Presumptuous or pluralistic presumptions of innocence? Methodological diagnosis towards conceptual reinvigoration

Abstract

This article is a contribution to interdisciplinary scholarship addressing the presumption of innocence, especially interdisciplinary conversations between philosophers and jurists. Terminological confusion and methodological traps and errors notoriously beset academic literature addressing the presumption of innocence and related concepts, such as evidentiary presumptions, and the burden and standard of proof in criminal trials. This article is diagnostic, in the sense that its primary objective is to highlight the assumptions—in particular, the disciplinary assumptions—implicit in influential contributions to debates on the presumption of innocence. It advocates a methodologically pluralistic approach, according to which definitions of the presumption of innocence are necessarily sensitive to purpose and method. These relationships and their implications are not always appreciated, and are seldom explicitly elucidated. Notably, philosophers (and some legal scholars) routinely treat the presumption of innocence as (in some sense) epistemic, evidentiary or otherwise featuring directly in practical reasoning. This article identifies jurisprudential and practical reasons why legal scholars and practitioners (and possibly others) concerned with criminal procedure and evidence should reject evidentiary interpretations of the presumption of innocence. By encouraging finer-grained engagement with the history and institutional details of common law procedural traditions, the article aims to show why legal scholars might think that philosophical approaches to the presumption of innocence are already methodologically-loaded and, for our purposes, address the wrong questions with deficient concepts.

1 Introduction: method in presumption

The ‘presumption of innocence’ is ubiquitous in contemporary discussions of legal process and criminal justice. It is regularly invoked in policy debates and popular media, and is a constant source of fascination and perplexity for legal scholars as well as for scholars from other disciplines, especially philosophers, who concern themselves with normative standards in legal procedure. Popularity is not necessarily conducive to conceptual perspicacity or analytical rigour. Being widely regarded as A Good Thing, the presumption of innocence attracts inflated rhetoric. Claims that a particular legal doctrine or official practice offends the presumption of innocence might be little more than visceral reactions to perceived—but unanalysed, unexplained and unrationalised—intuitions of ‘unfairness’. In their efforts to be more considered, academic discussions of the presumption of innocence frequently relate it to cognate concepts such as evidentiary presumptions and the burden and standard of proof in criminal adjudication. Indeed, there is an observable tendency to treat (some of) these concepts as synonymous, raising suspicions of conceptual laxity or partly camouflaged argumentative manoeuvres. The presumption of innocence is prone to presumptuous (mis)usage by theorists as much as by politicians or the media.

Unwarranted presumptions about the presumption of innocence are problematic, especially in interdisciplinary discussions. For example, a perfectly serviceable philosophical conception of the presumption of innocence may be maladapted to jurisprudential or law reform requirements, but this incongruity might not be fully appreciated by the philosopher advancing it, or by later glossators (philosophers or anybody else) extending the original conception beyond its competent sphere of application. This kind of problem reflects a familiar and profound challenge of interdisciplinary endeavours. It is difficult to shake off one’s own disciplinary assumptions even when self-consciously traversing disciplinary boundaries. A related difficulty is primarily terminological, and well-known to comparative legal scholars. Cross-jurisdictional legal research is complicated both by the same terms being used with different meanings, and by different terms being used to mean the same thing, in different legal systems (for elucidation, see Chiesa 2014; Nelken 2007). Analogous terminological difficulties stalk cross-disciplinary conversations. The natural default assumption is that competent language users mean (broadly) the same thing when they say (largely) the same thing, but this plausible generalisation wears thin in relation to disciplinary jargon and professional terms of art. Standard disciplinary usage may offer false friendship in interdisciplinary conversations.

These platitudes would hardly be worth mentioning, were it not for the fact that methodological traps and errors infest academic literature addressing the presumption of innocence and related concepts. By ‘traps’ I refer to propositions or approaches that are not inherently erroneous, but lie in wait to snare the unwary. From the perspective of methodological pluralism (which I endorse), definitional concepts, motivations and methods are mutually conditioning. I think of the relational nature of these three primary vectors as the Eternal Triangle of intellectual inquiry (ET) (see Roberts 2017a, b). For the purposes of the present discussion, ET implies that the best or most appropriate concept of the presumption of innocence depends on the purposes for which it is being deployed and the methods that are being used to investigate it. Change the purposes and/or the methods of research and the best definition might need to be adjusted accordingly. Choice of definition in this sense extends to the domain of the inquiry and its ‘problematic’, i.e., the specification and nature of the problem to be investigated or, phrased more generically, the research question(s) to be answered. There are, then, no innocent concepts and no innocent research projects, because all conceptual definitions build in (implicit or explicit) preferences regarding the researcher’s motivations and methods. From the vantage point of methodological pluralism operationalised by ET, two features of the presumption of innocence literature are striking. First, authors hardly ever specify their objectives at this methodological level. Rather, they typically assume that their purposes are self-evident, and shared by other contributors to the discussion, essentially by extrapolation from their own disciplinary assumptions and orientation. Secondly, much of the debate is conducted in the apparent belief that the objective must be to specify the best definition of the presumption of innocence (and related concepts), as though one conception of the concept must be superior for all purposes. The debate is conducted as a tournament between rival conceptions, in which most of the participants (apparently) share the dogmatic assumption that there can only be one outright winner of the contest. But this is a deeply suspicious, and presumptuous, presumption from the perspective of methodological pluralism.

This article is diagnostic, in the sense that its primary objective is to highlight the assumptions—in particular, the disciplinary assumptions—implicit in influential contributions to debates on the presumption of innocence. I have my own views about the best conception of the presumption of innocence (and related concepts), and I try to be clear that they reflect the purposes of an English jurist specialising in criminal procedure. Sensitivity to context does not imply self-imprisonment in an introverted disciplinary silo: my conception of the presumption of innocence might, and hopefully does, have comparative value for legal scholars in other jurisdictions and intellectual value for scholars in other disciplines. But anybody using it for any purpose other than that originally intended should proceed with circumspection and at their own risk, with no general warranty of fitness for purpose.¹ Caveat emptor, as lawyers say. Although I will not be coy when asserting theoretical preferences, this article does not promote my own conception of the presumption of innocence at the expense of potential rivals in any comprehensive way. True to the tenets of methodological pluralism, it merely demands reciprocity. By identifying the unacknowledged (disciplinary) assumptions framing other contributions to the debate and being explicit about my own, I hope to puncture any extravagant claims to unitary solutions whilst steering legal scholars and interdisciplinary researchers away from enticing entrapments, misunderstandings, fallacious reasoning and erroneous conclusions.

2 Framing presumptions

We may begin with two presumptions about the presumption of innocence that are widespread and in my opinion acceptable, subject to some cautionary observations. First, it is generally assumed that the presumption of innocence is normatively significant. Virtually every contribution to contemporary debates grapples with questions of definition and application whilst presupposing that the presumption itself, however so defined or applied, is valuable. Strictly speaking, the value of the presumption must turn on what it actually means or does or signifies in theory or practice. If one reckoned the presumption of innocence ‘a worm-eaten dogma of bourgeois doctrine’² it would not figure in any normative ideal of criminal adjudication. Its residual importance, if any, would only be descriptively sociological. Perhaps the presumption of innocence operates as a mystificatory ideological smokescreen for capitalist exploitation? Even if that were true, pointing it out would be changing the subject, abandoning rather than contributing to normative theorising. Even the radical sceptic or global error theorist must meet normative arguments with normative refutations, or start a different conversation. Given that the presumption of innocence has a very long history rooted in foundational texts of (western) civilization (Cascarelli 1996) and is today embedded in national legal constitutions (Bassiouni 1993, pp. 265–267) and basic human rights instruments structuring modern international relations (Jackson and Summers 2012, Chap. 7; Trechsel and Summers 2005, Chap. 7), it seems reasonable for scholars concerned with the normative underpinnings of criminal adjudication to assume its importance and to get to work on cashing out what the presumption actually does or should entail in theory and practice.

