed one, and the honest version says both.
• Sex-related traits. Height, hormone levels, muscle mass, fat distribution, voice pitch. Individually continuous and typically bimodal, with substantial overlap: within-group variation often exceeds between-group difference. empirical
• Sex as assigned at birth. A practical clinical or legal classification, usually read off genitalia, that mostly tracks gametic sex but is a separate act of categorization with its own edge cases. definitional

The popular "sex is bimodal, not binary" slogan is half right and contains a real equivocation, which the binary model's defenders are correct to flag. contested The bimodality is a property of sex-related traits, each with its own distribution; it is not a property of gametic sex, which does not sit on a single continuous axis you could be bimodal along. "Sex is bimodal" smuggles the genuine continuity of traits like height onto the categorical question of gamete type, where it does not belong. The honest statement is layered: gametic sex is near-binary for the typical case (with the teleological caveat above), the correlated traits are continuous and overlapping, and assignment-at-birth is a separate classification that usually but not always agrees with the first. Most "is sex binary" arguments are two people pointing at different layers.

4. Intersex and transgender are different phenomena, and the numbers matter

An earlier version of this essay leaned on intersex prevalence to carry weight for the transgender case. That was an instance of the exact conflation this piece exists to expose, and it is worth correcting in public. The two are distinct.

4.1 Intersex: variation in the biological inputs

Intersex conditions, or differences of sex development, are cases where the biological markers of sex (chromosomal, gonadal, hormonal, anatomical) do not align in the standard way. definitional They bear on gametic and categorical sex directly. And the prevalence is contested by nearly a factor of a hundred, entirely because of definition. contested The widely circulated 1.7% figure (Fausto-Sterling; Blackless et al. 2000) uses a broad definition that counts conditions such as Klinefelter (XXY), Turner (45,X), and late-onset adrenal hyperplasia, which most clinicians do not treat as sexually ambiguous at birth. The narrow clinical definition (Sax 2002), restricting "intersex" to cases where chromosomal sex conflicts with phenotype or the phenotype is not classifiable as male or female, yields roughly 0.018%, about one in 5,500. National self-report surveys land in between, near 0.3 to 0.4%. (A terminological note: the clinical field has largely shifted from "intersex" to "differences" or "disorders of sex development" (DSD), partly to name specific diagnoses rather than a blanket category. That shift is not neutral either: many of the people it describes reject "disorder" as pathologizing. This piece keeps "intersex" for continuity with the prevalence literature, while noting that the label itself is contested from more than one direction.)

Crux 1 (definitional, with an ontological commitment underneath). The intersex prevalence dispute is not about data. Both numbers trace to real conditions; it is a dispute about which conditions "intersex" should name. But it would be too quick to call it purely definitional and move on. The choice of definition is not arbitrary: it encodes a prior commitment about whether atypical development is a variation within a binary system or evidence against the binary framing. The broad definition treats the binary as a human imposition on a continuum; the narrow definition preserves the binary and classes the remainder as disorder. So the rhetorical payload lives in the definitional choice, and the definitional choice in turn reflects a prior ontology. The honest move is to state the definition first, the number second, and the ontological commitment third, rather than letting the number smuggle the ontology in unannounced.

4.2 Transgender: a mismatch at a different level

Transgender identity is not a variation in gametic or chromosomal sex. definitional It concerns a mismatch between gender identity and natal sex in people whose biological sex markers are typically unambiguous. Using intersex prevalence to argue about transgender people therefore proves nothing about them; it imports the edge cases of one phenomenon to soften the boundary of a different one. The binary model's defenders are right to reject that move, so this essay rejects it too.

