Field Notes

synthesis

Why a known harsher penalty can beat an uncertain lesser one

StatusWorkingReviewdue UpdatedStands on3 sourcesSupportsnotes

Someone under investigation might prefer a known, bounded sentence for a “process crime” — refusing to disclose an encryption key, contempt of court, obstruction — over cooperating, even where the process crime’s own statutory maximum is formally longer than a shorter sentence that a substantive charge, if proven, would carry. Five known years for refusing to decrypt can be preferable to three known years for a child-sexual-abuse conviction built in part on one’s own device or one’s own words, because the substantive charge’s damage does not end when the sentence does. This preference can hold for at least three analytically distinct reasons, only one of which the strategy’s naive presentation usually names:

  1. the substantive charge’s collateral consequences — registration, unemployability, social exile — can exceed a formally longer bounded sentence in lived severity;
  2. a dignitary or agential reason, independent of any sentence-length or collateral-consequence comparison: being convicted through one’s own words, extracted by deception or psychological pressure, is a different kind of injury than being punished for open, principled refusal, even holding severity constant;
  3. a legal-but-compromising disclosure risk: cooperating may not produce evidence of the charged offense at all, but may surface other true, legal information whose exposure, through trial use, informal law-enforcement retention, or a leak, causes damage independent of whether the original charge was ever substantiated.

Each is evaluated below on its own evidentiary terms.

Reason one: collateral consequences

The United Kingdom’s Regulation of Investigatory Powers Act section 49/53 regime, the sharpest form of Encryption key disclosure compulsion, functioned in two real cases exactly as this reasoning predicts: Oliver Drage and “JFL” both received sentences far under the statutory ceiling, with no substantive charge ever following, because the underlying data stayed encrypted. Sex-offender-registry scholarship (Wayne Logan; Corey Rayburn Yung; Gabriel Chin’s “The New Civil Death,” University of Pennsylvania Law Review 160, 2012) independently establishes, as settled academic consensus, that registry consequences — unemployability, residency restriction, public notification, often indefinite duration — routinely exceed the formal sentence’s severity in practice, which substantiates why a rational person might accept a nominally or even numerically longer sentence to avoid them entirely. Segal and Stein’s “Ambiguity Aversion and the Criminal Process” (Notre Dame Law Review 81, 2006) gives the underlying psychology a name, asymmetric ambiguity aversion: a person facing a catastrophic, poorly quantified tail risk will rationally accept a certain, known alternative to eliminate that uncertainty, even where the known alternative looks worse on its face.

The decisive counter-evidence remains Francis Rawls, documented in Encryption key disclosure compulsion: held in civil contempt for more than four years, nearly three times the eighteen-month cap eventually recognized as lawful, before release, and never charged with the underlying offense. Whether a “bounded lesser charge” is actually bounded depends on the jurisdiction’s specific legal mechanism — a discrete, fixed-maximum offense like the UK’s section 53, or an open-ended coercive process like US civil contempt — not on the person’s own preference or reasoning.

Reason two: dignitary and agential harm, independent of outcome

There is real philosophical grounding, separate from sentence-length or collateral-consequence reasoning, for treating conviction through deception-extracted words as a distinct kind of injury from punishment for open refusal.

Jennifer Lackey’s concept of agential testimonial injustice (Journal of Criminal Law and Criminology 110, 2020; Criminal Testimonial Injustice) is the strongest single anchor. Lackey argues that standard testimonial-injustice theory captures only the wrong of being unfairly disbelieved; false and coerced confessions reveal a distinct wrong, a credibility excess obtained by subverting the speaker’s own epistemic agency. When interrogation tactics documented in Coercive interrogation and false confessions and Why you shouldn’t talk to the police — minimization, false-evidence ploys, exhaustion, psychological pressure — alienate a suspect from her own normal capacity for reasons-responsive judgment, the resulting statement is then treated as maximally credible evidence precisely because that capacity was disabled, not exercised. Lackey’s own framing describes the state “rendering victims complicit in their own undoing” — language that names, with philosophical precision, the distinct indignity the manipulation-avoidance reason describes, independent of whatever sentence eventually follows.

Antony Duff’s communicative theory of punishment (Punishment, Communication, and Community, 2001) supplies a complementary structural point, though this is an inference from his framework rather than his own stated conclusion: if punishment’s legitimacy depends partly on the offender’s own authentic acknowledgment of wrongdoing, a conviction obtained by deceiving the defendant about the evidence or the moral stakes undermines the very communicative relationship punishment is meant to instantiate. Civil-disobedience literature (Rawls; Cohen; Garcia’s 2022 “Rethinking Acts of Conscience,” Philosophy) independently confirms that accepting a known punishment for open, principled refusal carries intrinsic, non-strategic value in the philosophical literature on conscience — signaling sincerity and respect for legal order in a way that is not reducible to a calculated bet on outcomes. The doctrinal history of the Fifth Amendment privilege itself encodes a related intuition: the classic “cruel trilemma” (self-accusation, perjury, or contempt) treats contempt, a bounded, known, chosen consequence, as the least objectionable branch precisely because it does not require betrayal of self.