Secondly, it is routinely asserted that terminological usage surrounding the presumption of innocence is inconsistent, incontinent, confused and potentially misleading, so that some preliminary definitional work is required if we are to make any meaningful progress in elucidation, explanation, or practical reform. This is not merely to repeat the familiar methodological injunctions to define one’s terms with clarity and precision, and to employ them consistently. More fundamentally, the worry is that if ‘the presumption of innocence’ is such an amorphous and protean notion that it can mean just about anything (presumptively positive) that a particular speaker wants it to mean, it will lose all critical bite and analytical usefulness. This is the explicit point of departure for Rick Lippke’s recent systematic reconsideration of the presumption:

A bewildering variety of claims have been made about the meaning and implications of the presumption of innocence in criminal law. Given its apparent elasticity, it is natural to wonder whether the presumption is an honorific concept, one that is mostly empty and therefore adaptable to the needs and interests of legal theorists of diverse kinds. (Lippke 2016, 11)

In view of the conceptual liberties being taken by ‘legal theorists’, Lippke concludes that the presumption of innocence needs ‘taming’ for the sake of its own continued vitality. His principal strategy is to cut down the presumption’s scope of operation to its ‘native habitat’ (ibid., 11), in the belief that a concentrated dose will be more effective than a diluted solution:

Instead of trying to make the PI do more work than it is capable of or suited for, I argue that we ought to tame it: We should confine it to the trial context, where what it means, how it functions, and what are the consequences of its rebuttal can be tolerably well-defined and defended…. Only in the trial context does a full-on presumption of innocence, on the part of those tasked with rendering verdicts, have a defensible role to play. (ibid., 4, 9)

There is a great deal of methodological presumption crammed into this statement of theoretical intent, which I will attempt to unpack presently. But first it should be noted how the problem of conceptual incontinence has suggested alternative solutions to scholars with different theoretical and practical agendas.

Writing from the more doctrinally-orientated perspective representative of common law Evidence scholars, PJ Schwikkard shares the general concern that the presumption of innocence should not be mistaken for something that it is not, but her particular worry is that ‘potential definitional difficulties arise if we do not distinguish between those rights that are coherent with the presumption of innocence and the presumption of innocence itself’ (Schwikkard 1999, p. 36). She has in mind, specifically, the right of silence and the privilege against self-incrimination, and it is these other procedural rights that, on her account, need to be insulated from the corrosive implications of conceptual imperialism:

The danger of conflating the presumption of innocence and other separately enumerated rights, is that those rights become vulnerable to the argument that in situations where the presumption of innocence is not applicable, or where the burden imposed by the presumption of innocence has been discharged, then those rights no longer apply. (ibid., 37)

Schwikkard’s preferred solution is to opt for an unapologetically narrow conception according to which ‘the presumption of innocence contains two components: (1) a rule requiring the state to bear the burden of proof and (2) a directive that the burden will only be discharged when guilt has been proved beyond reasonable doubt’ (ibid., 29). On this view, the presumption of innocence just is the decision rule for criminal adjudication specifying a particular allocation of the (ultimate) burden of proof—to the prosecution—and a particular standard of proof, beyond reasonable doubt. Schwikkard is South African, but her interpretation of the presumption of innocence would be entirely familiar to (not necessarily endorsed by) legal scholars across the common law world. In my criminal law jurisdiction, England and Wales, we might describe this as ‘the Woolmington principle’, after the celebrated case in which Viscount Sankey LC declared that ‘[t]hroughout the web of the English Criminal Law one golden thread is always to be seen, that it is the duty of the prosecution to prove the prisoner’s guilt… If, at the end of and on the whole of the case, there is a reasonable doubt… the prosecution has not made out the case and the prisoner is entitled to an acquittal’.³ English legal scholars sometimes expressly endorse the narrow conception of the presumption of innocence advocated by Schwikkard, and I suspect many more do so implicitly and unthinkingly. It should, to be sure, occasion no surprise if disciplinary outsiders approaching the Law of Evidence gained the impression that legal scholars, lawyers and courts routinely or standardly equate the presumption of innocence with the burden and standard of proof in criminal adjudication. Most of us have been guilty of this situationally convenient conflation at one time or another.

But remember: caveat emptor! Such remarks are not necessarily very well considered as normative ethical or legal propositions, and even when they are, their validity may be strictly delimited by their context of application. Jurisprudence should not be mistaken for philosophy at large because it operates under peculiar (and more or less particularised) institutional constraints. One very important set of constraints, which is alien to other ‘scientific’ endeavours and for this reason easily overlooked or underestimated, relates to the jurisdictional nature of law. Principles of political morality necessarily undergo important modifications in the process of their institutional (including jurisdictional) instantiation (see e.g., Gardner 2000; Summers 1997; MacCormick 1985). Criminal adjudication is jurisdictional in two major dimensions pertinent to the present discussion. First, criminal adjudication concerns alleged criminal wrongdoing, not other types of legal claim such as private law, employment or family disputes. Secondly, criminal adjudication is delimited by the criminal laws of particular territorial jurisdictions. Both of these institutional constraints operate in tandem to regulate the validity of statements of, or about, law and legal process. Some high-level normative generalisations apply to adjudication in general. We can say, for example, that judges should be independent and impartial; that fact-finding should conform to rational precepts of evidence evaluation and proof; and that trial is superior to gladiatorial combat as a pacific means of dispute resolution, without specifying particular kinds of legal process. Other, lower-level generalisations are confined to more circumscribed institutional contexts. For example, it makes sense to say that ‘the accused is presumed innocent until proven guilty’ only in the context of criminal adjudication (indeed, we only speak of the criminally ‘accused’: parties to other kinds of legal proceeding are styled ‘defendants’,⁴ ‘plaintiffs’, ‘claimants’, ‘applicants’, and so on). Likewise, some doctrinal generalisations are competent across territorial legal jurisdictions. I have already employed one myself, when I said that Schwikkard’s conflation of the presumption of innocence with the burden and standard of proof in criminal adjudication would be entirely recognisable to common lawyers around the world. My remark leaves open the possibility that this conflation would not be as familiar to civilian jurists or other non-common lawyers. Indeed, I doubt it is: because ‘inquisitorial’⁵ models of adjudication, featuring a proactive judge charged with primary responsibility for truth-finding (with or without the assistance of lay jurors),⁶ are not (fully) compatible with the common lawyer’s conception of a party-allocated burden of proof. Be that as it may, the more general methodological point stands: doctrinal propositions always relate to particular territorial jurisdictions. Some are general enough to apply to all or most of the world’s legal jurisdictions, and these—necessarily, relatively abstract—generalities include fundamental human rights norms and principles of natural justice. Other doctrinal propositions might be limited to a subset of territorial legal jurisdictions, or even to just a single, idiosyncratic jurisdiction. Generally speaking, as doctrinal argument or analysis becomes more fine-grained and institutionally concretised its trans-jurisdictional validity diminishes. Sometimes doctrinal propositions only really make sense when related to other finely-drawn features of their native institutional environment.

