Quaestio facti. Revista Internacional sobre Razonamiento Probatorio
Quaestio facti. International Journal on Evidential Legal Reasoning
Sección: Conjeturas y refutaciones
2025 l 9 pp. 241-250
Madrid, 2025
DOI: 10.33115/udg_bib/qf.i9.23149
Marcial Pons Ediciones Jurídicas y Sociales
© Giovanni Tuzet
ISSN: 2604-6202
Recibido: 04/05/2025 | Aceptado: 26/06/2025 | Publicado online: 30/06/2025
Editado bajo licencia Reconocimiento 4.0 Internacional de Creative Commons
THE BARD STANDARD: FROM HISTORICAL SOURCES TO NEW CHALLENGES. A COMMENT ON DELLA TORRE
Giovanni Tuzet
Bocconi University, Milan
Abstract: The work comments on Della Torre’s essay about the evolution of the reasonable doubt criminal standard of proof. By going through the main claims of the essay, which is both historical and theoretical, the work contends that the essay clarifies (and hopefully soothes) some badly framed discussions on that standard, furthermore that it provides valuable reasons to prefer the reasonable doubt formulation over some revolutionary ambitions, and that it sheds light on the reasons for taking the standard not only in the strict reading of an evidentiary threshold but also in the broad one of a method or even a principle of criminal justice. Finally, the work suggests some refinements to Della Torre’s arguments and addresses the tension between the reasonable doubt safeguards and the efficiency of the criminal justice system.
KEYWORDS: criminal justice; evidence; reasonable doubt; standards of proof.
SUMMARY: 1. AN AMAZING WORK.— 2. THE MAIN CLAIMS AND CHALLENGES.— 3. REFINEMENTS AND FINAL QUESTIONS.— REFERENCES.
In Taking the Evolution of the Standard of Proof for a Criminal Conviction Seriously, Jacopo Della Torre (2025) provides an account of the criminal standard of proof consisting in being beyond a reasonable doubt about the accused’s guilt. The account is amazing. It is so in that Della Torre provides an extraordinary piece of scholarship, at the same time historical and theoretical. It is extraordinary in this first sense, since it is rare to find works that deal with historical and theoretical questions with equal clarity and acumen. And it is extraordinary for the remarkable level of detail and perspicuity with which it deals with the subject under consideration.
The title of the work suggests that there has been a sort of “evolution” in the way the criminal standard of proof has been conceptualized and operationalized. The steps of this evolution can be traced in both common-law and civil-law systems, contrary to the opinion that the beyond a reasonable doubt (BARD) standard is essentially a product of the former. Della Torre identifies and examines a large number of historical sources that allow him to reconstruct that process, focusing in particular on Roman-canon law, which in general the English-speaking public knows only superficially. It is a major merit of this piece of scholarship that it makes available to the international audience several sources of information that are not easy to find and assess.
Another major merit of the work is that it hopefully puts an end to some badly framed discussions about the BARD standard and similar standards of proof in criminal matters. As the work shows, it is distorting to treat the BARD standard as a purely subjective standard of proof centered on the mental state of the decision-maker. If there are factfinders and commentators who take it this way, they are simply wrong. If only because the standard does not speak of any doubts, but of reasonable ones. Taking it as licensing any doubt, or the overcoming of any doubt, is simply wrong. Something similar applies, even if to a lesser degree, to the intime conviction French standard, which concerns the factfinder’s mental state given the trial evidence, not regardless of it. No legal system has ever been so perverse as to make criminal proof depend on the mental state of the individual decision-maker regardless of the available evidence and its probative value. After recognizing this, and taking into account the global role that the BARD standard now has, it is time to set the discussion in new directions.
In the following I proceed through the main claims of the work and notably contend that Della Torre also provides valuable reasons to prefer the reasonable doubt formulation over some revolutionary ambitions, and that the work sheds light on the reasons for taking the standard not only in the strict reading of an evidentiary threshold but also in the broad reading of a method or even a principle of criminal justice (§ 2). This brings to the fore a set of new challenges, especially tied to a broad understanding of the standard. Finally, I recommend some refinements to Della Torre’s arguments and address the tension between the reasonable doubt safeguards and the efficiency of the criminal justice system, suggesting to adopt in particular a flexible understanding of the standard (§ 3).
