The paper’s starting point is what Di Giorgio calls the ontological fracture: the gap between the lived, contextual fabric of social reality and the finite vocabulary of variables and categories into which any computational system must translate it. Encoding race as a discrete feature, or substituting the six-point Fitzpatrick skin-type scale in computer vision evaluations, does not dissolve the problem; it merely relocates the decision about which distinctions count as relevant. Two epistemic consequences follow. The first is a resolution limit, which the paper terms epistemic grain: like the grain of photographic film, a model’s feature space determines the finest distinctions it can register, and phenomena below that grain are unavailable in principle to any analysis conducted in the system’s own terms. The second is more insidious. Once categories are fixed, whatever exceeds them registers not as evidence of an inadequate category scheme but as residual error, noise to be reduced with more data. The paper calls this systematic misreading computational fog: the distortions introduced by representational choices are not representable within the representation that produced them. Gender classifiers trained on a binary scheme illustrate the structure, since a person whose identity does not fit the binary confronts an inadequacy of the scheme itself, which the system can only record as noise.

What makes the computational case distinctive, the paper argues, is not abstraction as such, which all modelling shares, but the timing of precisification. Drawing on Friedrich Waismann’s notion of open texture and H.L.A. Hart’s account of legal adjudication, Di Giorgio observes that social concepts such as race, merit, risk and need are never delimited in all possible directions; human decision-makers sharpen them case by case, in view of particulars, and can revise the sharpening when the next case differs. A computational system cannot do this. It fixes the precisification of every open-textured concept once, ex ante, at design time, and applies it uniformly at scale, with no interpreter positioned to notice that the present case exceeds the scheme and re-open it. This also explains why the familiar reassurance that a human remains in the loop underdelivers: an overseer reviewing outputs under the EU AI Act’s Article 14 can check decisions, but cannot re-open the precisification itself, because the information that would motivate re-opening has typically been discarded below the system’s epistemic grain.

The paper’s central contribution is a criterion for deciding when representational discrimination is remediable, borrowed from an unexpected source: the less-discriminatory alternative construct of United States disparate-impact doctrine. In litigation, a challenger who shows a facially neutral policy produces disparate impact can prevail by exhibiting an alternative policy serving the same legitimate purpose with a smaller disparity. Di Giorgio shifts that comparative logic one level up, from models to operationalisations. A disparity is contingent, relative to a legitimate purpose, a measure of disparity, and a class of admissible designs, when some alternative operationalisation within that class serves the purpose while reducing the disparity. It is constitutive when no such alternative exists. Nothing is constitutive in the abstract; every classification is elliptical for the purpose, the disparity measure, and the design space against which it is made. Debates that sound metaphysical, whether a bias is inherent to computation, thereby dissolve into three tractable questions: what the purpose is, what the relevant disparity is, and what the admissible design space actually contains.

The criterion’s logical form carries real teeth. Contingency is verified by exhibition: produce one admissible alternative and the classification is settled. Constitutivity, by contrast, is a negative existential claim, that no alternative exists, and can never be conclusively verified over a design space that is not surveyable. It is established only by provisional exhaustion of the known alternatives and remains permanently open to refutation by a single new exhibit. This asymmetry generates a burden structure: the critic must show contingency by exhibit, while the deployer claiming a disparity is unavoidable owes documented exhaustion of the alternatives considered and the grounds for excluding each. Neither burden is discharged by assertion. The result converts an unfalsifiable rhetorical endpoint into a procedure that specifies in advance what would show any constitutivity claim to be wrong.

Three case studies test the criterion adversarially, and the first delivers a striking reversal. The ImageNet benchmark inherited, from the WordNet lexical taxonomy beneath it, derogatory categories under the person synset, so photographs of real people were assigned to demeaning classes because the conceptual vocabulary admitted them. The bias sat exactly where the paper’s framework looks, in the category scheme itself, upstream of any training, which would ordinarily invite the label constitutive. Yet the criterion returns the opposite verdict: an admissible alternative, filtering and rebalancing the person subtree while preserving the dataset’s benchmarking purpose, was not merely available in principle but actually carried out by researchers in 2020. Upstream, the paper concludes, does not mean constitutive, and the slide from upstream to irremediable is the single most productive source of misclassification in the field.

The second case, tokenisation in large language models, resists a single verdict and forces a decomposition. Byte pair encoding builds vocabularies from corpus statistics, so well-represented languages receive compact tokenisations while under-represented ones fragment into many small units; research shows the same content can require up to fifteen times more tokens in some languages than others, compounding into higher costs, shorter outputs and less context for speakers already economically constrained. The contingent component, attributable to the construction method and allocation policy, has demonstrable remedies. The residual component is subtler: even character-level and byte-level encoding shows more than a fourfold difference in length for some language pairs, reflecting the differing information densities of scripts and the uneven byte costs of UTF-8. Within the class of systems that segment all languages with one shared inventory and price by segment count, every known alternative relocates the disparity rather than eliminating it. The paper classifies that residual layer as constitutive only relative to that class and to current knowledge, hedged exactly as the criterion demands, and refutable by any scheme achieving cross-linguistically uniform encoding costs.