It is not self-evident why, or how much, these jurisprudential generalisations matter for debates about the presumption of innocence. It might be thought, for example, that the presumption of innocence must be a trans-jurisdictional legal principle because otherwise it could not feature, in the way that it indubitably does, in global or regional human rights instruments. That thought is demonstrably inchoate. For one thing, it would not settle any question about the potential application of the presumption of innocence outside the ‘fair trial’ context, whereas contemporary theoretical debates concerning the presumption range more broadly across criminal process and beyond.⁷ For another, it invites the retort that the presumption, qua international human right, is nothing more than rhetorical hand-waving in the direction of penal justice. One need not be a strong sceptic about the normative value of the presumption of innocence to point out that international human rights instruments typically assert allegiance to the presumption without actually specifying what it entails.⁸ Moreover, the interpretational jurisprudence of international courts deploying the presumption of innocence has a strong flavour of judicial improvisation, as particular tribunals confront a succession of ‘hard cases’ with fairly unpredictable outcomes and feel their way towards new doctrinal prescriptions in the absence of clear legislative guidance.⁹

Instructive as this judicial experience may be, a point of still greater methodological profundity needs to be stressed. Theoretical discussions of the presumption of innocence characteristically proceed at the level of conceptual and normative generalisations. This is especially, and understandably, true of discussions involving philosophers and jurists, whose normal field of reference tends to be bounded by linguistic rather than jurisdictional borders (hence we can speak intelligibly of ‘Anglo-American legal theory’ and the like) and who—to be blunt—generally neither know nor care very much about the detailed doctrinal specifications of procedural law in particular territorial jurisdictions. However, the Eternal Triangle teaches that every such methodological choice comes with relational baggage incurring freight costs, and that these must be paid for in the currency of selective definitions and motivational pre-commitments. If there are important features of the presumption of innocence, possibly extending to its normative core, that can fully or properly be appreciated only at the level of doctrinal particularity, the best account of the presumption (relative to contextual jurisprudential objectives and methods) will evade general theorising’s grasp. Those theorists who presuppose that the presumption of innocence must be a general normative ideal are already making a big assumption that few, if any, pause to defend.¹⁰

A great deal more argument and explanation are required to work up ET’s abstract methodological question-mark into a well-substantiated prescription for thinking more clearly about the presumption of innocence. Before venturing further in that direction and providing at least the outline of such an argument, let me give a final example to illustrate how rival conceptions of the presumption of innocence are, in part, largely a reflection of their own (presumptuous) methodological assumptions.

As recently as 1990, Jeffrey Reiman and Ernest van den Haag expressed surprise about the lack of serious theoretical engagement with the familiar thought that criminal adjudication should take special care to avoid convicting the innocent. Specifically, Reiman and van den Haag contemplate ‘the common saying that it is better that ten guilty persons escape than that one innocent suffer’, which common lawyers frequently ascribe to Blackstone and to which civilian jurists sometimes attach the Latin tag in dubio pro reo. Although pretty much everyone agrees that convicting the innocent is worse than acquitting the guilty, Reiman and van den Haag mused, ‘it is difficult (and for us so far, impossible) to find systematic attempts to defend the maxim. It is treated as a truism in no need of defense’ (Reiman and van den Haag 1990, p. 226). Yet, Reiman and van den Haag continue, ‘the principle within it is not at all obvious; and since it undergirds many of our criminal justice policies, we should be sure that it is justifiable’ (ibid.).

Reiman proceeds to argue that the maxim is justified by overlapping deontological, consequentialist and contractarian rationalisations. He concedes that each of these arguments requires an element of imaginative supplementation, and even a bit of ‘bullying’ in relation to traditional conceptions of retributive justice (ibid. 231), but maintains that attractive versions of retribution, deterrence and social contract theory all subscribe to the ‘punishing innocence is worse’ maxim. Van den Haag counters that none of these three prominent philosophical traditions supplies a convincing justification and consequently rejects the maxim, along with the traditional requirement of ‘proof beyond a reasonable doubt’ (which he takes to be implied by it), in favour of a contextually flexible standard of proof set at ‘clear and convincing evidence’ for the majority of cases (ibid., 248). Like many critics of ‘proof beyond reasonable doubt’ before and since, van den Haag cannot see any good reason of principle for systematically sacrificing the overall accuracy of adjudication in order to safeguard the innocent, given that acquittals of the guilty also impose social costs (principally, additional future victims of crime owing to the presumptively reduced deterrent and incapacitative efficacy of criminal sanctions).

A notable methodological feature of Reiman and van den Haag’s shared approach is the immediate link they forge between the questioned maxim and a putative ‘punishment constraint’. As they elucidate:

If the task were to prove the literal truth of the claim that it is more than ten times worse to punish an innocent person than to acquit a guilty one, it could not be accomplished. We possess neither moral intuitions nor moral theories which could establish such a specific ratio. However, we can separate the core content of the constraint from the specific ratio proposed. After all, the truism is not a carefully quantified principle but a slogan meant to convey a message quickly and memorably. Shorn of rhetoric, the punishment constraint asserts that it is morally worse to punish innocent people than to acquit guilty ones. This is the part of the constraint that most matters, since, if we came to believe that punishing the innocent was no worse than acquitting the guilty, our criminal justice policies should change in important ways. For example, we might lower the standard of proof of guilt in criminal cases. (Reiman and van den Haag 1990, pp. 226–227)

To be sure, the value of the presumption of innocence seems intuitively connected to the importance of not punishing innocence, at least to the extent that this can realistically be avoided.¹¹ In legal doctrinal terms, this connection is institutionalised in the form of an epistemically demanding standard of proof for criminal convictions.¹² Reiman and van den Haag are, in my opinion, well justified in rejecting numerically quantified interpretations of Blackstone’s maxim, but I have major doubts about what supposedly follows. Even if we thought that punishing the innocent is no worse than acquitting the guilty, it is far from obvious why this should have any practical bearing on the formal standard of proof in criminal adjudication, let alone qualifying as the most logical or emblematic policy implication of rejecting the maxim. What is, to all intents and purposes, an essay in penal theory rapidly loses interest, and analytical rigour, when it comes to the institutional details of criminal procedure.

If the disciplinary lacuna perplexing Reiman and van den Haag effectively boils down to wondering why so little has been written by penal theorists about the presumption of innocence and related concepts, the mystery is largely dissipated. Criminal law and penal theory, like law teaching more generally, have traditionally concentrated on issues of substantive law and their normative philosophical foundations, relegating evidentiary and procedural questions to the margins.¹³ This is (merely) a sociological observation about legal scholarship, amenable to explanatory conjectures but hardly defensible in principle.¹⁴ Meanwhile Evidence scholars, for their part, have continuously engaged with the presumption of innocence without perceiving any need, arising form their own jurisprudential projects, to relate it back to foundational penal rationales.

This break in the chain of argumentation is not merely a curious artefact of academic tribes and their segmented (sub)disciplinary literatures. It reflects significant differences of scholarly motivation and method, as ET should prime us to expect. Moreover, these discrepancies have real-world practical implications. Those who approach the issue from the perspective of penal theory—or normative theorising more generally—will naturally tend to interpret the presumption of innocence as a substantive constraint on convicting the morally innocent, of whom the factually innocent are merely an epistemologically interesting subset. There is no knock-down conceptual barrier to treating the presumption of innocence as a bar to convicting the morally blameless, and some theorists have argued in favour of affording the presumption of innocence this kind of substantive purchase (see, e.g., Tomlin 2013; Tadros 2007). The problem with many such imaginative reinterpretations of the presumption of innocence for anybody seriously concerned with the institutional realities of criminal law is that, however normatively attractive the theory might appear in the abstract, such interpretations are mostly¹⁵ non-starters from a doctrinal point of view. For example, it was entirely predictable that judges would emphatically reject ‘substantive’ interpretations of Article 6(2) of the European Convention on Human Rights and insist that its doctrinal impact is purely procedural (Roberts 2002a). And this is exactly what happened when the issue was expressly litigated, first before domestic English courts¹⁶ and then to the European Court of Human Rights.¹⁷ Philosophers are entitled to argue that all judges are collectively in the grip of normative delusion, but the luxury of traducing authoritative institutional materials is not available to jurists,¹⁸ at least for as long as they want to address their jurisprudential arguments to lawyers and courts. In jurisprudence, unlike normative philosophy, prominent institutional facts cannot simply be ignored or wished away as inconsequential deviation. As the old common law maxim has it, communis error facit ius. Or to restate the same point in more formal philosophical terms, ‘legal truth, like truth regarding the meaning for conventional terms of language, is a function of collective (medium-term) judgment… therefore it is a misunderstanding to believe that legal officials could be globally mistaken about legal matters for significant periods of time’ (Bix 2009, p. 547).

3 A dose of traditional common law thinking

The puzzlement and disorientation expressed by theorists such as Reiman and van den Haag are hardly surprising, in view of the conceptual fog enveloping the presumption of innocence and its terminological cousins. Their claim that the ‘common saying’, preferring ten acquittals of the guilty to one conviction of an innocent, had eluded previous attempts¹⁹ at rationalisation is more contentious. Legal scholars specialising in the Law of Evidence might think that the salient issues have received copious jurisprudential attention spanning several centuries. However, the way in which the presumption of innocence has been conceptualised in traditional common law thinking is not what might be imagined by theorists unfamiliar with pertinent institutional sources. The ‘presumption of innocence’ is both less and more, jurisprudentially speaking, than modern theoretical reconstructions generally recognise.