Let me now summarize Della Torre’s (2025) work and comment on it by going through its main claims. It starts by addressing the nature and functions of standards of proof (p. 157ff), and divides them into three macro-categories, namely: (a) “propulsive standard”, used to decide whether to prosecute or to proceed from pre-trial to trial; (b) “incidental standards”, used at particular sub-stages of the proceedings; and (c) “decision-making standards in the strict sense”, used for the final decision on conviction (p. 159). BARD as traditionally understood belongs of course to the third macro-category. The historical reconstruction provided by Della Torre, from Roman and Roman-canon law onwards, shows that Western legal systems have been constantly concerned with the definition of a demanding standard of proof in criminal matters, to be reached either with the satisfaction of legal proof rules or with an appropriate belief confidence in the context of free proof systems. But interestingly one step in the evolution was motivated by the difficulty to obtain full proof under the legal proof rules, when fewer convictions than desirable were the case (p. 166-167). This fueled a discussion on the role of “unquestionable circumstantial evidence”, and on the “extraordinary punishment” compromise solution, consisting in less severe punishment when the evidence was telling but not up to full proof (p. 167-169) 1.
The “moral certainty” and the “BARD” standards resulted from that process, which had significant theological components and a kind of irony in it (p. 171ff). In some late modern Italian jurisdictions, in particular, the standard of full proof was lowered to face particular crime waves (p. 174ff): the irony of the story is that something like the BARD standard was introduced to make criminal conviction easier, not harder to obtain as we now think when we conceive of BARD as a demanding standard (certainly more demanding than the civil standard of the preponderance of evidence, or proof on the balance of probabilities). Della Torre mentions among other things a 1743 decree of Pope Benedict XIV (p. 175), recommending conviction when the circumstantial evidence removed from the mind of the judge “any reasonable hesitation that the crime might otherwise have been committed” 2; this, according to Della Torre, “marked a step backwards with respect to the classical structure of the Roman-canon systems of proof”, since “the difficulty of reaching the standard of full proof was such that, in order to meet the greater repressive demands of modern absolute regimes, a downward adjustment of the decision-making rules for a criminal conviction was imposed” (p. 176).
That “hesitation” had to be “reasonable”, and this made the standard different from the French post-revolutionary idea of conviction based upon the intime conviction of the factfinder (p. 180ff), which did not mention reason or reasonableness and by the way was inspired by the English system of jury trial. Della Torre observes that such “intimate conviction” standard still has a double function of standard of proof and criterion of evidence assessment, and that, while on the one hand it only makes reference to the internal forum of the individual—with no indication of how firm the belief must be (p. 181)—it is on the other hand “inextricably linked” to the trial evidence (p. 183). It is unfortunate, I would add, that many scholars continue to depict it as a purely subjective standard of proof, disregarding not only its historical sources but also the normative texts that link it to the trial evidence presented against and in favor of the accused 3. That uncharitable account of the French standard, rather blindly and sometimes stubbornly iterated in the literature, should come to an end.
Then Della Torre focuses on the specific development of the BARD standard, as it took over moral certainty in the Anglo-American world and beyond (p. 185ff) and is nowadays challenged in England by the “being sure” standard, given the troubles that juries have in understanding its content (p. 189). This is indeed a problem of the BARD standard. But such doubts on the reasonable doubt standard do not necessarily invite lengthy articulations of the standard of proof, since according to some it is unwise to elaborate on it (p. 190) and the “being sure” formula is certainly simpler than the BARD one and basically conveys the same content (note in this sense that the “being sure” option is just revisionary, not revolutionary).
So the present situation is this: on the one hand, BARD is becoming a global standard, adopted also by civil-law jurisdictions either by legislation (with possible constitutional coverage) or as a matter of case law, and utilized by international courts too (p. 190ff); on the other hand, there are concerns over its reading and understandability, especially when juries and not judges are supposed to understand and apply it (p. 202).