5. The brain-evidence detour, and why neither side needs it

A recurring inclusive move points to "the transgender brain": neuroimaging said to show trans people's brains resembling their identified gender. The honest read is that this literature is mixed and cannot bear the weight placed on it. contested Large studies, including a mega-analysis of over 800 subjects, do find group-level structural differences, but the patterns are heterogeneous across regions and studies, confounded by sexual orientation, hormone exposure, and comorbidity, and subject to the replication problems endemic to neuroimaging. That mega-analysis concluded that transgender persons present, if anything, their own distinct brain phenotype rather than a clean shift toward the identified sex, which is the opposite of the tidy "born in the right brain" story the evidence is often recruited for. More fundamentally, the brain is not a cleanly dimorphic organ; reported male-female differences are group averages with heavy overlap, "differences that do not differentiate." There is no settled "male brain" or "female brain" template for a trans brain to match or miss.

The deeper problem cuts against the inclusive side's own commitments. definitional The search for a neural correlate concedes a premise it should reject: that a felt psychological state needs a brain-scan signature to count as real. On any physicalist view the felt state already is a physical state of the organism. Demanding the correlate to license the identity is a quiet dualism, treating the experience as second-class until a structural proxy validates it. The brain evidence is thus a detour for both sides: too weak to settle anything for the side that invokes it, and unnecessary to the position it is meant to support.

One distinction must be held firm here, because a common objection turns on blurring it, and it is easy to overclaim in the other direction. To call a gender identity real in this context is a narrow descriptive claim: that it is not feigned and not a fleeting whim, but a stable, persistent disposition. empirical That is all the word carries here, and it does no work toward the legal question, because stability does not distinguish gender identity from stable preferences or even stable false beliefs, and "it is a physical state" is trivial under physicalism, where whims and delusions are physical states too. The substantive question, whether gender identity is a kind with the sort of causal structure that should ground a legal category, is genuinely open and is itself part of the normative crux below; it is not settled by noting that the identity is not a whim. So granting the descriptive reality of gender identity is not covertly taking the inclusive side on the semantic question. One can hold that gender identity is a real, stable psychological state and that a legal category should track natal sex; careful gender-critical thinkers hold exactly this combination. The objection that affirming the first must concede the second conflates a fact about a person with a decision about a word, and the reverse overclaim, that stability alone earns the legal category, makes the same conflation from the other direction.

6. The argument that is not about facts

The hottest exchanges concern sex-segregated spaces, sports, records, and language. They are usually conducted as factual disputes about what sex or gender "really is." They are not. normative They are disputes about which classification an institution should use for a given purpose, and which interests that purpose should weight. Given the scope premise, that is a question about goals, and the facts do not contain the goal.

This is the cleanest result in the piece. Once the facts above are fixed (gametic sex near-binary for the typical case, traits continuous, intersex rare and definition-dependent, transgender a distinct identity-level phenomenon, brain evidence inconclusive and unnecessary), none of them selects a policy. To get from any fact to "therefore this space should be organized this way" you must add a goal: what the space is for, and whose interests it should weight when they conflict. That goal is not in the biology or the psychology.

This is the right place to answer the objection that fact and value are entangled, that "harm" and "fairness" are partly empirical, and that strong enough evidence of harm would move almost everyone. That objection is correct and the scope premise predicts it rather than being embarrassed by it. Because normativity here is goal-relative, the facts constrain hard once a goal is fixed: if a policy demonstrably produced a harm the parties already disvalue, the facts would bind the conclusion tightly. What the facts still cannot do is adjudicate between the goals when the parties weight competing interests differently. Fairness-to-whom and safety-against-which-risk are where the goals diverge, and that divergence is not a measurement that better data resolves.

Crux 2 (normative, goal-relative, out of scope here). The policy disagreement is real, but it is not settled by getting the biology or psychology right, because it is a conflict between agents protecting different interests in a shared institution. One side seeks to protect interests organized around natal sex (privacy, safety, and fairness in sex-segregated contexts); the other seeks to protect interests organized around gender identity (recognition, access, and freedom from exclusion). Both are protecting interests and both are recognizing identities; the dispute is over which interests and which identities a given institution should weight when they conflict. Even this even-handed phrasing is contestable, since each side would describe its own goal and the other's in different terms, and that descriptive disagreement is itself part of the fight. The established facts underdetermine the choice between the goals. A person can accept every factual claim in this document and land on either side, because the facts constrain each goal's implications without selecting the goal. Pretending the fight is about facts is what keeps it unwinnable: each side fields true statements that cannot reach the conclusion, and reads the other's refusal to be moved as a denial of reality, when it is a difference in goals.