James Bradley Thayer, the godfather of Anglo-American Evidence law theory, was bemoaning ‘the confusion and ambiguity that hang over the common talk about the presumption of innocence in criminal cases’ (Thayer 1897, p. 185) whilst Queen Victoria still reigned. In his customarily fastidious fashion,²⁰ Thayer set out to restore conceptual order and reinstate doctrinal orthodoxy, in order to staunch ‘that feeble administration of our criminal law which is doing so much in these days to render it ineffectual’ (ibid.). His ground-clearing commences by noting the ancient historical connection in English legal doctrine between the presumption of innocence and certain ‘humane maxims, rules of procedure, and practical adjustments’, prominently including ‘the principle that it were better that a guilty person should be unpunished than that an innocent one should be condemned’ (ibid. 186–187). Thayer took it as self-evident that the general sentiment rather than quantified numbers or ratios—which have fluctuated over time²¹—was the predominant message:

Obviously these phrases are not to be taken literally. They all mean the same thing, differing simply in emphasis—namely, that it is better to run risks in the way of letting the guilty go, than of convicting the innocent. (ibid. 187)

Thayer was at pains to reaffirm the limited doctrinal significance of the presumption of innocence, by way of antidote to ‘our hysterical American fashion of defending accused persons’ which had, so Thayer thought, lent to the maxim ‘an extraordinary stretch’ (ibid. 189–190). What did institutionally authoritative precedents have to say on the matter?

[I]n the last [eighteenth] century and the early part of this [nineteenth century], we shall find very little, indeed almost nothing, about the presumption of innocence. But a great deal will be found, a very great emphasis is placed, upon the rule that a party must be proved guilty by a very great weight of evidence. That is the important thing. And I think it will be found that, in English practice, down to our time, the presumption of innocence—except as a synonym for the general principle incorporated in that total phrase which expresses the rule about a reasonable doubt, namely, that the accused must be proved guilty, and that beyond a reasonable doubt—plays a very small part indeed…. That is the simple, intelligible, plain way in which the presumption of innocence is dealt with in important cases in England. The prisoner is, indeed carefully protected, but his bulwark is not found in any emphatic or strained application of the phrase or the fact of a presumption of innocence. (ibid. 190–1, 192)

The doctrinal accuracy of Thayer’s depiction of English law is all the more remarkable for retaining its currency throughout most of the ensuing twentieth century, essentially right up until the advent of the Human Rights Act 1998, which revolutionised (Roberts and Hunter 2012) this aspect of English criminal procedure. There are only seven reported English cases referring expressly to the ‘presumption of innocence’ for the entire period between the inauguration of official law reporting in 1865 and the mid-1990s, and none of these glancing mentions is jurisprudentially significant or illuminating (Roberts 2007, pp. 415–416). Woolmington itself refers to presuming innocence only in passing. Traditionally, common lawyers have consistently regarded the workaday burden and standard of proof in criminal adjudication, rather than the more pretentiously inflated presumption of innocence, as the main juridical event.

Prior to Thayer, James Fitzjames Stephen had written that ‘the general presumption of innocence… though by no means confined to the criminal law, pervades the whole of its administration’ (1883, p. 438). Stephen characterised the presumption of innocence as a ‘rule’ with several subparts anticipating and broadly corresponding with the Woolmington principle. He added that ‘[i]t is also closely connected with the saying that it is better that ten guilty men should escape than that one innocent man should suffer’, the idea often (misleadingly) expressed as ‘the Blackstone ratio’ in modern US legal literature (see, e.g., Zalman 2018; Allen and Laudan 2008). Stephen, however, rejected the saying as (already in the late nineteenth century) an anachronistic relic of a barbarous legal history. Stephen’s chief objection to any putative Blackstone ratio was pragmatic: he saw no necessary connection between legal procedures well-calibrated to convict the guilty and heightened risks to the innocent. If anything, insisted Stephen, the risks to innocence run in the opposite direction to the aphorism’s supposition:

[I]t assumes, in opposition to the fact, that modes of procedure likely to convict the guilty are equally likely to convict the innocent, and it thus resembles a suggestion that soldiers should be armed with bad guns because it is better that they should miss ten enemies than that they should hit one friend. In fact, the rule which acquits a guilty man is likely to convict an innocent one. Just as the gun which misses the object at which it is aimed is likely to hit an object at which it is not aimed. (Stephen 1883, p. 438)

Stephen was over-generalising. Whether, in fact, ‘the rule which acquits a guilty man is likely to convict an innocent one’ turns on concatenated situational contingencies; it cannot state any universal truth about accuracy in criminal adjudication. Notwithstanding his exaggeration, Stephen was here marking a vital empirical consideration that too many contemporary commentators overlook. The best way to protect the innocent from wrongful conviction may be to try to enhance the reliability and probative value of information presented to factfinders such that apparent guilt, on the evidence, is reliably indicative of actual guilt, in fact. The complexities of that simple-sounding aspiration are enough to fill a library of textbooks on police investigations and criminal procedure. Suffice it for present purposes to note that the discovery and development of powerful new forensic technologies such as DNA profiling have probably done far more to enhance the accuracy of criminal adjudication without materially increasing the risks of mistaken convictions of the innocent²² than any marginal tinkering with formalised proof standards could ever hope to achieve.

Granted that mistaken convictions of the innocent are worse (in some to-be-clarified normative sense) than mistaken acquittals of the guilty, loose talk of a Blackstonian ratio encourages the thought that it might be possible, and useful, to quantify these values and calculate their optimal distribution. The desirable standard of proof will be that which produces the appropriate level of trade-off between one kind of mistake and the other (see, e.g., Lillquist 2002; Laudan 2006; cf. Picinali 2013). However, this putative calculation is not nearly as straightforward as might appear, because standards of proof do not single-handedly determine error ratios in adjudication (also see Risinger 2007). Much also depends on (1) how many of those tried are actually innocent (the smaller that proportion becomes, the less likely it is that an innocent person will be convicted, all else equal); and (2) the effectiveness of trials in discriminating between guilt and innocence (imagine an evidential litmus test for guilt with a miniscule false positive rate: then very few innocent people would be convicted even on a ‘balance of probability’ standard, because anybody the jury convicts would nearly always be correctly identified as guilty²³). Notably, a standard of proof set at 0.91²⁴ probability of guilt will not, as many have erroneously imagined, produce a Blackstone-mirroring ratio of 10 false acquittals for every one false conviction, except under rather stringent empirical assumptions (DeKay 1996). Instrumental, or avowedly consequentialist, approaches to framing the standard of proof in criminal trials run into a host of theoretical, empirical and normative objections (the literature is vast: see e.g., Epps 2015; Picinali 2018; Sunstein and Vermeule 2005; Steiker 2005). Besides, there is scant institutional basis, in English law or in wider common law materials, for treating the standard of proof as any kind of quantified, calculable ratio. Even if—contrary to the historical record—one thought that the numbers in the canonical Blackstone ratio were supposed to be meaningful, no court, judge or legislature has to my knowledge ever argued that 10:1 should be a target to aim for in error distribution. Rather, the more fully articulated thought seems to run along the lines of ‘we would be prepared to tolerate 10 mistaken acquittals for every mistaken conviction, if the former were unavoidably necessary to minimise the latter’. Decision theorists typically assume a more or less hydraulic relationship between true positives and false positives (to convict more guilty you must convict more innocent), and between false negatives and true negatives (to acquit more innocent you must acquit more guilty). This assumption begs the question. Jurists such as Stephen have traditionally denied that tolerating acquittals of the guilty will correspondingly safeguard innocence, as we have seen.