Facing that situation, some views that one could qualify as “revolutionary” call for a replacement of BARD: being hard to understand (by juries) and boiling ultimately down to the mental state of the factfinder (as a subjective standard), it would be better replaced by a standard centered on the epistemic warrant for accepting a given hypothesis (taking this warrant as an objective standard) 4. Della Torre finds some merit in such revolutionary views, but at the same time he resists them. His reservations focus on the inherent vagueness of language (which would not spare the revolutionary proposals), on the success of BARD in the present world, and on “the fact that many of the most advanced proposals for new standards are very similar to the rational reinterpretations of the BARD standard actually used in some legal systems” (p. 198). The last point is particularly important, since it refers not so much to the succinct ways in which the standard is explained to juries in judicial instructions (when these are legitimate), as rather to the ways in which judges interpret its formula in providing the reasons for their decisions (especially in civil-law jurisdictions, of course) 5. In brief, one thing is the standard applied by a jury rendering an unmotivated verdict, another is the standard applied by professional factfinders such as judges giving reasons for decisions. The evolution that BARD is undergoing also consists in moving from a standard used in jury trials to a standard used by courts that give reasons for decisions. In this regard, the motivation for revolutionary ambitions is significantly weakened.
More specifically, Della Torre advances the following three reasons to prefer BARD to alternative formulations:
First, it does not refer linguistically to potentially absolute mental states (such as moral certainty or the English sure standard), and this fits well with the fallible nature of evidential reasoning. Second, it does not contain explicit references to the internal forum or to morality, but to human reason. Finally, by requiring the “exclusion” of reasonable doubt, it lends itself to interpretation from a perspective related to eliminatory induction, i.e. the elimination by progressive evidence of the possibility that there are elements capable of founding a hypothesis on the disputed facts for which the defendant may be considered innocent. (p. 198, fn. 198)
The last point is of particular importance and deserves special comment. The fact that for a decision against the accused one has to exclude reasonable doubts on the accused’s guilt favors a reading of the standard that mandates a comparative assessment: one has to compare the guilt hypothesis advanced by the prosecution to other accounts of the evidence that are compatible with the accused’s innocence. If such alternative accounts are ruled out as unreasonable or unsupported by the trial evidence, whereas the evidence supports the guilt hypothesis, only the latter survives the “elimination” process. For Della Torre this means taking BARD as a “reasoning procedure” or “method of reasoning” (p. 199) centered on eliminative induction (p. 195-196). The idea is not new in fact. Interestingly, it can be already found in English works 6 that refer to the reasoning process of eliminative induction and, though less explicitly and more indirectly, in American works that advocate a “relative plausibility” account of criminal factfinding 7. The idea has been expanded in the Italian literature, in particular, following the introduction of the BARD standard in the criminal procedure code in 2006 and various judicial decisions and theoretical works that have broadened its scope (p. 184-185, p. 191, p. 201). Several Italian scholars presently take BARD not only as a standard of proof but also as a method or criterion of evidence assessment and of evidential reasoning, or even as a principle of factfinding that enjoys constitutional protection as it is a necessary component of a “fair trial” 8. I must admit that when I first found claims of this sort I was puzzled, since I was committed to the idea of BARD as a standard of proof, or epistemic threshold, and nothing else. Now I see the reasons for such an expanded understanding of it, which is particularly suited to a legal context of reasoned judicial decisions. For purposes of analytic clarity one can maintain that there is a strict reading of the formula (BARD as a standard of proof in the traditional sense) and a broad reading of it (BARD as a method of reasoning and assessment, or even a principle of factfinding and of the criminal justice system). This should not come as a surprise when one considers that, to put that differently, there is functional connection between evidence assessment methods and standards of proof 9. Some assessment criteria or methods better fit some standards and vice versa. Trivial as it may sound, the reasonable doubt standard is better served by a reasonable assessment of the evidence than by a system of legal proof rules.
