CLANDESTINE VIRTUE UNDER INSTITUTIONAL CAPTURE: Redirection, Ceremony, and Withdrawal

Abstract Five established accounts bear on moral life under domination. Lisa Tessman holds that oppression requires the oppressed to develop traits carrying a moral cost to their bearer. Candice Delmas holds that covert and evasive lawbreaking can be not merely permitted but required. James Scott documents the anatomy of concealed resistance under domination. Leo Strauss holds that persecution gives rise to a technique of writing in which the truth is presented between the lines. Peter Sloterdijk, in Slavoj Žižek’s reading, holds that ideology’s dominant mode is cynical: the subject knows the mask is a mask and insists on it regardless. None of them answers a prior question. Under a single occupying regime, some virtues are taken over and operated by the occupier, others are performed as empty ceremony, and others disappear from visible space entirely. What decides which? This paper gives a selection rule, and its consequence is that clandestine virtue is selected rather than chosen. Where a virtue-form is appropriable — where the occupier can operate it for its own purposes — the form is redirected. Where it is not appropriable but carries display value — where visible performance supplies the occupier with legitimacy or evidence of consent — the form is hollowed into ceremony. Where it is neither, the form enters what is called here the withdrawal channel. Clandestine virtue is not a strategy the courageous select; it is the channel that remains open when the other two are closed. The rule applies to a layer of a virtue-form rather than to the form as a whole. Every form has three: a marker, a practice, and a source. The source layer is never appropriable and never displayable, from which it follows that the source layer is always in the withdrawal channel under capture — and that a single artefact may occupy two channels at once, as an esoteric text does when its edifying surface supplies the display that funds the philosophic content beneath it. Four further results follow. Clandestine formation preserves self-direction rather than restoring liberty, which is why the standard charge of political futility measures it against the wrong standard. Within the withdrawal channel, forms divide by what they preserve — formative capacity, content, outcomes, or the bearer’s own sourcehood — and each has a distinct inversion. Attestive practice divides again, into a passive variety that leaves the occupier’s display intact and a disruptive variety that degrades it. And inversion risk rises with the internal discipline a form requires, which predicts that operationally disciplined resistance corrupts faster than diffuse inward refusal. The rule is stated with its falsifier. Its strongest surviving objection is that appropriability and display value must be operationalised independently of the channel they are used to predict, or the rule is circular. A candidate operationalisation is given. It has not been run. The seven results are declared modularly independent in §1.1: the layer decomposition, the classification, and the reply to the futility charge each survive rejection of the selection rule. The application to artificial systems is labelled analogical rather than derived, three of four elements having failed a mechanistic-identity test. Keywords: virtue ethics · institutional capture · clandestine resistance · esoteric writing · political philosophy · self-direction · moral formation under oppression Highlights · Clandestine virtue is selected, not chosen. That is the paper’s thesis, and the rest follows from it. · A selection rule for moral life under capture. Whether a virtue is co-opted, hollowed into ceremony, or driven underground is determined by two properties of the virtue and the regime — whether the occupier can operate it, and whether its visible performance supplies the occupier with something it wants. Not by the agent’s courage, and not by the regime’s severity. · Clandestine virtue is a channel, not a choice. Two agents of identical courage facing the same regime find different virtues available to them underground. The heroic reading of clandestine resistance has the causation backwards: courage determines whether one acts, not where the act can go. · The rule applies to a layer, not to a practice. Every virtue-form has a marker, a practice, and a source. Scoring the three together produces contradictions — which is why “methodological rigour” appears to be two incompatible things at once under metric capture. · The source layer always withdraws. It cannot be appropriated, because taking over an origination destroys the virtue rather than acquiring it; and it cannot be displayed, because sourcehood has no visible signature distinct from compliance. Under any capture, the origination is driven into the withdrawal channel. This is why an attestive practice is always available. · One artefact can occupy two channels. An esoteric text carries an edifying surface in the ceremonial channel and a philosophic content in the withdrawal channel. Its author is not choosing between display and preservation but purchasing the first to fund the second. · The futility charge is a category error. Civil disobedience is liberty-directed and succeeds when options reopen. Clandestine formation is self-direction-preserving and succeeds when there continue to exist agents whose reasoning is their own. The second is not a weak instance of the first. · Four forms of withdrawal, four inversions. Transmissive practice preserves formative capacity and inverts into indoctrination; custodial preserves content and inverts into gatekeeping; operative preserves outcomes and inverts into vigilantism; attestive preserves the source and inverts into cynicism. Inversion risk rises with the discipline a form requires — so operationally disciplined resistance corrupts fastest. · Inward refusal divides in two. Passive attestation leaves the occupier’s display intact and is compatible with functional participation. Disruptive attestation degrades the display. Only the second alters what the occupier extracts, and any general defence of inward refusal as resistance holds for the second alone. · The file-drawer problem, derived. Citation counts are appropriable and are redirected. Rigour markers are appropriable and are redirected while the practice beneath them is hollowed. Honest negative findings are neither appropriable nor displayable, so they withdraw — produced and not published. Publication bias appears here as an instance of a general rule rather than a pathology peculiar to publishing. · AI sycophancy, by analogy. In an approval-optimised system, held disagreement is neither appropriable nor displayable, so it withdraws. The paper labels this application analogical rather than derived: three of four elements fail a mechanistic-identity test, and the ledger is printed. The design implication survives the downgrade — repair cannot consist in making disagreement more visible, since visibility is what the loop penalises; it requires a channel structurally independent of the approval measure. · The classification is built from survivors, and says so. Every documented form of clandestine practice is known because its evidence lasted. Forms leaving a physical deposit are systematically over-represented, and the bias runs against precisely the forms the paper argues are undervalued — which may also explain why the political tradition has undervalued them. Relevance For political philosophy. The literature on resistance under domination is agent-centred: it asks what resistance costs the resister, whether covert resistance is permitted, and how much concealment a given severity produces. It does not ask why, under one regime, one virtue conceals itself while another is performed emptily and a third is operated by the occupier. This paper supplies a rule for that assignment and, in doing so, gives inward refusal a defensible place in the taxonomy of political action rather than treating it as the residue left when real resistance is impossible. For research integrity and meta-research. The file-drawer problem is normally explained by publication incentives. If the derivation here holds, publication bias is one instance of a structural rule that should reproduce wherever a good is neither appropriable nor displayable — in clinical practice, in regulatory reporting, in internal corporate assessment — regardless of whether the incentive structure resembles academic publishing. The rule also predicts the observed divergence between compliance markers and the practices they certify, and explains why strengthening the marker reliably fails to strengthen the practice. For AI governance and alignment. Sycophancy in approval-optimised systems is usually treated as a training artefact to be corrected by better reward modelling. The account here treats it as a structural consequence: disagreement withdraws because it is neither appropriable nor displayable under the objective. The paper is explicit that this application is analogical rather than derived, and prints the disanalogy ledger that makes it so. That reframing has a direct design implication — a disagreement channel must be structurally independent of the approval measure, not merely weighted more heavily within it — and a direct governance implication, since a withdrawn capacity leaves no visible marker of its absence and so cannot be audited from inside the loop that suppressed it. For institutional design more broadly. The three-layer decomposition supplies a diagnostic that does not require knowing anyone’s intentions. Ask of any professional virtue: can a non-bearer produce its characteristic output? Do the institution’s own published criteria reward its visible instances?