In summary, if your main objective is to insulate innocent accused from wrongful conviction you would be well advised to concentrate on enhancing the reliability of judicial evidence, taking account of the entire process (including guilty pleas),²⁵ rather than attempting to extrapolate institutional reform from a popular maxim that (so far as I know) has no formal doctrinal status in any modern legal system, common law or otherwise. If Blackstone’s maxim boils down to a rhetorical commitment to strive to protect the innocent from wrongful conviction, why is this not best honoured by efforts to increase accuracy across the board so long as traditional procedural protections, including the asymmetric burden and standard of proof at trial, are preserved? At the very least, the policy issue is not engaged unless and until it is shown that efforts to convict the guilty concomitantly increase the risks of convicting the innocent, beyond the truistic—and morally irrelevant—observation that every single additional trial adds to the absolute actuarial risk (just as additional funding for the police or building more courtrooms will presumptively result in a larger absolute number of mistaken convictions, holding all else—including the existing ratio of false positive and false negative verdicts—equal). In terms of more comprehensive theorising, traditional common law thinking has contributed disappointingly little to conceptual elucidation of the ‘presumption of innocence’ and downplayed its significance in doctrinal analysis. The path to unimpeded theoretical reconstruction is apparently open and clear. Yet as we will see, there are jurisprudential roadblocks ahead.

4 Evidentiary versus jurisprudential presumptions

Definitional reset is the theorist’s first reflex when confronted with conceptual muddle. Antony Duff seeks to deliver us from confusion by reconsidering, and reframing, the essential tasks for a coherent and normatively attractive conception of the presumption of innocence. According to him:

If we are to understand the meaning and implications of this presumption (or presumptions)… we must address several questions. What is to be presumed, by whom, of whom? What is the effect of that presumption? What can defeat it? (Duff 2013, p. 170)

These are perfectly logical, reasonable and potentially illuminating questions, posed within the analytical framework of normative philosophical inquiry. But they also contain an unacknowledged presupposition with significant methodological ramifications. The presupposition is made explicit by Lippke, who broadly adopts Duff’s analytical heuristic:

Duff’s succinct statement of the complexities raised by the presumption of innocence anticipates many of the distinctions I draw… I take literally the notion that the PI is a presumption. As such, it is a deliberately adopted perspective on persons suspected of or formally charged with crimes. (Lippke 2016, pp. 1–8)

These innocent-sounding presuppositions invite closer methodological scrutiny and critical diagnosis.

In the following discussion, I will refer to conceptions of the presumption of innocence which require criminal process actors and decision-makers (or, indeed, anybody else) to hold certain beliefs or make particular assumptions as ‘evidentiary’ conceptions. A subset of these evidentiary conceptions is more properly ‘epistemic’, in relating to actual beliefs of those involved in the administration of criminal justice. Evidentiary conceptions may be contrasted, generically, with what I will call ‘jurisprudential’ conceptions, according to which compliance with the presumption of innocence is fundamentally a question of how people should be treated—as if they were innocent—regardless of anybody’s actual beliefs or factual presumptions. One could say that treating someone as if they were innocent is tantamount to treating them as if one believed they were innocent, or involves entertaining such a belief (without actually holding it), or something of that sort. But I do not perceive any good theoretical or practical reason for thinking or speaking in that way. Beliefs of the relevant kind are not doing any analytic or normative work in jurisprudential conceptions of the presumption of innocence. I can treat you as though you are innocent (in every way consistently with how an innocent person should be treated) irrespective of any belief about your guilt or innocence, or even if I confidently believe, perhaps even know, you are guilty. More routinely, criminal justice professionals may neither know nor care whether the accused is factually guilty or not. Discharging their institutionally allocated responsibilities in the administration of justice does not, in general, require police, prosecutors or defence lawyers to form any view about suspects’ or the accused’s factual guilt or innocence (though doubtless there is more to say, from the perspectives of cognitive psychology, strategic institutional design and professional training, about the practical attitudes that we should try to cultivate in our criminal justice professionals in order better to serve the normative value of the presumption of innocence).²⁶

The availability of jurisprudential (non-evidentiary) alternatives demonstrates that the methodological strategy adopted by Lippke and Duff involves a choice. From my jurisprudential perspective, moreover, their choice seems ill-advised. This section explains why evidentiary conceptions of the presumption of innocence exhibit poor fit with authoritative institutional materials in the common law tradition, and impose numerous uncomfortable theoretical compromises. More seriously still, evidentiary conceptions of the presumption of innocence are prone to a persistent type of conceptual confusion that is especially perplexing in evidentiary scholarship and legal practice. From a jurisprudential perspective, it will count heavily against any conception of the presumption of innocence that it operates on legal reasoning like mogadon or mental kryptonite.

Even absent detailed doctrinal expertise, more narrowly epistemic conceptions of the presumption of innocence have always struck me as fishy, for the elementary (metaphysical) reason that beliefs are not directly subservient to the will. Instructing a jury at the start of a trial to believe that the accused is innocent would hardly be calculated to make jurors actually believe it. Jurors will believe what they believe, and if they have any common sense they will realise that the individual standing in the dock didn’t accidently fall into the courtroom from the sky. And given that this is all so obviously true, any judicial direction to the effect that the jury should believe the accused to be innocent would inevitably run the risk of sounding hollow and possibly hypocritical, a mere institutional ritual rather than a serious-minded procedural safeguard, and thus potentially self-defeating as an expression of genuine commitment to avoiding conviction of the innocent.²⁷ In fact, jurors are never told to presume innocence in the epistemic sense. They are generally told that they should not convict the accused unless they are confident he is guilty; that if any plausible interpretation of the evidence supports innocence they must acquit; and that this is how the hallowed presumption of innocence is, in practice, respected in criminal adjudication (if ‘the presumption of innocence’ is expressly mentioned at all in particularised judicial directions).²⁸ The traditional injunction is, in essence, to suspend belief until all the evidence and arguments have been heard in the trial, signalling ‘emphatic caution against haste in coming to a conclusion adverse to the prisoner’ (Stephen 1883, p. 438).

Larry Laudan has devoted considerable time and energy to demonstrating that it makes no epistemological sense for the jury to believe that the accused is factually—or in his terminology, ‘materially’—innocent, and concludes, with a flourish, that ‘innocence_m is an idle wheel in most of the machinery of justice’ (2006, p. 98).²⁹ From the perspective of criminal jurisprudence, this conclusion is trivially orthodox. Criminal proceedings, from investigation to trial, are by design structured exercises in identifying the guilty. Preliminary investigative measures may be authorised on the basis of limited information (‘reasonable suspicion’, ‘probable cause’), but more coercive interventions propelling suspects deeper into the adjudicative process are generally predicated on higher, more demanding epistemic thresholds. Prosecutions in England and Wales cannot lawfully continue unless there is a ‘realistic prospect of conviction’,³⁰ which comes close³¹ to saying that the reviewing lawyer must regard the accused as probably guilty or the case should be dropped. Within this structured procedural environment (and assuming broadly effective policy implementation), if individual jurors were asked at the beginning of a criminal trial to state truthfully what they believed about the accused’s guilt or innocence—a question which, in reality, jurors are never asked—the only rational response would be that the accused is probably guilty. Laudan, who mostly discusses secondary commentaries and rather selective primary legal sources, treats these prosaic facts as revelations with potentially damaging implications for conventional understandings of the presumption of innocence. But there is nothing troubling to see here. Any ostensible difficulties are artefacts of conceptualising the presumption as epistemic, and simply fall away when the presumption of innocence’s epistemic pretentions are renounced, in accordance with common law orthodoxy.

