2. A tangled (textual) tale: the Panel's Analysis of the Art. 2.2–5.1/5.7 SPS Relationship

Norm classification in EC–Biotech was a ‘preliminary issue’Footnote ⁵ within the analysis of EC Member State safeguard measures restricting imports of certain genetically modified products.Footnote ⁶ The need to classify norms did not arise under burden-of-proof arguments, although the parties addressed them,Footnote ⁷ and the panel noted implications in this regard.Footnote ⁸ Rather, the central question was substantive: which SPS provision(s) apply when assessing the WTO-consistency of measures?Footnote ⁹ The complainants advocated Art. 2.2/5.1 review, requiring stringent risk assessment based on sufficient scientific evidence, with Art. 5.7 playing a subsidiary role as a defence.Footnote ¹⁰ The EC, however, argued that the safeguards were provisional measures to be assessed exclusivelyFootnote ¹¹ under the more lenient terms of Art. 5.7, permitting provisional measures based on available pertinent information. Despite the considerable substantive implications, the parties' arguments soon aimed at tactical, procedural gains. For the EC, complainants had not presented an explicit Art. 5.7 violation claim, and so if the measures should be assessed only under Art. 5.7, the ‘wrong provision’ had been invoked.Footnote ¹² The complainants naturally argued that Art. 5.7 was an ‘exception’ (so the initial burden-of-proof would be the respondent's).Footnote ¹³ In approaching Arts. 2.2–5.1/5.7, procedural arguments thus juxtaposed the substantive.

Within this substantive-procedural hodgepodge the panel ‘mutated’ rule–exception–right terminology from procedural tool to substantive formula. It navigated the parties' competing references to burden-of-proof jurisprudence, while ignoring its main task: interpreting the substantive relationship between WTO rights and obligations in context. The result is a serpentine set of rationalizations that presents no unifying logic. The following observations condense and criticize the panel's analytical method as overly reliant on its idea of the rule–exception–right distinction.

First, in approaching Arts. 2.2–5.1/5.7, the panel took the role of taxonomist rather than legal interpreter. Adhering to parties' arguments, it modified the question of applicable law to read: ‘Art. 5.7 of the SPS Agreement – right or exception from the “general obligation” under Art. 5.1?’,Footnote ¹⁴ thus embracing the limited assertion of a relationship ‘of exclusion, not exception’.Footnote ¹⁵ Yet it did not clarify why these narrow terms, derived from the rule–exception–right burden-of-proof jurisprudence,Footnote ¹⁶ establish the correct framework for analyzing the substantive relationship between provisional and definitive SPS measures. Second, in its taxonomical framework the panel assumed that ‘autonomous rights’ constitute a separate class of norms, distinct from both ‘exceptions’ and ‘positive rules’.Footnote ¹⁷ Indeed, it presumed that any provision subjected to a textual litmus test (discussed below), will reveal itself as belonging to one of these categories.Footnote ¹⁸ For both procedure and substance, the panel found that Art. 5.7 ‘establishes a qualified right’. Third, in thus ‘characterizing’ Art. 5.7, the panel took a two-step approach, first analyzing the Art. 2.2–5.7 relationship, then the Art. 5.1–5.7 relationship. The necessity of this division is unclear. The AB has found that Arts. 2.2/5.1 should ‘constantly be read together’,Footnote ¹⁹ and has confirmed that Art. 5.1 may be viewed as ‘a specific application of the basic obligations contained in Art. 2.2'.Footnote ²⁰ The panel might have therefore regarded Arts. 2.2/5.1 as a single obligation, so that an analysis of their relationship to Art. 5.7 should have satisfied the panel's taxonomical effort. The bifurcated approach reflects the panel's geometrical understanding of provisions as two points between which a single line may be drawn; it raised its statutory sextant twice, to sight Arts. 2.2 and 5.1 separately in order to fix the position of Art. 5.7 as either ‘right’ or ‘exception’.Footnote ²¹ For the panel, its provision-specific readings of Arts. 2.2 and 5.1 confirmed each other,Footnote ²² each identifying Art. 5.7 as a ‘right’. Yet at no point did the panel form a coherent understanding of Arts. 2.2–5.1/5.7 from which the provisions' substantive relative applicability may emerge.