For the same reason, an older argument this essay used to make is dropped: "if it were about protecting women, women would be the main constituency, and exclusion always leads to violence." The first is a non-sequitur (stated goal and demographic constituency need not track); the second is a slippery slope. Both try to win a goal-level disagreement with pseudo-empirical shortcuts, the precise move this section names. The incidence questions underneath them (how often a given harm actually occurs under a given policy) are genuinely empirical and answerable, and they are the one place better data does bind, but they are a separate inquiry this document does not run.

7. Where this sits, and what it is not

This is philosophy of classification, in the lineage that distinguishes natural kinds from human kinds and studies how classifying people changes the people classified (Hacking's looping effects; Dupré on the purpose-relativity of biological categories). It is not biology, not clinical science, and not advocacy. It deliberately does not adjudicate the live philosophical dispute between gender-critical accounts (Stock) and trans-inclusive accounts (Dembroff; Jenkins, building on Haslanger's ameliorative analysis), because once the equivocations are cleared that dispute reduces to Crux 2: a goal-relative disagreement about which classification should govern which institution. Both camps are internally varied, not monolithic; "gender-critical" alone spans positions from Stock's to Lawford-Smith's, and the representative here is a representative, not a survey.

The claim is only this. The debate has the shape it has because a high-accuracy heuristic is widely mistaken for a law, because one word carries three referents, because two different phenomena get merged, and because a goal-level disagreement is dressed as a factual one. Fix the first three and the factual surface area shrinks dramatically. What remains is the fourth, and on the scope premise of this piece the fourth is not a fact-shaped object. It is a conflict of goals between agents, which is exactly why no amount of getting-the-science-right dissolves it.

8. Summary, and where I think the evidence leans

1. The two models share their home questions peaceably but genuinely collide on one: what a category like "woman" should pick out for legal and social purposes. That collision is normative, not cross-talk (1).
2. The binary is a strong regularity widely mistaken for a law, which explains the public reaction without endorsing the law-reading; the considered gender-critical position is not this error but a different weighting of interests (2).
3. "Sex" denotes at least three things: near-binary gametic sex for the typical case (with a teleological caveat that frays at CAIS and Swyer), continuous overlapping traits, and a separate assignment-at-birth category. Most "is sex binary" arguments confuse these layers (3).
4. Intersex prevalence is contested ~100x purely on definition (broad ~1.7% vs narrow ~0.018%), and the definition encodes a prior ontology; transgender identity is a distinct identity-level phenomenon and cannot borrow intersex edge cases (4).
5. Neuroimaging is mixed, confounded, and in any case unnecessary, since demanding a neural correlate concedes a dualism a physicalist should reject; the descriptive reality of gender identity (stable, not feigned) is separate from, and does no work toward, the semantic question of what "woman" denotes (5).
6. The policy fights are goal-relative and normative; facts constrain each goal's implications tightly but do not select the goal, which is why fixing the science does not end them (6).

The cruxes sort into three bins. One is genuinely empirical and open: the incidence questions behind specific policies (deferred here). One is definitional with an ontology underneath: what "intersex," and "sex," should name. One is goal-relative, does the loudest work, and is out of scope by design.

Here is the lean promised at the start, placed where it can be read correctly. On the empirical sub-claims the two sides actually make about the world, the identity-inclusive side is, in my judgment, less wrong: the binary-as-exceptionless-law claim, the dismissal of trait bimodality, and the framing of transgender identity as delusion are the factual errors that recur most, and they cluster on one side. That is a claim about the empirical bin only. It does not transfer to Crux 2, and the structure of this piece is what blocks the transfer: being less wrong about the facts cannot win a dispute that the facts do not decide. A reader who infers "inclusive side is less wrong empirically, therefore inclusive policy should win" has made exactly the fact-to-goal leap the whole document argues is invalid. The empirical lean and the normative neutrality are not in tension; they are on opposite sides of the firewall, and saying both out loud is more honest than hiding either.