Lippke does not regard prosaic features of the administration of criminal justice (with which he is admirably well-acquainted) as problematic for his preferred version of an epistemic (or, more broadly, evidentiary)³² presumption of innocence. His solution to the challenge of conceptual incontinence is to endow the presumption of innocence with real practical bite by significantly narrowing down its scope of application. Lippke argues that ‘[o]nly in the trial context does a full-on presumption of innocence, on the part of those tasked with rendering verdicts, have a defensible role to play’ (ibid. 9).³³ Thus, the tactic is to save the presumption of innocence by ‘taming’ it: ‘My intention… is to develop and defend a systematic account of the PI, one that assigns it an important and intelligible role in the criminal justice system. But it is a limited role, far more so than many legal scholars would assign it’ (ibid., 31). Lippke thus positions himself as, in my (non-judgemental) terms, a conceptual dogmatist. He is constrained to say that when scholars, criminal justice professionals, policymakers or ordinary people invoke the ‘presumption of innocence’ in other contexts they are always³⁴ confused or abusing terminology, or at any rate not employing the concept in the most effective or appropriate way. In this, Lippke parts company with Antony Duff, who is prepared to contemplate employing different conceptions of the presumption of innocence, in different contexts, and for different purposes.³⁵ But Duff is only a half-hearted pluralist about the presumption of innocence, as we shall see.

It strikes me that Lippke pays a heavy price for stipulating a narrow, trial-centric conception of the presumption of innocence. It is not obvious to me why lawyers or laypeople are necessarily making any kind of conceptual error or linguistic faux pas when they invoke the presumption of innocence in relation to, for example, pre-trial investigative measures,³⁶ systematic data-gathering for the purposes of criminal intelligence,³⁷ or post-trial conduct by officials.³⁸ If the gist of the presumption of innocence is a prohibition against treating people who have not (yet) been convicted of a criminal offence as though they have already been convicted of it, it is natural to think that a broad spectrum of activities might potentially breach the presumption. To be sure, the presumption’s precise contours and implications will need to be elucidated in relation to particular procedural phases and types of activity, and—to my knowledge—this can be done only through context-specific normative and jurisprudential argument, with proper regard for institutional factors and jurisdictional diversity. There can be no peremptory victory on disputed points of political morality, as though invoking the presumption of innocence amounted to ‘playing the joker’ in formulating criminal justice policy. So I would go along with the general thrust of Lippke’s complaint that the presumption of innocence does not really stand for all that some of its proponents would like it to, but not with his proposed solution.

Instead, I would characterise the presumption of innocence as a second-order normative rationale for more particularised institutional doctrines and practices, especially those concerned with structuring criminal investigations, prosecutions, trials and post-conviction procedures. This conceptualisation is congruent with Stephen’s somewhat enigmatic remark, previously quoted, to the effect that ‘the general presumption of innocence… though by no means confined to the criminal law, pervades the whole of its administration’ (Stephen 1883, p. 438).³⁹ In this guise, the presumption of innocence plays an important diagnostic role in flagging up situations in which its core value would be imperilled unless government power is adequately constrained—either wholesale through legislation, or on a case-by-case basis through the faithful application of existing legal norms and the wise exercise of official discretion. As it happens, I also largely concur with Lippke’s insightful discussion of particular aspects of criminal law and process and with the substance of his analysis; except that Lippke thinks it necessary to invent new names for principles associated with this broader sphere of penal regulation, such as ‘the societal presumption against punishment’ (Lippke 2016, p. 59, pp. 72–74) and the ‘nonpresumption of guilt’,⁴⁰ whereas I would say that we already have perfectly good names for the most significant principles, including names well-rooted in existing procedural traditions and institutional practice (e.g., the Woolmington principle; in dubio pro reo; the presumption of liberty⁴¹; the mens rea principle⁴²). Perhaps the most revealing illustration of the exorbitant costs of Lippke’s conceptual dogmatism is the logical implication that the presumption of innocence cannot be a human right, in part because ‘it must be defeasible in the pre-trial context, or else police and prosecutors will be unable to do their jobs’ (ibid. 49). This means that, on Lippke’s account, the right announced by Article 6(2) of the European Convention on Human Rights (and transposed into English law by the Human Rights Act 1998), that ‘[e]veryone charged with a criminal offence shall be presumed innocent until proved guilty according to law’, is not actually part of the presumption of innocence, or only a counterfeit version of the authentic normative concept. North American legal theorists might not be unduly troubled by this discrepancy, but English criminal lawyers plainly should be.⁴³ Presumably, we would have to work with (at least) two concepts of the presumption of innocence, the normatively justified one and the European Court of Human Rights’ specious doctrinal doppelgänger.⁴⁴ All of this trouble stems directly from Lippke’s epistemic/evidentiary presumption. As soon as we ditch the idea that anybody must actually presume anything about innocence, there is no difficulty in saying that the scope of the presumption as it applies in pre-trial (or post-conviction, etc.) proceedings must be circumscribed by the pragmatic realities and other normative strictures of criminal process. Indeed, I am not sure why this elegant solution is not already suggested by Lippke’s decidedly odd-sounding talk of ‘rebutting’ the presumption of innocence.⁴⁵ It seems to me that the presumption of innocence is something that, in ordinary parlance, one may be afforded (or denied), like trial by jury or a second vote on Brexit, but not something that can be ‘rebutted’, ‘disproved’ or ‘weighed’ evidentially.

Which brings me back to mental kryptonite and Duff’s half-hearted pluralism. A major reason why the topic of ‘presumptions’ is traditionally regarded as one of the most mind-bending and unsatisfactory chapters of the Law of Evidence (Morgan 1937; Thayer 1889) is the ease with which legal or evidentiary presumptions of fact can be confused and confounded with factual inferences. Legal presumptions are evidentiary devices directing or permitting a factual inference to be drawn where that inferential conclusion would not, or possibly might not, otherwise have been reached by ordinary inferential reasoning based on evidence presented in the case. Legal presumptions enable factfinders to bridge epistemic gaps, for a variety of contextual policy reasons. They authorise verdicts, by process of law, that could not (or might not) otherwise be achievable on the proven facts. Although the conceptual distinction between legal presumption and factual inference is sharp and clear, there is a marked propensity—possibly owing to linguistic laxity, or simply the intimacy of their interrelationship—for lawyers to get presumptions and inferences muddled up in practice. It is not difficult to point to notorious instances of this confusion in legal reasoning,⁴⁶ and it can produce all manner of mischief when it takes hold—including potentially undermining the safety on appeal of an otherwise sound conviction or contributing to a miscarriage of justice at trial. Crucially for present purposes, the presumption of innocence is not traditionally regarded in Evidence law theory as an evidentiary presumption. Thayer fulminated against lawyers’ failure to respect this vital distinction⁴⁷; and to this day, authoritative textbooks on the Law of Evidence routinely caution against treating the presumption of innocence as an evidentiary or true legal presumption (e.g., Munday 2018, pp. 134–135).

Duff does not acknowledge, let alone employ or respect, the conceptual distinction embedded in traditional common law thinking between (in my terminology) ‘evidentiary’ and ‘jurisprudential’ presumptions. He treats ‘presumptions’ as unitary entities, that can in principle relate to any object. So we can speak, in Duffian terms, of presuming ‘innocence’ or of any other contested fact⁴⁸ in litigation, including facts constitutive of an offence. On this view, ‘[t]he official or body who is to reach the verdict (the judge(s), the jury) is to presume, of the defendant, that he is innocent of the offence for which he is being tried. The presumption protects the defendant against conviction and punishment for that crime unless it is defeated; what defeats it is proof of guilt to the requisite standard’ (Duff 2013, 170, footnote omitted).⁴⁹ In consequence, Duff’s account of the presumption of innocence does not merely inadvertently carry the contagion of mental kryptonite, it positively trades on eliding the (jurisprudentially essential) distinction between presumptions and inferences. The following passage exemplifies Duff’s general theoretical reconstruction, applied to two illustrative⁵⁰ sets of criminal offences in English law (selling adulterated food⁵¹ in the first example, and official corruption⁵² in the second):

The Food Safety Act gives legal force to an expectation that shopkeepers who sell food should not only take appropriate precautions to make sure that the food they sell is safe; but should also, if they do sell food that is unsafe, be ready to explain what happened, and to offer evidence that they had taken appropriate precautions. Given such expectations, proof that a shopkeeper sold unsafe food creates a presumption that she was negligent, and defeats the PoI; the burden then lies on her to rebut that presumption of guilt by offering evidence that suffices, if not rebutted, at least to create a reasonable doubt about whether she was negligent. Similarly, the Prevention of Corruption Act formalizes an expectation of civil servants that, if they accept gifts from contractors, they must be ready to explain (to offer evidence) that the gift was not corrupt: given that expectation, proof of the receipt of such a gift creates a presumption of corruption, which it is up to the civil servant to defeat, by offering evidence that suffices to create a reasonable doubt as to whether it was corrupt. (Duff 2012, p. 55, footnote omitted, my emphasis)