The panel thus chose a rigid construal of the rule–exception–right distinction as the definitive framework for scoping SPS provisions, and it is in this mindset that it turned to WTO jurisprudence for grounding. As its sextant, the panel relied primarily on what it called a ‘general test’,Footnote ²³ derived from the AB's EC–GSP report, whereby:Footnote ²⁴

The panel adopted this passage as a textual template for distinguishing permissive provisions that are ‘exceptions’ from those that are ‘rights’.Footnote ²⁵ It first confirmed the template by checking if it fit the Art. 3.1/3.3 relationship,Footnote ²⁶ which the AB had found establishes an ‘autonomous right and not an exception’.Footnote ²⁷ It then applied the template to Arts. 2.2–5.7, soon determining that Art. 5.7, too, is a ‘right’, and not an ‘exception’ from Art. 2.2.Footnote ²⁸ However, in this deployment of EC–GSP, the panel begged its own question. The requirement that ‘one of the provisions suggests that the obligation is not applicable’ is merely a literal transduction of the question, whether the Art. 2.2–5.7 relationship is one of exclusion, not exception, whether Art. 2.2 is applicable if the measure is reviewed under Art. 5.7. This, it will be recalled, is the very question that the panel had asked itself to begin with. Yet the panel used this constituent element of the template it identified in EC–GSP Footnote ²⁹ to proclaim that Art. 5.7 is a ‘right’, not an ‘exception’. And it did so by resorting to convoluted textualism; its only justification for deeming Art. 2.2 inapplicable in the face of Art. 5.7 is that the language of Art. 2.2 (‘except as provided for’ in Art. 5.7) ‘suggests’ that it does not apply to Art. 5.7 measures.Footnote ³⁰ Thus, the word ‘except’ was taken to imply not an ‘exception’ but an ‘exclusion’. Indeed, the panel's textual logic is such that had the phrase ‘except for …’ not been included in Art. 2.2, it could not have characterized Art. 5.7 as a ‘right’, because the text would not ‘suggest’ that Art. 2.2 is inapplicable. Yet later, the panel noted the failure of Japan–Agricultural Products II to use the word ‘exception’ to depict Art. 5.7 (using ‘exemption’ instead), as support for its characterization of Art. 5.7 as a ‘right’.Footnote ³¹ The root-word ‘except’, in presence and absence, was thus used by the panel to construct Art. 5.7 as ‘right’, not ‘exception’.

In subsequently applying the EC–GSP template to the Art. 5.1–5.7 relationship, the panel's textual entanglements were even more pronounced. In order to characterize Art. 5.7 as a right,Footnote ³² the panel relied most literally on part of the Australia–Salmon panel report depicting Art. 5.7 as an exception Footnote ³³ (establishing Art. 5.7 as a permissive provision in relation to Art. 5.1).Footnote ³⁴ But then, since there is no explicit textual cross-referencing between Art. 5.1 and Art. 5.7 that clearly ‘suggests’ the inapplicability of one in the face of the otherFootnote ³⁵ (such as the word ‘except’), the panel had no choice but to approach an analysis of the substantive distinction between Arts. 5.1 and 5.7 – namely, the existence or non-existence of scientific evidence sufficient for a risk assessment under Art. 5.1. Only here, at the end of its analysis of the Art. 2.2–5.1/5.7 relationship, did the panel briefly touch upon substance, relating, on a logical basis, to the parties' intentions in obligating themselves,Footnote ³⁶ thus breaking from its strictly literalist approach and rule–exception–right triangulation. This is more than a minor methodological inconsistency, but a failure of literalism sans paroles.

This reading of the panel's analysis of the Art. 2.2–5.1/5.7 relationship should suffice to demonstrate how its reliance on the rule–exception–right distinction drove it into dire, inconsistent, straits. A few words about the way the panel later stumbled on the basis of its characterization of Art. 5.7 as a ‘right’Footnote ³⁷ seem inevitable, though. This finding would seem to have comported with the EC's initial argument, that the Art. 2.2–5.1/5.7 relationship is one of exclusion, not exception, implying that procedurally, the burden of proving an inconsistency with the conditions attached to this right should be on the complainant; that the complainant's had invoked the ‘wrong’ provision; and that substantively, the Member State measures should be assessed only under Art. 5.7. And, indeed, the panel placed the burden-of-proof on the complainant, with respect to Art. 5.7 as well as Art. 5.1 – ostensibly a concession to the EC; but this procedural finding's effect on the report's outcome was negligible in real terms, because the panel later generally found that the EC's internal evaluation's of the Member State safeguards ‘established a presumption’ that the safeguards' were imposed ‘where relevant scientific evidence was not insufficient’,Footnote ³⁸ thus baldly shifting the burden to the EC, to prove otherwise. On the question of the proper basis of the complaint, the panel half-sidestepped the issue, apologetically referring only to a situation in which an SPS measure was adopted and ‘maintained consistently with the cumulative requirements of Art. 5.7’ – in which case a complaint under Art. 5.1 ‘could not succeed’ – and so ‘in such a case’ it would be inaccurate to say that the complainant's had invoked the ‘wrong’ provision.Footnote ³⁹ This ignored more difficult questions, e.g. whether a measure maintained inconsistently with Art. 5.7 despite its applicability, is not a violation of that provision, rather than Art. 5.1.Footnote ⁴⁰