Della Torre endorses that broad reading of BARD and conceives of it in a pluralist framework of (i) reasoning tools, including plausibility considerations and Bayesian techniques, and (ii) procedural remedies and safeguards, including motivated judicial decisions, appeals, and panel decisions rather than individual ones (p. 202-203). In sum, he stresses the importance of the criminal justice system as a whole, to make the presumption of innocence and the BARD operation effective.
As a challenge that that view brings to the fore, one can wonder whether it is possible to be more explicit on the criteria for eliminative induction that best suit the context of legal factfinding. And, as a related challenge, one can wonder about the dynamics that should govern a unified or integrated model of factfinding where a plurality of methods and tools would be at home. It is reasonable to open the door to such diverse things as the use of statistics, of Bayesian tools, and of AI systems along with plausibility considerations, common sense wisdom, and traditional sources of information like human testimony. The difficult part of the story is to combine all of that, and to govern the cases in which different tools or methods pull in different directions. Pluralism is just fine until conflict emerges. To illustrate, what to do if a facial-recognition system (operating on statistical information and the images captured by a camera) indicates that most likely the shooter is the accused while a human witness says it was another person? The solution is rather easy if the human testimony is credible: then the testimony is sufficient to generate at least a reasonable doubt on the accused’s guilt. But what if the value of the testimony is not that clear and the prosecution provides statistical data (e.g. on lighting conditions) which put it into question? How should a BARD principle govern a situation like this?
On AI in particular, Della Torre mentions various concerns that its use in factfinding raises in our present days. Lack of transparency is a major concern, as the information processing leading to AI outputs is most often opaque. Inequality of arms is another concern, as the prosecution usually has better financial resources to access and use the relevant technology. A way to counterbalance this is to attribute to AI evidence a purely circumstantial value. Another idea, which Della Torre takes seriously, is to use “negative rules of corroboration” (p. 205-206) as criteria for evidence assessment, especially to avoid convictions based solely on evidence provided by AI tools. The idea is that uncorroborated AI evidence would not be sufficient for conviction. For example, facial recognition by an AI tool should be consistent with other evidence. Uncorroborated AI evidence, as a result, would not go beyond a reasonable doubt on the accused’s guilt. This should not be read as an obscurantist program. Della Torre recommends no ban on AI evidence. Rather, he recommends caution and safeguards:
it would only be a question of introducing a limited number of exceptions to the general principle of free evaluation of evidence, which, on closer examination, would be motivated by the desire to safeguard its most profound meaning: that of ensuring that, in the future too, judicial decisions will always be based, at least in part, on rationality and human logic, and not only on that of machines, which is often incomprehensible to us (p. 206).
This is a reasonable and sensible attitude, that neither demonizes the use of AI in matters of evidence and proof nor uncritically welcomes its outputs.
To conclude this comment on a different note, after praising Della Torre’s work in many respects, let me first suggest some refinements that would make it more precise, and then ask an additional question, or better a set of related questions.
As to the refinements, consider that Della Torre takes standards of proof as rules for decision (e.g. 2-4) 10. This is an abridged claim that one can frequently find in the literature. But strictly speaking it is wrong. Standards of proof are standards, not decision rules. Rather, they are incorporated in decision rules. The rule that mandates conviction when guilt is proven beyond a reasonable doubt is a decision rule whose antecedent component incorporates that standard, as a rule for the factfinder: if guilt is proven beyond a reasonable doubt, then convict. Consider next that “the factual assertions made by the parties in legal proceedings” are not “mere hypotheses”, as Della Torre has it (p. 157). The litigated points amount to hypotheses when seen from the factfinders’ perspective, true. But from the first-person perspective of the accused they are usually more than that: by asserting “I was elsewhere”, for instance, the accused is not making a hypothesis but claiming instead that things were definitely so (and the accused knows if the claim is true of false). Additionally, and more generally speaking, the parties’ stories include more than hypotheses and litigated points: they comprise statements about the context and elements of common knowledge without which the relevant hypotheses would be floating in the air. Finally, as a refinement, consider that a lamentable confusion is not between standards of proof and “legal proof rules” (p. 199, fn. 201) but between standards of proof and assessment criteria or assessment methods in general (not just legal proof ones). Even when claiming that there is a functional connection between them, one should not conflate them.