The third pair of cases shows the criterion operating where layers are entangled. Amazon’s abandoned CV-screening tool penalised indicators of being a woman, and neutralising explicit indicators failed against subtler proxies; under the criterion, that documented disparity was contingent along at least one dimension, since outcome-based targets and structured competency assessments exist within the class of automated screening, while the failure of particular remedies counts as evidence about the depth of the disparity, never as proof of inevitability. The clinical case supplies the empirical anchor: Obermeyer and colleagues’ landmark 2019 study found that a widely deployed algorithm operationalising health need as health cost systematically underserved Black patients, because unequal access depresses spending conditional on illness. Re-operationalising the target through measures of health reduced the disparity by 84 percent, a remedy invisible at the model level. The paper even rereads the famous impossibility theorems through this lens: they are a constitutivity result, but one proved relative to a class of designs that holds fixed a population partition and base rates which are themselves representational commitments, not primitives of nature.

The regulatory payoff is a proposal the paper calls ontological auditing: an upstream evaluation of representational choices during design, documented in an Ontological Impact Statement. For each socially consequential construct, the statement would record what the variable purports to represent, which dimensions it excludes, what proxy relationships link it to protected characteristics, which alternative operationalisations were considered and why each was excluded, and, for any disparity the deployer deems unremediable, the parameters and exhaustion supporting that claim. Filed before deployment, such a statement in the clinical case would have exposed the cost-as-need choice while it was still cheap to revise. The paper locates the practice within the EU AI Act’s risk management and data governance provisions, noting that current obligations discipline data quality and outputs but never ask the question disparate-impact doctrine has long asked: was a less-discriminatory way of achieving the same end available? The proposal is explicitly untested, and its administrative cost unknown.

What the framework ultimately buys is a change of register. Only one of the disparities examined is classified as constitutive, and that classification is relative to a class and hedged to current knowledge, because the design space expands with technique and unhedged inevitability claims are exactly what the criterion forbids. Where a representational choice is genuinely forced, the paper asks that the fact be established, documented and weighed openly rather than discovered after deployment. Where it is not forced, and the evidence suggests that is the more common case, the criterion poses the simpler question that the vocabulary of inevitability has tended to obscure: what else could have been built, and why was it not?

Subject of Research: A criterion for distinguishing contingent from constitutive algorithmic discrimination arising from upstream representational choices in AI systems

Article Title: When algorithms cannot be fair: constitutive and contingent discrimination and the limits of less-discriminatory alternatives

Article References: Di Giorgio, A. (2026). When algorithms cannot be fair: constitutive and contingent discrimination and the limits of less-discriminatory alternatives. AI & SOCIETY. https://doi.org/10.1007/s00146-026-03384-0

Image Credits: AI Generated

DOI: 10.1007/s00146-026-03384-0

Keywords: algorithmic fairness, algorithmic discrimination, less-discriminatory alternative, open texture, ontological auditing, EU AI Act, ImageNet, tokenisation, clinical risk scoring, disparate impact, computational opacity, AI & Society

The technical stakes of this distinction are considerable. In empirical science, the status of a claim — whether it functions as a hypothesis to be tested or as an established fact to be assumed — determines everything downstream: which research questions get asked, which methodologies are deployed, which alternative explanations are considered, and which findings are treated as anomalies rather than confirmations. Redstone’s argument is that when a causal claim is treated as settled, scholars lose any reason to explore competing explanations for observed disparities, and simultaneously acquire every reason to suspect the motives of colleagues who propose them. The assumption, she contends, has become so foundational to the practice and teaching of sociology in the United States that it steers research agendas, shapes student training, defines what counts as serious scholarship, and silences debate before it can begin.

Tracing the intellectual genealogy of this shift, Redstone identifies an unexpected catalyst: the Supreme Court’s 1971 decision in Griggs v. Duke Power Co. In that landmark case, the court ruled that the energy company’s employment policy — which conditioned hiring and job transfers on holding a high school diploma and passing two written tests — violated Title VII of the Civil Rights Act by limiting job opportunities for Black candidates. The ruling established what legal scholars call disparate impact doctrine, holding that employment practices need not be intentionally discriminatory to be unlawful if they produce unequal outcomes across racial groups without sufficient business justification. According to Redstone, this decision functionally embedded a causal claim — that disparate outcomes are produced by discrimination — into American law, and attached a moral valence to that claim in the process.