Finally, when it came to resolving the applicability of Arts. 2.2, 5.1 and 5.7 – the fundamental ‘preliminary’ question before it – the panel had no qualms about discarding its painstaking analysis, finding that ‘in the specific circumstances of this case, the critical legal issue’ is consistency with Art. 5.1, not Art. 5.7.Footnote ⁴¹ The panel's subsequent order of analysis is to examine each safeguard's consistency with Art. 5.1, only then with Art. 5.7, distinctly like a classical rule–exception construct.Footnote ⁴² Perversely, the panel agreed with the EC that Art. 5.7 is not an ‘exception’, but then followed the complainants' order of analysis.Footnote ⁴³ The panel sought support in Japan–Agricultural Products II and Japan–Apples Footnote ⁴⁴ – a bruised reed to trust upon: in both cases, the panels (and AB) examined Art. 2.2 consistency first, then the application of Art. 5.7 as part of Art. 2.2 obligations, only then turning to Art. 5.1.

3. The rule–exception–right distinction in WTO jurisprudence: genes(is) and outcrossing

With this critical reading in mind, let us revisit the rule–exception–right distinction's prior record. It has no statutory basis, explicit or otherwise,Footnote ⁴⁵ and its sources in general international law have never been adequately traced.Footnote ⁴⁶ It was not identified or reasoned at any single point in time. Rather, its contours, as drawn by the EC–Biotech panel, evolved piecemeal through many disputes while allocating the burden-of-proof. Rereading relevant milestones of this judicial narrativeFootnote ⁴⁷ suggests that the EC–Biotech panel's framework misinterprets earlier WTO decisions. The distinction was never taxonomized so rigidly; substance and procedure were never so confused; and, ‘autonomous rights’ never constituted a separate category of norms.

First, there were only rules and exceptions. In US–Wool Shirts, faced with a burden-of-proof argument,Footnote ⁴⁸ the AB opined in a much-quoted passage based on previous GATT/WTO jurisprudence,Footnote ⁴⁹ that: ‘Arts. XX and XI:(2)(c)(i) [GATT] are limited exceptions from obligations under certain other provisions of the GATT 1994, not positive rules establishing obligations in themselves. They are in the nature of affirmative defence.’Footnote ⁵⁰ Yet the AB offered no universal guidance towards determining the defensive ‘nature’ of certain WTO provisions. Did it really mean to identify a specific category of rules and establish (or consolidate) a taxonomy of norms? In the face of prior rulings, the AB was decisive in labeling Arts. XX and XI:(2)(c)(i) GATT as ‘exceptions’, but was far less clear on the provision at issue, Art. 6 of the Agreement on Textiles and Clothing (ATC). The AB placed the burden of proving that US measures were inconsistent with Art. 6 ATC on India, the complainant, noting that GATT Reports relating to exceptions were not ‘relevant in this case’. It did not, however, classify Art. 6 ATC as a ‘positive rule’, or find that it was not an ‘exception’,Footnote ⁵¹ and, indeed, it did not establish a generally applicable relationship between Arts. 2/6 ATC, saying that these provisions reflect a ‘carefully drawn balance of rights and obligations’.Footnote ⁵² This is hardly a bright-line normative distinction.Footnote ⁵³ It is far from clear that the AB envisioned a rigid norm-classification determinative of the burden-of-proof, let alone of more substantive legal issues.Footnote ⁵⁴

In EC–Hormones, the panel considered Art. 3.3 SPS to be an exception to the Art. 3.1 obligation to base SPS measures on international standards, and that the burden of proving Art. 3.3 compliance (and hence with Art. 5 risk assessment) was on the party defending its SPS measure. The AB rejected this, finding instead that Arts. 3.1/3.3 don't hold a rule–exception relationship.Footnote ⁵⁵ In explaining why Art. 3.3 was not an ‘exception’, the AB offered the following, oft-repeated language:

The probative burden regarding Art. 3.3 was duly allocated to the complainants, but more importantly for present purposes, the norm-category of ‘autonomous rights’ was textually born, establishing the impression that there are three categories of norms among WTO rules, along with what seems like a legal definition distinguishing ‘autonomous rights’ from ‘exceptions’ – provisions that another provision ‘simply excludes’ from its scope of application.