As to the final question, note that Della Torre briefly criticizes the understanding of BARD as a flexible standard, an understanding suggested by the vagueness of its formula which makes it open to more or less demanding interpretations depending on the seriousness of the criminal charge and its consequences. The intuition is that the more serious the charge, the more demanding the standard should be. Those who have a pragmatist inclination do not dislike this reading, but Della Torre contends that the same protection of the presumption of innocence should be granted to all defendants (p. 199, fn. 202). This sounds unrealistic in fact. Factfinders likely interpret the formula according to the stakes. In principle, following Della Torre’s understanding, a minor criminal offense should be proven BARD as a major one. Shoplifting, for instance, should be proven BARD as murder. In practice sensible factfinders adjust the requisite threshold to the seriousness of the act and the severity of its legal consequences. Consider in this regard that a major virtue of standards is flexibility. The BARD formula is flexible, as is the “reasonable care” one used in civil cases to assess negligence. A sensible factfinder requires more, compared to a shoplifting case, when the prosecution charges someone with murder. Also the investigation resources that are normally spent on a murder case are greater than what is spent on a shoplifting case. This makes perfect economic sense. And it should not come as a surprise. Nominally the standard is the same, but the seriousness of the case justifies the investment of more resources and a more demanding threshold or proof. It is the same logic that differentiates the preponderance of the evidence standard from the clear and convincing evidence standard applicable to “serious” civil cases 11.
The picture is complicated by the distinction, mentioned above, between propulsive, incidental and decision-making standards in the strict sense. One could maintain that BARD, as a decision-making standard in the strict sense, remains the same all across the board, and what changes is the operation of a propulsive or incidental standard, or the operation of some form of negotiated justice, or at the very beginning of the process the decision to invest or not to invest in investigation. For sure, it seems unrealistic to think that the same standard, with the same content, governs a variety of different cases and activities, in light of the presumption of innocence as Della Torre appears to claim.
However, the attitudes of actual factfinders adjusting the content of the formula to the seriousness of the charge may happen to be simply wrong, if the standard should be understood in the same way given the presumption of innocence and regardless of the specific criminal charge. Then the question is whether Della Torre’s claim is not only unrealistic but also unreasonable given the demands of law enforcement and the efficiency pressures on the criminal justice system. Should the safeguards and components of the BARD standard taken in its broad reading count as the same for all criminal charges and situations across the board? Should we rather calibrate them by taking into consideration the desiderata of law enforcement and economic efficiency? The justice system is notoriously a costly machine. Should the same BARD safeguards apply to shoplifting and to murder, for instance? Notice also that, as a compromise, a form of asymmetrical flexibility is possible if some minimal BARD safeguards are granted to all defendants while additional ones are reserved for particularly serious charges. Della Torre himself has worked on the topics of procedural efficiency and negotiated justice 12, so it would be of great interest to understand how, in his view, the operation of BARD in its broad reading (as a method of reasoning and assessment) legitimately impacts such distinct desiderata of the justice system.
Allen, R. and Pardo, M. (2021). Inference to the Best Explanation, Relative Plausibility, and Probability. In C. Dahlman, A. Stein and G. Tuzet (eds.), Philosophical Foundations of Evidence Law. (p. 201-214). Oxford University Press.
Caprioli, F. (2009). L’accertamento della responsabilità penale “oltre ogni ragionevole dubbio”. Rivista Italiana di Diritto e Procedura Penale, 52(1), p. 51-92.
Carlizzi, G. (2018). Libero convincimento e ragionevole dubbio nel processo penale. Storia prassi teoria. Bonomo.
Conti, C. (2020). Il BARD paradigma di metodo: legalizzare il convincimento senza riduzionismi aritmetici. Diritto penale e processo, 6/2020, p. 829-842.
Della Torre, J. (2019). La giustizia penale negoziata in Europa. Miti, realtà e prospettive. Wolters Kluwer-Cedam.