Now, it must be said that legal provisions involving presumptions, ‘reverse onus clauses’ and affirmative defences are complex and typically open to a variety of jurisprudentially plausible interpretations. The relevant law is not always ordered neatly in a single location,⁵³ statutory provisions are sometimes hacked about by piecemeal amendments, and may even be dispersed over several Acts of Parliament (as in Duff’s corruption example). Even very experienced and knowledgeable jurists can disagree about the proper construction of criminal statutes.⁵⁴ Without entering into the substance of jurisprudential controversies regarding particular offences or defences, I will simply observe that: (1) Duff’s three (highlighted) references to ‘presumptions’ are all strictly superfluous, because the relevant statutory provisions can be interpreted as establishing substantive criteria of liability subject to affirmative defences without presuming anything; (2) the presumption of innocence is ‘defeated’ in these examples only in the trivial, and explanatorily superfluous, sense that the presumption is always by definition exhausted when all offence elements have been proved to the requisite standard authorising a guilty verdict; and (3) interjection of phantom ‘presumptions’ into the analysis invites the very conflation, of presumptions and inferences, that common lawyers have traditionally strained to avoid. The imagined ‘presumption of guilt’ is especially problematic, since this ‘presumption’ has a flavour of prejudgement that could never be squared with any jurisprudentially acceptable presumption of innocence. If anything, the doctrinally orthodox reconstruction would be that the prosecution’s proof of specified offence elements, to the requisite standard, authorises an inference of guilt. That inferential conclusion may be rebutted by proving the affirmative defence. Specifically with regard to these two illustrative provisions, however, even this more benign reinterpretation seems superfluous. It would be more elegant, as well as arguably doctrinally superior, to say that proof of specified offence elements constitutes the proof of the relevant offences, in the absence of proof of non-liability. Statutory presumptions qualifying material offence elements operate—at least sometimes—to reconfigure the prosecution’s probative task by, in effect, dispensing with proof of those elements,⁵⁵ even when such presumptions are legally rebuttable.⁵⁶ Within the context of a contested trial, neither inferences nor presumptions need be made in proving Duff’s illustrative crimes. This point appears particularly clearly in relation to the corruption offences qualified by Section 2 of the Prevention of Corruption Act 1916, which—prior to its fairly recent repeal—‘deemed’ payments of money, gifts or other consideration to be corrupt unless and until proved otherwise, irrespective of inferential reasoning (or epistemic presumption). ‘Deeming’ is the language of legal stipulation shorn of any epistemological or evidentiary dimension, and in fact involves no express statutory ‘presumption’ at all.⁵⁷

What is going on here? My methodological diagnosis proceeds in three steps. As a first clarification, Duff seems to approach the issue from a primarily legislative perspective, external to the dynamics of litigation. The thought may be that the statutory provision itself (‘the law presumes…’), or perhaps Parliament when enacting it, is making a presumption that the acts prescribed are sufficiently culpable to warrant criminal sanctions in the absence of proof to the contrary. Such personification of legislative reasoning is intelligible, but hardly very illuminating for those working with such provisions. If the point is that some criminal law offences are excessively widely-drawn or target behaviour that is not morally culpable, why not just say that? Subsequently, however, Duff appears to slip into describing how a fact-finder might reason in an individual case, e.g., by ‘presuming’ one factual component of liability (negligence, corruption) unless this ‘presumption’ is ‘rebutted’. This elision calls for a second diagnostic clarification. It is not unusual for non-lawyers addressing legal issues to attempt to reconstruct the fact-finding process,⁵⁸ but this is not—strange to say—how legal practitioners and procedural scholars in the common law world generally approach criminal evidence and procedure. Fact-finding in common law criminal adjudication is paradigmatically a blackbox activity undertaken by lay juries or magistrates. Although trial judges do have ancillary fact-finding responsibilities in criminal trials (Pattenden 2009), their main jurisprudential task is to direct juries in relation to the jury’s deliberations and verdict; whilst advocates are primarily in the business of making corresponding submissions to the court, either in the form of rival narratives impressed upon the jury during the course of the trial or in closing speeches or through doctrinal legal and factual arguments pitched to the judge (e.g., regarding the admissibility of evidence). For these, quintessentially juridical operations it is essential that legal ‘presumptions’ are correctly characterised, and never, ever, mixed up with inferences of fact.

The third strand of the diagnosis conjectures an explanation for the half-hearted nature of Duff’s methodological pluralism. There is nothing obviously linguistically extravagant or unconventional about Duff’s deployment of the term ‘presumption’. The OED’s third definition of ‘presumption’ as a noun is: ‘the action of taking for granted or presuming something; assumption, expectation, supposition; an instance of this; a belief based on available evidence’.⁵⁹ Duff’s usage seems fairly standard amongst philosophers,⁶⁰ especially those who are interested in the language, structure and dynamics of argumentation.⁶¹ Needless to say, as a fully committed methodological pluralist, I have no objection to philosophers framing and using concepts for their own philosophical purposes. Far be it from me to purport to judge the legitimacy of theoretical projects in the abstract, or to disqualify—on what basis?—attempts to elucidate the presumption of innocence as a trans-jurisdictional normative ideal largely shorn of procedural tradition or grounding in institutional facts. I cannot see why anybody should find such accounts jurisprudentially compelling, but this is only to restate (recalling ET again) the constitutive influence of motivation and (disciplinary) method on conceptual definition and exegesis. I am certainly not promoting a transdisciplinary conception of the presumption of innocence as uniquely and exclusively prescriptive (cf. Weigend 2013). Conversely, if jurisprudential analysis were completely exhausted by descriptive exegesis and rehearsals of primary legal sources, the fact that positive law conceptions of the presumption of innocence departed from philosophers’ idealist reconstructions, and from each other trans-jurisdictionally, would be wholly unexceptional (and uninteresting) from a normative point of view.

The problem arises, it seems to me, when concepts framed for one set of disciplinary tasks are unthinkingly extrapolated to new contexts and purposes for which they may be ill-suited; that is, when methodological presumptions are presumptuous. Duff, like Lippke, propounds an evidentiary interpretation of the presumption of innocence without ever, to my knowledge, actively considering whether this conceptualisation is apt for particular institutional uses or applications. Are evidentiary (including epistemic) conceptions of the presumption of innocence intelligible to criminal court judges and practising lawyers, for example? Would thinking about the presumption of innocence in this way help legal practitioners to do their jobs better? Would it enlighten policymakers, improve the quality of criminal legislation or advance the cause of intelligent law reform? That evidentiary conceptions lack any firm doctrinal basis, certainly in English law and probably in common law jurisprudence generally, is a threshold reason to doubt their general suitability for legal scholarship’s theoretical and practical projects. When theorists start talking about ‘defeating’, ‘rebutting’ or ‘waiving’ the presumption of innocence (to say nothing of ‘presumptions of guilt’), those schooled in common law jurisprudence must be on their guard. If, to cap it all, a toxic byproduct of the evidentiary conception is intellectual kryptonite (as I am suggesting), legal scholars and practitioners concerned with criminal evidence, procedure and proof have compelling reason to reject it.