Despite the eminently quotable passage, did the AB really intend to establish a new category of norms – ‘autonomous rights’ – to which special rules might apply in dispute settlement? If one observes how the AB actually applied the law in EC–Hormones, it clearly did not, for in practice it regarded Art. 3.3 as a provision that in the proper context of Art. 5 SPS establishes obligations of a standing at least equivalent to those of Art. 3.1. Arts. 3.1 and 3.3 act as separate, alternative rules governing different situations: the former applies to SPS measures ‘based on’ international standards, the latter applies to measures resulting ‘in a higher level’ of protection.Footnote ⁵⁷ The burden of proving an inconsistency with either provision rests in any case on the complainant who claims that one of them has not been complied with by the respondent. In terms of norm classification, both provisions are obligations,Footnote ⁵⁸ the scope of whose application is mutually exclusive. In another, virtually uncited segment of EC–Hormones the AB states outright that the panel should have:

The AB thus treated Art. 3.3 SPS as an ‘obligation’ and even referred to it as such. Supporting this understanding is the fact that the AB, having found that the EC measures were inconsistent with Art. 5.1, and, hence, with Art. 3.3, did not derive from this a violation of Art. 3.1, indeed reversing the panel's finding of violation.Footnote ⁶⁰ In substance, the AB found that Arts. 3.1 and 3.3 are independent ‘positive rules’, each of which may be violated under its own terms (and cannot be violated contemporaneously by the same measure). Which of them applies to a given situation depends on the watershed question, whether the challenged measure is ‘based on’ an international standard or not, a question of substantive interpretation, not procedure. The AB did not intend, and did not need, to create a new set of norms; but the powerful textual legacy of EC–Hormones is the phrase ‘autonomous rights’, suggesting a distinct normative class, setting the stage for future complication.Footnote ⁶¹

What did the AB actually mean by the term ‘autonomous right’? In general legal parlance, ‘autonomous’ right usually refers to the autonomy of the right-holder – that the right is an expression or an extension of the autonomy of the bearer of the right.Footnote ⁶² This is likely what the AB intended by using the phrases ‘autonomous right of a Member …’ and ‘projected for itself’ (emphases added)– referring to the element of choice that exists in Art. 3.3 (and in ‘the kinds of situations’ it covers), between international standards and higher ones. In EC–Biotech, Canada correctly pointed out that this aspect of ‘free choice’ is absent in Art. 5.7, whose application depends on an objective or exogenous condition – the absence of sufficient scientific evidence.Footnote ⁶³ But Canada overshot in arguing on this basis that Art. 5.7 was an exception and that any recourse to the Art. 3.1–3.3 relationship as discussed in EC–Hormones would be ‘inappropriate’.Footnote ⁶⁴ At the same time the EC was insisting that Art. 5.7 was an ‘autonomous right’, not an exception.Footnote ⁶⁵ Thus, both complainant and respondent integrated the ‘autonomy’ of the right into the provision's classification.

Moreover, a literalist approach bent on finding a neat taxonomy of norms, distinguishing between ‘exceptions’ and ‘rights’, would read into the phrase ‘autonomous rights’ another meaning, that would refer to the autonomy of the right, not the right-holder – i.e., that the provision, because its scope of application has been excluded, not because of the right-holder's choice, is an ‘autonomous’, self-standing permissive norm, which unlike an ‘exception’, is not dependent upon the existence of an opposing obligation or prohibition.Footnote ⁶⁶ It seems that at least the EC–Biotech panel adopted this understanding, allowing it to discard Canada's arguments;Footnote ⁶⁷ and yet, the panel was likely aware of the ambiguity of the term ‘autonomous’, because in its own conclusion it eschewed it, using instead the term ‘qualified right’ for Art. 5.7.Footnote ⁶⁸