The moral dimension is central to Redstone’s analysis. Labeling an outcome discriminatory, she writes, is not merely describing a statistical pattern; it is assigning blame, even when the practice that produced the outcome was adopted without any intent to exclude. This is a subtle but consequential move in the logic of social science. A descriptive claim — that a disparity exists — is verifiable through data collection and statistical analysis. A causal claim — that discrimination produced the disparity — requires additional evidence ruling out alternative explanations. A moral claim — that someone is responsible for the disparity — carries an entirely different evidentiary and normative structure. Redstone argues that the post-Griggs environment collapsed these three distinct claim types into one another, so that observing an unequal outcome became sufficient grounds for asserting both its discriminatory cause and its moral condemnation.

Before the court’s decision, Redstone notes, sociological research of the 1960s and 1970s was methodologically pluralistic enough that researchers investigating racial disparities could and did explore alternative explanations. The hypothesis that institutional racism might be causing some or all of the differing outcomes was one possibility among many in their analytical toolkits, competing alongside explanations rooted in economics, geography, family structure, institutional resource allocation, and other factors. The Griggs decision, in her account, justified treating as settled the question of whether disparate outcomes were caused by discrimination or by other factors — and the ramifications extended far beyond law into academia, altering prevailing conceptions of harm, justice, group identity, and knowledge itself.

To assess the effects on contemporary scholarship, Redstone examines findings from several studies published over the past 25 years. The authors of those studies, which explored racial disparities in student discipline, health outcomes, criminal justice, and generational wealth accumulation, each concluded that the differences they documented were indicative of institutional racism. Redstone is careful to state that her point is not that these conclusions are necessarily wrong. Each of them, she writes, may be capturing something real about institutions’ resource allocations and about how disadvantage persists or grows across time. Her objection concerns the inferential structure: once it became widely accepted that inequities were themselves evidence that institutional racism produced them, the implications compounded, changing what discrimination was understood to look like and how pervasive it was believed to be. The reasoning became circular — disparities prove discrimination, and discrimination explains disparities.

According to Redstone, this circularity gave rise to a new organizing principle in which injustice became measured around group identity. Any negative outcome associated with a group that could be categorized as disadvantaged became presumptively attributed to the disadvantage that the identity carries. In this framework, identity does double duty: it not only indicates who experiences a given outcome but also explains why they experience it. The causal and moral weight attached to group identity subsequently reorganized scholarship across virtually every domain of sociological research, including health, education, criminal justice, culture, and labor markets. Conversely, outcomes experienced by majority groups came to be interpreted as the products of the same system operating in reverse — as unearned advantage — a reading that, Redstone argues, disregards the possibility that these results might stem from other complex factors deserving independent investigation.

The costs of dissent within this framework are asymmetric and severe. Challenging the assumption that institutional discrimination is the primary explanation for racial disparities carries different consequences depending on who voices the challenge, but for anyone, Redstone observes, dissent can be read as indifference to injustice or as complicity in it. Skeptics risk being accused of blaming the victim, and their skepticism itself may be treated as evidence of the very bias they are questioning — a self-sealing logical structure in which denial of the claim confirms the claim. Once disparities are presumed to be produced by discrimination, proposing a non-discriminatory explanation is, by definition, denying that discrimination is occurring. Redstone emphasizes that this moral logic cannot be softened or set aside while the original claim remains settled, because it follows directly from the claim itself. This is precisely why the claim’s epistemic status — hypothesis or fact — matters so much for everything built upon it.

The irony at the heart of the paper, Redstone concludes, is that sociology’s wholesale adoption of this central framework has compromised the discipline’s own capacity to examine its institutional processes and their impact — a reflexive blind spot in a field whose core mission is precisely that kind of institutional self-scrutiny. Redstone explores these themes in greater depth in her forthcoming book, Presumption of Guilt: How Equating Inequality with Injustice Fractured American Democracy, to be published in April 2027 by Pitchstone Publishing. In the book, she analyzes the implications of the broader ideological shift for American democratic institutions, arguing that they undermined a key element of their survival by treating the contested moral claim that inequality is inherently unjust as a settled question. Once inequality is equated with injustice, she argues, it morally recodes an entire worldview concerning justice, oppression, identity, and a wide range of topics extending well beyond inequality itself. The paper, published on 27 August 2026 and based on content analysis, invites sociologists to restore the empirical discipline that once allowed competing hypotheses about disparity to be weighed on their evidence rather than judged by their moral implications.

Subject of Research: The treatment of systemic racism as a default causal explanation for racial disparities in American sociology

Article Title: Paper explores systemic racism as the default explanation for disparities

Article References: Paper explores systemic racism as the default explanation for disparities. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: sociology, systemic racism, racial disparities, Griggs v. Duke Power, disparate impact, Theory and Society, Ilana Redstone, research methodology, group identity, academic debate, Civil Rights Act, moral epistemology