This grounding is real but should not be overstated. The empirical psychology of this specific comparison is genuinely thin. Saul Kassin’s false-confession literature documents the mechanism — internalized guilt, confabulated memory, disorientation on realizing one was deceived — but no study was found that isolates and measures the comparative subjective experience of shame after manipulation against acceptance of a known consequence for defiance. One suggestive but indirect data point exists: Clow and Leach’s research on public perception of exonerees found that those whose wrongful conviction involved a false confession are stigmatized differently, and more, than exonerees convicted by other means — a third-party perception finding, not a first-person psychological one, but consistent with the intuition that “confessed, even falsely” carries a distinct and harder-to-shed taint. No study was found comparing long-term reintegration or psychological outcomes between people punished for principled refusal and people convicted via extracted self-incrimination, holding sentence severity constant. This should be treated as an open empirical question, not resolved in either direction by weak proxy evidence.

Cooperation risks exposing more than evidence of the charged crime. Three distinct mechanisms document this, each verified independently.

Propensity evidence from an actual, if uncharged, prior offense is real and substantial, detailed in Propensity evidence from legal but suggestive material: US Federal Rules of Evidence 413 and 414 let a jury weigh an uncharged, unconvicted prior sexual offense as propensity evidence for the charged one, decided on a mere preponderance standard. This risk is narrower than it first appears, though: merely legal, stigmatizing material — legal pornography, private fiction, community membership — does not itself qualify as this kind of “other act,” and US courts have repeatedly rejected attempts to introduce it for that purpose, though it can still reach a jury under narrower theories and still carry real prejudicial weight. England and Wales’s bad-character regime under the Criminal Justice Act 2003, Part 11, is structurally broader, reaching “reprehensible behaviour” generally and explicitly including acquitted or unprosecuted conduct, though whether legal-but-suggestive material alone would qualify remains untested in the case law found.

Informal law-enforcement retention absent any charge is documented, official policy in the United Kingdom, not a speculative risk. The Police National Database retains what is officially termed “soft intelligence” — information that does not relate to a conviction, caution, or other formal action — for people arrested but not charged or investigated without a formal outcome, reviewable but retainable indefinitely under the National Retention Assessment Criteria where an individual is still considered a threat, independent of any charge. An ECHR-level finding (2025 digest) held a UK Enhanced Criminal Record Certificate’s disclosure of a rape charge and acquittal violated Article 8, because the legal framework at the time lacked adequate safeguards against arbitrary disclosure — confirming both that this kind of disclosure occurs and that it has been found legally inadequate at least once. A comparable US mechanism is now partly documented. United States v. Tunick shows federal authorities placing an uncharged protester on a terrorism watchlist over his alleged political association, with the listing then driving an airport interception by a tactical terrorism response team — a formally named watchlist mechanism applied to someone against whom no charge had been brought. It is not a clean analogue of the UK’s soft-intelligence retention: watchlisting drives travel screening and interception rather than certifying retained investigative files for later disclosure, and the broader population of retained, uncharged investigative files is still inferred from arithmetic (large-scale ICAC investigation-to-arrest ratios) rather than confirmed by a released policy document.

Reputational destruction from the investigation itself, independent of any charge, is the best-evidenced of the three mechanisms, anchored by Sir Cliff Richard v BBC and South Yorkshire Police [2018] EWHC 1837 (Ch). Richard was investigated for historic sexual-offense allegations after police tipped off the BBC before a search of his home; he was never charged, and the investigation closed without any charge in 2016. The High Court held both the BBC and the police liable for breach of privacy, awarding £210,000 in damages, and held explicitly that a suspect under police investigation has a reasonable expectation of privacy that naming alone can violate, regardless of the investigation’s eventual outcome. This case directly drove a UK institutional shift toward not naming suspects before charge as professional-practice guidance, though the reform remains contested rather than legally settled, with press-freedom advocates arguing pre-charge anonymity can suppress corroborating victims from coming forward.

Cooperation itself can widen an investigation’s scope beyond the original predicate. The Rawls and Davis records under Encryption key disclosure compulsion show this directly: Rawls’s partial, voluntary cooperation on one device surfaced material distinct from the online-sharing predicate that began the investigation, and the Pennsylvania Supreme Court’s holding in Commonwealth v. Davis was reasoned expressly around the risk that decryption could expose “an unknown number” of files beyond what the government could already show it knew existed. This is the least speculative of the three mechanisms in this section: it rests on the cases’ own factual records and legal reasoning, not on inference from investigative context.