Another relevant case, Japan–Agricultural Products II was referred to by the EC–Biotech Panel because it specifically addressed Arts. 2.2, 5.1, and 5.7 SPS.Footnote ⁶⁹ In this dispute, to simplify, the US cited violations of Arts. 2.2 and 5.1, and Japan, as respondent, invoked the ‘provisional measures’ clause, Art. 5.7. Arguments were made on the burden-of-proof, but neither Panel nor AB were concerned with classifying Art. 5.7 as either ‘exception’ or ‘right’. The ‘qualified exemption’ language was used by the AB not taxonomically but rather in substantively interpreting Art. 2.2 – the logic being, that because Art. 5.7 applies where ‘sufficient scientific evidence’ is absent, an ‘overly broad and flexible interpretation’ of the sufficiency requirement ‘would render Art. 5.7 meaningless’.Footnote ⁷⁰ Examining the way the panel/AB applied Art. 5.7 itself as well as the burden-of-proof, it seems clear that both treated the provision as an obligation. To the panel, Art. 5.7 is a specification of obligations under Art. 2.2, complementary to Art. 5.1.Footnote ⁷¹ Even though Japan, the respondent, had invoked Art. 5.7, the initial burden of proving its violation was placed on the complainant, the US.Footnote ⁷² Thus, the panel and AB discarded the textual baggage of EC–Hormones: obligations are obligations, not ‘autonomous rights’.

In Brazil–Aircraft,Footnote ⁷³ the panel faced burden-of-proof arguments relating to the export subsidy prohibition under Art. 3.1(a) SCM, on the one hand, and the special and differential treatment provisions under Art. 27 SCM, on the other. Canada, the complainant, argued that the latter were an ‘exception’ and that the burden was on Brazil. Brazil disputed this, arguing instead that Art. 3.1(a) SCM did not apply to it, as a developing country complying with the conditions of Art. 27 SCM. The panel (as upheld by the AB)Footnote ⁷⁴ engaged in a relatively contextual examination of the respective provisions, methodologically diametrically opposed to the taxonomical approach in EC–Biotech Footnote ⁷⁵ The panel did not ask whether Art. 27 is a ‘right’ or ‘exception’, but rather more specifically asked itself whether Art. 27.4 SCM is ‘an element of the claim of inconsistency with Art. 3.1(a)’, in the light of ‘the relevant text of the Agreement in its context, and in light of the object and purpose of the SCM Agreement’.Footnote ⁷⁶ It considered that Canada, to succeed in its claims under Art. 3.1(a) SCM, would need to demonstrate that the provision applies to the respondent in the particular situation,Footnote ⁷⁷ and so would need to show that Art. 27 does not apply to Brazil because its conditions had not been met. Although it noted that Art. 27.2(b) ‘recognizes the autonomous right’ of certain developing countries to maintain export subsidiesFootnote ⁷⁸ – and certainly the use of the word ‘autonomous’ here is appropriate, because it refers to the right-holder's freedom to exert its right at will – the panel characterized Art. 27.4 as establishing obligations to ‘be considered in conjunction with Art. 3.1(a) of the SCM Agreement in order to establish a claim of violation of that provision’.Footnote ⁷⁹ In short, the panel did not resort to ‘autonomous rights’ as a third norm-category, but regarded the provision as an obligation that is both alternative in its application and contextually related to the obligations of Art. 3.1(a) SCM.

Finally, while the impact of EC–GSP on the EC–Biotech panel's inconsistencies is clear,Footnote ⁸⁰ what remains is to review EC–GSP to see whether it supports the EC–Biotech panel's understanding, establishing a textual template for ‘rights’ and ‘exceptions’. In EC–GSP, the characterization of the Art. I:1 GATT/Enabling Clause relationship arose as a burden-of-proof question and was accompanied by the tactical argument that the complainant had not made a proper claim, having not claimed a violation of the Enabling Clause.Footnote ⁸¹ However, even more than in EC–Biotech, the procedural question was tied to a far-reaching substantive one: do the MFN obligations of Art. I:1 GATT continue to apply to measures permitted under the Enabling Clause (hence preventing discrimination among GSP beneficiaries), or do they dissolve as a result of the Enabling Clause?Footnote ⁸² Notably, the EC–GSP panel understood the underlying question as a binary one (the Enabling Clause is either ‘rule’, or ‘exception’), to be answered in a contextual and purposive manner.Footnote ⁸³ On this basis, it deemed the Enabling Clause an ‘exception’, and found that the burden-of-proof rests on the respondent, yet MFN obligations continue to apply where there is no inconsistency between the provisions.Footnote ⁸⁴ The AB, too, examined the contextual relationship between Art. I:1 and the Enabling Clause,Footnote ⁸⁵ and then looked ‘to the object and purpose of the WTO Agreement and the Enabling Clause to clarify whether the Enabling Clause was intended to operate as an exception to Art. I:1’.Footnote ⁸⁶ The AB thus disposed of an argument whereby ‘exceptions are provisions that allow measures that pursue objectives’ other than the WTO's.Footnote ⁸⁷ In so doing, it employed language that clearly distinguishes the substantive analysis of rules and their relationships from the procedural classification for burden-of-proof purposes:

This stands in stark contrast to the textual template pursued in EC–Biotech Footnote ⁸⁹ and it is on this basis, not a techno-textual one, that the AB confirmed the EC–GSP characterization of the Enabling Clause as an ‘exception’ to Art. I:1 GATT, placing the burden-of-proof on the EC.Footnote ⁹⁰ The formulaic passage in EC–GSP – EC–Biotech's decisive reference – was but an attempt to generalize previous findings on the burden-of-proof, obiter dicta mistaken for ratio decidendi.Footnote ⁹¹ This analysis reveals several gaps between the panel's use of the rule–exception–distinction and preexisting jurisprudence: the term ‘autonomous rights’ used in certain WTO cases was not designated as a separate norm-category; the distinction was never construed as inflexibly, taxonomically, as it was in EC–Biotech; and it was previously employed for procedure, not for substantive interpretation. In these senses, the panel's method of analyzing the Art. 2.2–5.1/5.7 relationship is simply wrong, as a matter of WTO law.

4. Restoring (genetic) order: logic and the Art. 2.2–5.1/5.7 relationship

The EC–Biotech panel report is not a freak ruling; it is an exaggeration of familiar currents in WTO judicial practice. Its incongruities are confounding but ingenuous, based on fragmentary literal readings of prior jurisprudence. They are the WTO's folly, taking literalism to new extremes. Where substantive, contextual interpretation of rules should determine their treatment, obscure textualism has distorted the way their interactions are interpreted. In international law, substantive questions on obligations and permissions are not unique to the WTO, but it is unlikely that a tribunal would apply a procedural-textual template to divine emergency derogations from international human rights,Footnote ⁹² or the applicability of humanitarian law in non-international armed conflicts.Footnote ⁹³ Yet, the EC–Biotech panel resorted to a rigid, literal path in addressing a substantive legal relationship that is arguably no less important for human welfare. This should trouble anyone concerned with the efficacy, reflexivity, and legitimacy of the WTO and its dispute settlement system.

In closing, I will suggest how order may be restored in this area, turning to a logic that validates the analysis above, then to SPS substance. Deontic logic – the logic of obligation and permission – supports the intuition that distinguishing between ‘rules’, ‘exceptions’ and ‘rights’, ‘exemptions’, ‘autonomous rights’, and ‘conditional rights’, on a purely textual basis, is both futile and false. Permissive norms, however phrased, can all be understood as equivalent to ‘exceptions’:

Thus, the very idea of ‘autonomous rights’ as self-standing permissive rules distinguishable from ‘exceptions’ has no logical or normative utility. Furthermore, permissive norms are intimately connected to contingent obligations, making even the distinction between ‘rights’ or ‘exceptions’ and ‘obligations’ a difficult one. Von Wright, the deontic school's founder,Footnote ⁹⁵ distinguished between ‘weak permissions’ and ‘strong permissions’:Footnote ⁹⁶ ‘An act will be said to be permitted in the weak sense if it is not forbidden; and it will be said to be permitted in the strong sense if it is not forbidden but subject to norm.’ Many, if not most, WTO rules may thus be understood as ‘strong permissions’ rather than ‘pure obligations’. The entire SPS construct permits states to adopt SPS measures, ‘subject to norm’, and as such its provisions are ‘strong permissions’, rather than obligations – after all, the adoption of SPS measures is not mandatory. Indeed, the SPS agreement is a detailed ‘exception’ to the GATT, a super-strong permissive complex of rules – recall that the SPS Preamble states that it is intended ‘to elaborate rules for the application of the provisions of GATT 1994 which relate to the use of sanitary or phytosanitary measures, in particular the provisions of Art. XX(b)’ (emphasis added), and under Art. 2.4 SPS, a measure to which the SPS agreement applies (in that it satisfies the definition of an SPS measure) and which complies with the provisions of the SPS agreement is presumed to comply with the GATT in the same sense.Footnote ⁹⁷