References

1. Fausto-Sterling, A. (2000). Sexing the Body: Gender Politics and the Construction of Sexuality. Basic Books. Source of the broad-definition 1.7% intersex estimate, developed from Blackless et al. (2000).
2. Blackless, M., Charuvastra, A., Derryck, A., Fausto-Sterling, A., Lauzanne, K., and Lee, E. (2000). How sexually dimorphic are we? Review and synthesis. American Journal of Human Biology, 12(2), 151 to 166. The study underlying the 1.7% figure.
3. Sax, L. (2002). How common is intersex? A response to Anne Fausto-Sterling. The Journal of Sex Research, 39(3), 174 to 178. Argues the clinically standard definition yields ~0.018%, roughly two orders of magnitude below 1.7%.
4. Hacking, I. (1999). The Social Construction of What? Harvard University Press. Natural kinds versus human kinds; the looping effect of classifying people. Located by argument; specific page references not given here.
5. Dupré, J. (1993). The Disorder of Things: Metaphysical Foundations of the Disunity of Science. Harvard University Press. On the purpose-relativity of biological classification, the backbone of the "models answer different questions" point.
6. Stock, K. (2021). Material Girls: Why Reality Matters for Feminism. Fleet. Representative of the gender-critical philosophical position tying the relevant category to natal sex. Cited for positioning, not endorsement.
7. Dembroff, R. (2020). Beyond binary: Genderqueer as critical gender kind. Philosophers' Imprint, 20(9). Representative trans-inclusive analysis of gender as a kind. Cited for positioning, not endorsement.
8. Haslanger, S. (2000). Gender and race: (What) are they? (What) do we want them to be? Noûs, 34(1), 31 to 55. The foundational statement of ameliorative analysis of gender and race concepts, on which Jenkins (2016) builds.
9. Jenkins, K. (2016). Amelioration and inclusion: Gender identity and the concept of woman. Ethics, 126(2), 394 to 421. Extends Haslanger's ameliorative approach to be trans-inclusive.
10. Mueller, S. C., Guillamon, A., Zubiaurre-Elorza, L., et al. (2021). The neuroanatomy of transgender identity: Mega-analytic findings from the ENIGMA Transgender Persons Working Group. The Journal of Sexual Medicine, 18(6), 1122 to 1129. doi:10.1016/j.jsxm.2021.03.079. Mega-analysis (n=803); finds group differences but concludes transgender persons present a distinct phenotype rather than a clean shift toward the identified sex.
11. Frigerio, A., Ballerini, L., and Valdés Hernández, M. (2021). Structural, functional, and metabolic brain differences as a function of gender identity or sexual orientation: A systematic review of the human neuroimaging literature. Archives of Sexual Behavior, 50, 3329 to 3352. Reviews the mixed and confounded state of the literature.
12. Baxendale, S. (2025). Brain sex: Differences that do not differentiate. Archives of Sexual Behavior, 54, 3793 to 3796. doi:10.1007/s10508-025-03306-z. On the non-dimorphic brain ("differences that do not differentiate") and replication problems in transgender neuroimaging.

Author: Andraž Đurič, independent researcher, Slovenia. Written in collaboration with Claude (Anthropic), with earlier feedback from other AI systems. This is a diagnosis of a disagreement and takes no policy position; it rests on one stated and rejectable premise, that normativity is agent-relative and goal-dependent rather than a free-standing fact. Empirical and definitional claims were checked against the cited literature, and contested figures are given with the dispute attached. Licensed CC BY 4.0. Open to revision; corrections and counterexamples are welcome.