Della Torre, J. (2025). Taking the Evolution of the Standard of Proof for a Criminal Conviction Seriously. Quaestio Facti. International Journal on Evidential Legal Reasoning, 8, p. 155-216. https://doi.org/10.33115/udg_bib/qf.i8.23112
Ferrer Beltrán, J. (2021). Prueba sin convicción. Estándares de prueba y debido proceso. Marcial Pons.
Gialuz, M. and Della Torre, J. (2022). Giustizia per nessuno. L’inefficienza del sistema penale italiano tra crisi cronica e riforma Cartabia. Giappichelli.
Laudan, L. (2003). Is Reasonable Doubt Reasonable? Legal Theory, 9(4), p. 295-331. https://doi.org/10.1017/S1352325203000132
Laudan, L. (2006). Truth, Error, and Criminal Law. An Essay in Legal Epistemology. Cambridge University Press.
Redmayne, M. (1999). Standards of Proof in Civil Litigation. The Modern Law Review, 62(2), p. 167-195. https://www.jstor.org/stable/pdf/1097022.pdf
Roberts, P. (2022). Roberts & Zuckerman’s Criminal Evidence (3rd ed.). Oxford University Press.
Tuzet, G. (2021). Evidence Assessment and Standards of Proof: A Messy Issue. Quaestio Facti. International Journal on Evidential Legal Reasoning, 2, p. 87-114. https://doi.org/10.33115/udg_bib/qf.i2.22480
Tuzet, G. (2023). Certainty Beyond a Reasonable Doubt. A Pragmatist Understanding of the Criminal Standard of Proof. Contemporary Pragmatism, 20(4), p. 398-423. https://brill.com/view/journals/copr/20/4/article-p398_005.pdf?srsltid=AfmBOopfjf-cMQeT3NNFdEh1dI5RaZj7ahLhDYAozaB3kOcUdw-CNgAx
Códe de procédure pénale. https://www.legifrance.gouv.fr/codes/texte_lc/LEGITEXT000006071154/
1 “The problem was that the standard of “full proof ” turned out to be so demanding that it was often impossible to meet in practice, which favoured the development of creative solutions, one of the most important of which was precisely that of extraordinary punishments.” (Della Torre, 2025, p. 170)
2 Interestingly, the hesitation of a “prudent person” appears in one of the American jury instructions on reasonable doubt; see Laudan (2003, p. 302-303) (criticizing it as subjective). Cf. Tuzet (2023) for a pragmatist reading of this.
3 See Arts. 304, 353 and 427 of the French Code of Criminal Procedure. Cf. Tuzet (2021, p. 97-98).
4 For such “revolutionary” views see Laudan (2006) and Ferrer Beltrán (2021).
5 “On the one hand, standards are essential to enable triers of fact to justify their decisions in a non-arbitrary way. On the other hand, the imposition of specific motivational burdens favours a path of clarification of the threshold to be reached.” (Della Torre, 2025, p. 201)
6 See especially Roberts (2022, p. 280ff) on “beyond reasonable doubt” as a “reasoning procedure”.
7 See Allen and Pardo (2021) claiming that the BARD standard is met when the evidence makes the guilt hypothesis plausible and there is no plausible account of it compatible with innocence.
8 Cf. among others Caprioli (2009); Carlizzi (2018); Conti (2020).
9 Cf. Tuzet (2021), where I used the phrase “assessment criteria” which is perhaps not the best formula in English, insofar as the word “criteria” does not refer to methods or guidelines but rather to clear-cut ways of discriminating things.
10 Consider this passage: ““standards of proof ” are rules, either explicit or implicit, that specify the minimum threshold that must be reached for a hypothesis to be accepted as sufficiently proven by the trier of fact.” (Della Torre, 2025, p. 158) Or this one: “standards provide the material conditions for the application of so-called “decision rules” in the strict sense, i.e. those rules which resolve procedural uncertainty—i.e. the absence of the quantum of evidence required by the standard—in a manner favourable to one party or the other.” (p. 159).
11 See Redmayne (1999) recommending the adoption of the American “clear and convincing evidence” standard in England for civil cases that can be qualified as “serious” given the stakes.
12 Cf. Della Torre (2019); Gialuz and Della Torre (2022).