So, most WTO ‘permissions’ may be seen as rights or exceptions to obligations, themselves including a contingent obligation. Furthermore, deontic logic shows us that, textually, the same prescriptive results are attainable exclusively through obligatory semantics, without the use of any distinctively permissive language:

The textual structuring of provisions tells us little about either their substantive relationship or their burden-of-proof classification; the same normative results can be achieved through different textual formulations. This confirms that the substantive relationship between provisions cannot be left to text, or as the EC–GSP panel put it, ‘Whatever its characterization, a provision of the covered agreements must be interpreted in accordance with the ‘customary rules of interpretation of public international law’.Footnote ⁹⁹ Most WTO norms are obligations, contingent or otherwise, and so by default, any claim of violation should rest on evidence produced by the complainant.Footnote ¹⁰⁰ Exceptions are, put simply, exceptional. A provision is an ‘exception’ for burden-of-proof purposes on the basis of an interpretation that characterizes that provision, in its context and circumstances, as ‘intended to operate’Footnote ¹⁰¹ as an ‘exception’ or ‘affirmative defence’, to be proven by the respondent, not the complainant. This does not mean that the same provision cannot function, in other circumstances, and substantively, as a ‘positive rule establishing an obligation’.Footnote ¹⁰²

What does this imply for Arts. 2.2–5.1/5.7? I have focused on interpretative method, with dispute settlement in mind, and so I may be as culpable for decontextualization as the panel, but I will now briefly trace how the panel should have applied Arts. 2.2–5.1/5.7, in my view. The panel should have considered Art. 5.7 in the context of both Arts. 2.2 and 5.1 and not bifurcated the analysis. It should have noted that the provisions establish a ‘carefully drawn balance of rights and obligations’ recognizing two distinct situations: one, where there exists scientific evidence sufficient to base an SPS measure on risk assessment (whether risk assessment has been conducted, and whether the measure has been ‘based on’ it); the second, where scientific evidence is insufficient for this purpose.Footnote ¹⁰³ It ought then have conceived Arts. 5.1 and 5.7 as complementary and alternative obligations, each a separate ‘specific application’Footnote ¹⁰⁴ of Art. 2.2, ‘an element of a claim of inconsistency’.Footnote ¹⁰⁵ The panel should have found that the complainants had not invoked the ‘wrong’ provision, having claimed an Art. 2.2 violation implicating either Art. 5.1 or Art. 5.7. It should then have asked whether Art. 5.7 was ‘intended to operate’Footnote ¹⁰⁶ as an ‘exception’ for procedural burden-of-proof purposes. It would then order the analysis as in previous SPS disputes:Footnote ¹⁰⁷ Art. 2.2, then Art. 5.7, only then Art. 5.1.

Would this have altered EC–Biotech's result? On the case's facts – probably not; but the order of analysis would change, making the sufficiency of evidence pivotal, with significant prescriptive implications for state conduct and SPS disputes. The panel's order of analysis – Art. 5.1, then Art. 5.7 – allows it to derive the sufficiency of evidence that precludes Art. 5.7 measures, from risk assessments upon which the measures were not ‘based’, in a highly problematic manner. The objective question of sufficiency of evidence, a crucial substantive element of Art. 2.2, is never properly examined. Indeed, Art. 5.7 becomes redundant, because the panel examines whether any Art. 5.1 risk assessment was conducted (or not), simply assuming that scientific evidence was sufficient for this purpose. In such circumstances, the same measures could not possibly be subsequently justified by Art. 5.7, whose relevance depends on the insufficiency of evidence. This is judicial tenterhooking, reversing the rational connection between scientific evidence and risk assessment. By bypassing the sufficiency of evidence question, this mistaken approach neutralizes the Art. 5.7 role in stimulating research for new information. As per the panel, it seems that a pre-existing risk assessment (whether based on ‘sufficient evidence’ or not) cannot be trumped by new information suggesting risks but requiring additional research that might justify a provisional Art. 5.7 measure. Undoubtedly, the panel's legacy is therefore difficult in not only procedural but also substantive terms. Much of the EC–Biotech outcome will be intelligible only ‘to anyone that understands such things’.Footnote ¹⁰⁸