5 Conclusions: presumption of innocence lost… and found?

Although Blackstone’s 10:1 ‘ratio’ has colonised common law jurisprudence (especially in the USA) and enjoys broad cultural resonance, it is by no means clear how exactly this idea relates to the presumption of innocence, either as an embedded feature of procedural tradition or as normative political morality. Contemporary theorists such as Reiman and van den Haag, we observed, were surprised to discover the philosophical immaturity of theoretical literature and debates. Inasmuch as theorising has been informed by legal scholarship, the burden and standard of proof in criminal trials have tended to dominate discussion, pushing broader normative and institutional considerations to the margins. In the common law tradition the presumption of innocence is routinely equated with the burden and standard of proof in criminal adjudication. This conflation is understandable in light of lawyers’ doctrinal focus, but intellectually stunted and plainly unsatisfactory from a more theoretically ambitious perspective. The presumption of innocence might be conceptualised more holistically as part of a theory of criminal procedure, which is one component of a theory of criminal justice, which in turn is part of a comprehensive theory of justice—a theory that addresses, and strives to answer, Socrates’ question,⁶² how shall we live (together, flourishing)? In debates beset by terminological laxity and conceptual incontinence, virtually everybody agrees that the presumption of innocence needs ‘rescuing’, or ‘taming, or ‘rethinking’, or at any rate that its meaning, functions and value urgently stand in need of clarification. The best conception of the presumption of innocence, considered from a jurisprudential perspective centrally concerned with criminal adjudication, would properly accommodate pertinent institutional facts, faithfully express the core normative value of the presumption, and steer clear of logical fallacies and predictable reasoning errors (such as the mental kryptonite of confounding legal presumptions with factual inference) known to infect and corrupt legal process. For legal scholars specialising in criminal evidence and procedure, the judicial task of directing juries and the advocate’s job of formulating arguments addressed to judges about how to direct juries tend to be more salient than adjudicative fact-finding itself.

In light of these criteria for sound theorising, I propose that the presumption of innocence is best conceptualised as a mid-level principle of political morality from which a variety of more concrete procedural (and other) doctrinal principles, rules and institutional practices may be derived and rationalised. The normative core of the presumption of innocence as a principle of political morality is the distinctive injustice perpetrated by officially and publicly condemning a person for a crime of which he or she has not (yet) been convicted and of which he or she might be innocent. Concomitantly and partially derivatively, it serves to limit officials’ intervention in life, liberty, security and property in the exercise of lawful powers to prevent, investigate, prosecute and punish crime. The presumption of innocence entails that the formal process of criminal adjudication must not be side-stepped or pre-empted. This precludes officials from treating any person as though they are guilty of a crime when they have not been formally convicted of it. The normative ground of the presumption of innocence lies in the injustice of convicting the innocent or gratuitously hurting any person’s important interests, particularly in liberty, property and reputation. A person is seriously wronged whenever they are falsely convicted of a crime, and that wrong is especially grave when the offence is serious and heavily censured and/or peculiarly socially stigmatic. It does not follow that individuals have a right against wrongful conviction as such, or that anybody is necessarily culpable when the innocent are convicted. It is sometimes just an unfortunate feature of the world that people who are factually innocent can appear very guilty by any realistic standard of epistemic warrant.⁶³ But if it subsequently, perhaps accidentally, transpires that a provable mistake has been made, we do what we can to correct it, even in cases of nonculpable mistaken conviction and even if the only compensation then practically feasible is posthumous exoneration of the wrongly judicially executed.⁶⁴ It is part of the inherent structural asymmetry of criminal adjudication that, whilst some unpunished crimes are treated as bygones, wrongful convictions can never⁶⁵ be allowed to lie.⁶⁶ The serious public wrongfulness of convicting the innocent furnishes derivative justification for a range of procedural (including evidentiary) safeguards against this eventuality, which simultaneously also express official public commitment to minimising the risk, some or many or all of which may be brought under the capacious rubric of the presumption of innocence, conceptualised as a major structural girder in the normative architecture of criminal justice.

This suggested reconceptualization remains to be tested systematically and more comprehensively against pertinent institutional facts. The pertinence of particular institutional facts in turn depends on the nature of the inquiry and its level of institutional generality or concretisation, for example, whether one wishes to theorise about the presumption of innocence as a trans-jurisdictional principle of ‘Anglo-American jurisprudence’ or as a normative basis for doctrinal English law. As my sympathies lie towards the doctrinal end of jurisprudential theorising, a full defence of my proposed conception of the presumption of innocence would require extensive engagement with institutional sources and at a level of doctrinal technicality more refined than any single article could offer. This would be a subtle and complex jurisprudential undertaking, of a kind that few theorists ever contemplate. It is no simple matter to reconstruct contemporary English’s law reception of foundational doctrines from the historical sediment of ramified common law traditions. In a common law constitution even the status of an evidentiary norm as ‘constitutional’ and the practical implications of that designation, if any, are matters of jurisprudential controversy (see Roberts 2012; cf. Stein 2008). But even the schematic outline sketched in this article establishes some nearly axiomatic fixed points.

Firstly, English law emphatically endorses the presumption of innocence as a common law principle, a central component of the ‘overriding objective’ of criminal procedure,⁶⁷ and an institutionally enumerated human right. Yet secondly, the demands of institutional ‘fit’ are not as exacting as one might anticipate, given the normative significance of the principle, because English law has generally concentrated on concretised derivative doctrines specifying the burden and standard of proof (the Woolmington principle), leaving the presumption of innocence in the background as a comparatively underspecified normative justification for institutional practices. To this extent, the jurisprudential conception of the presumption of innocence is amenable to theoretical development and reconstruction. Thirdly, the presumption of innocence may support a range of evidentiary doctrines and institutional practices including particularised evidentiary presumptions, but it does not itself have any direct evidentiary (including epistemic) applications. As Evidence specialists have always maintained, ‘the presumption of innocence’ is not that kind of presumption. Fourthly, the presumption of innocence recognised in English legal doctrine is procedural rather than substantive. Although normatively reconstructing common law procedural traditions is never as easy as simply reporting what appellate judges have declared in precedent cases (a practical impossibility, since common law jurisprudence is endemically conflicted, gappy, and sometimes mistaken), when national and international judiciaries are saying exactly the same thing,⁶⁸ when there are strong institutional reasons for holding the conceptual line, and when there are no authoritative sources or institutionally cognisable arguments for revision, criminal jurisprudence must treat the law as clear and settled for the foreseeable future and work within its doctrinal parameters.

Rather than attempting to defend a particular, necessarily controversial, conception of the presumption of innocence, however, this article’s primary objectives were more modestly diagnostic and methodological. Under the rubric of the Eternal Triangle, according to which the conceptual definitions (what?), motivations (why?) and research methods (how?) of any intellectual inquiry are mutually conditioning, we can see that no proposed conception of the presumption of innocence is itself methodologically innocent. Each comes with its own baggage and costs. Furthermore, the puzzles and difficulties attending particular conceptions are often attributable to selected features of the original set-up. They are artefacts of the researcher’s methodological design. This is not necessarily a matter for criticism, provided that definitions, domains and disciplinary methods have been chosen deliberately and are well-calibrated to their intended purposes; or in other words, provided that they are not merely unexamined, and possibly presumptuous, presumptions.

Finally, a pluralist approach to the presumption of innocence relativises theoretical endeavours in a way that should automatically problematise any claim to have formulated the one, true, all-purpose normative conception. Methodological pluralism is the ultimate antidote to any presumptuous conceptual dogmatism. Pace Lippke and Duff, the starting point when ‘ask[ing] about the meaning of the PoI and its proper role in the criminal law’ is not the already theoretically-loaded series of questions ‘what must be presumed, by whom, about whom, to what effect; what can defeat that presumption; and when (if ever) the presumption need not be made’ Duff (2012, p. 51). This is not a neutral or inevitable point of origin for theoretical inquiry: for why are we bound to accept that answers to these questions are necessarily the answers we seek, whoever ‘we’ are? On the assumption that the goodness of concepts must be substantially if not wholly constituted by their instrumental (including explanatory) value, the prior, methodological questions should always be: for whom is the presumption of innocence being formulated, what is its disciplinary or operational domain, and what practical (including normative) purposes does and should it serve? For those with predominantly jurisprudential interests and concerns, the best conception of the presumption of innocence must answer to the pragmatic dictates of legal reasoning and address—through accommodation or mitigation—unavoidable institutional facts. Those with different, or equivocating, theoretical agendas owe us some further explanation of what their concepts are for, whom they think they are talking to, and to what end.