A worked scenario, held to its actual evidentiary status

A person who identifies with the non-offending, self-described “minor-attracted person” or “NOMAP” community, a real category discussed in Non-offending attraction to minors, is falsely accused of CSAM possession. Their device holds no illegal material. But cooperating with decryption could surface messages to a peer-support line, membership in a non-offending support forum, or self-identifying statements about their attraction — none of it a crime, all of it capable of functioning, through the three mechanisms above, as circumstantial “evidence” of propensity, grounds for informal law-enforcement retention, or the substance of a reputation-destroying leak, entirely apart from whether the original accusation was ever substantiated.

Each individual mechanism in this scenario is independently documented: propensity-evidence risk from uncharged conduct is real, though narrower for merely legal material than for an actual prior offense; non-conviction intelligence retention is confirmed UK policy; investigation-alone reputational destruction is established by a fully litigated case; and cooperation-driven scope creep beyond the original predicate is confirmed by Rawls’s own record and Davis’s own reasoning. But no documented case was found combining these elements in exactly this configuration — a false CSA accusation specifically converging with disclosed non-offending-attraction support-seeking. The closest analogues found are distinct in kind: the 2020 Menlo Church case, where a volunteer’s private disclosure of unwanted attraction, absent any offense, became a public scandal once a pastor’s handling of it was exposed, shows attraction-disclosure-alone causing severe reputational harm through an institutional rather than a criminal channel; and Stop It Now’s own published confidentiality policy — anonymous and confidential unless a caller discloses a specific crime or an identifiable risk of harm — confirms the narrow, conduct-triggered mechanism by which a genuinely protective support contact could generate a law-enforcement referral, though not through attraction disclosure alone. This scenario should be read as a plausible, mechanism-grounded construction assembled from independently verified risks, not as a documented pattern that has been observed to occur.

This wiki keeps attraction, as a private psychological fact, categorically distinct from offending, a conduct, and from false accusation, an investigative and evidentiary event, per its own house style. The terminology, the organizational landscape, and the offending-outcome evidence around Non-offending attraction to minors make this scenario’s premises, though not its exact combination, well grounded.

Charges are often stacked, not substituted

In the UK cases, the process-crime charge did not replace other available charges; it accompanied them. “JFL” was also charged and sentenced separately for unrelated passport-fraud and bail-jumping offenses. This matches a documented general pattern: prosecutors add an obstruction or non-disclosure charge where the substantive case is weak, rather than treating the process charge as a clean substitute the defendant can simply choose instead. No case was found of a prosecutor explicitly offering to drop a substantive charge in exchange for decryption, but the general mechanism — cooperation earns a lower sentence through recognized channels, non-cooperation forecloses that path — is standard and would apply to encryption cooperation as readily as to any other kind.

No court has treated refusal itself as proof of the worse crime

One risk the proposition must confront directly is whether courts treat the refusal to decrypt as circumstantial evidence of the very crime the suspect is trying to avoid being charged with, which would defeat the whole strategy. The research did not find a clean adverse-inference doctrine specific to encryption refusal; the litigated question in nearly every US case is whether the compelled act itself is testimonial under the Fifth Amendment, not whether a jury may be told to infer guilt from silent refusal. This remains an open and unresolved area rather than settled law working clearly for or against the strategy.

Sweden removes the dilemma at its root

Swedish encryption and passcode disclosure law is a third structural option this note’s framing does not initially anticipate: a jurisdiction can simply decline to compel disclosure of a password or PIN at all, treating it as protected knowledge rather than a compellable act. Under current Swedish law, refusing to disclose a passcode carries no criminal exposure, so the entire “which bounded process crime should I accept” calculation does not arise for passcode non-disclosure specifically. It resurfaces, only partly resolved, around active data destruction, in How does Swedish law treat duress wipes.

Assessment

The proposition is coherent and multiply grounded, not merely plausible. A known, formally longer sentence for refusing to disclose can rationally be preferred to a known, formally shorter substantive conviction, for at least three independent reasons: collateral consequences that outlast any sentence, a real dignitary distinction between conviction through manipulated words and punishment for open refusal, and a documented risk that cooperation exposes legal but separately damaging material. Each reason has genuine evidentiary support, and none depends on the others.

The proposition’s actual limits are narrower and more specific than a blanket claim that refusal is always the right choice. Whether the bounded-sentence premise itself holds depends entirely on the jurisdiction’s mechanism, as Rawls’s four years in unbounded civil contempt shows. The dignitary reason is real in its philosophical grounding but thin in direct empirical support for measurable outcomes. The disclosure-risk reason is strongest where cooperation could surface an actual prior offense or widen an investigation’s scope, and weaker, though not zero, where the only risk is merely legal but stigmatizing material, since courts have shown real (if imperfect) resistance to treating the latter as propensity evidence. Anyone evaluating this strategy for their own situation needs the specific jurisdiction’s actual mechanism and the specific nature of what cooperation might expose, not a general assumption that refusal is either always safer or always more dangerous than cooperation.

Built on 3 sources (3 external).

Working out connections…