The Place of Constitutional Conventions in the Constitutional Architecture, and in the Courts

Abstract The Supreme Court's recent invocation of the “constitutional architecture” in the Senate Reform Reference has led a number of scholars to question the status of constitutional conventions in the legal, as opposed to political, constitution. Has the Court, without expressly saying so, transformed at least some conventions into constitutional law? This would be a serious rupture, not only from existing precedent on the justiciability of conventions but also from the traditional understanding of conventions as binding political rules. In light of this recent scholarly debate, an exploration of the profound consequences of entrenching conventions in the legal constitution is warranted, as it implicates the meaning of constitutional conventions, their creation, their relation to law, and their enforcement. Judicial entrenchment of conventions would be a dangerous violation of the separation of powers and would have negative consequences for the functioning of Canada's system of government and for the future of constitutional change.

In its landmark opinion in Reference re Senate Reform (2014), the Supreme Court of Canada elaborated on the scope of the constitutional amending formula's various procedures to determine whether Parliament alone could validly enact certain changes to the Senate. In doing so, the Supreme Court relied on the amorphous concept of the "constitutional architecture" to note that "the Constitution must be interpreted with a view to discerning the structure of government that it seeks to implement" and which includes the "assumptions that underlie the text" (2014: para. 26). The Court explained that "amendments to the Constitution are not confined to textual changes. They include changes to the Constitution's architecture" (para. 27). The Court's conclusion that a proposed consultative elections scheme required provincial consent was in part premised on the notion that it was an attempt to bind future prime ministers to appoint the winners of senatorial elections. Without saying so explicitly, the Court found that any legislation to implement consultative elections amounted to an attempt to create a binding constitutional convention via a purportedly non-binding legal process.
The Court's decision and its invocation of the ill-defined constitutional architecture has led a number of scholars to question the status of constitutional conventions in the legal, as opposed to political, constitution (Glover, 2014;Pal, 2016;Hamill, 2016;Scholtz, 2018;Sirota, 2020;Macfarlane, 2021a). If constitutional conventions are part of the architecture of the Constitution such that any changes to them implicate the formal amending formula, then the Court has, without expressly saying so, transformed at least some conventions into constitutional law. This would be a serious rupture, not only from existing precedent on the justiciability of conventions but also from the traditional understanding of conventions as binding political rules of behaviour that are neither law nor enforced by courts.
In light of recent scholarship raising this question-including a number of scholars endorsing the justiciability of conventions-an exploration of the profound consequences of entrenching conventions in the legal constitution is warranted. This analysis requires not only revisiting debates about the meaning of constitutional conventions, their creation, and their relation to law, but also the crucial role of conventions for political accountability. I argue that judicial entrenchment of conventions as legal rules would be a dangerous violation of the separation of powers and would have negative consequences for the functioning of Canada's system of government and for the future of constitutional change. It would also contribute to the further judicialization of politics-understood as the willingness of, and reliance on, courts to exercise judicial review to settle social and policy disputes (Hirschl, 2011;Manfredi, 1997;Tate and Vallinder, 1995)-and would undermine the constitutional role of the executive and legislative branches in ensuring democratic accountability and the functioning of the overall system of government. Those who advocate judicial declaration or enforcement of conventions dramatically understate the ability of politics to address perceived breaches of convention. A review of recent cases and controversies also illustrates that some scholars overstate the extent and circumstances under which conventions can be established. The article concludes that, fortunately, there is an interpretation of the Court's holding in the Senate Reform Reference that allows it to back away from the dangerous jurisprudential precipice it clumsily approached in 2014.

The Status and Role of Constitutional Conventions
A. V. Dicey famously identifies a set of rules that he terms "conventions of the constitution," ([1885] 2013: 20) and notes that "though they may regulate the conduct of the several members of the sovereign power, of the Ministry, or of other officials, [they] are not in reality laws at all since they are not enforced by the courts." K. C. Wheare defines a convention as "a binding rule, a rule of behaviour accepted as obligatory by those concerned in the working of the Constitution" (1966: 15), and Geoffrey Marshall as "rules of non-legal accountability" (1984: 1).
Dicey's core law and politics distinction, with conventions as binding rules of political behaviour not enforced by courts, remains the dominant understanding among leading commentators (Jennings, 1959;Wheare, 1966;Marshall, 1984), although these authors also recognize that the relationship between conventions and law is complex and interwoven. Sir Ivor Jennings describes conventions as providing "the flesh which clothes the dry bones of the law; they make the legal constitution work; they keep it in touch with the growth of ideas" (1959: 81-82). Jennings saw a reciprocal and reinforcing relationship between law and convention, noting that "conventions presuppose the law. The conventions of Cabinet government, for instance, assume the legal relations between Queen and Parliament. . . . Nevertheless, though conventions are built in the first instance on the foundation of law, when once they have been established they tend to form the basis for the law" (83-84). In his view, "the two are inextricably mixed, and many conventions are as important as any rules of law" (84).
Conventions can modify or even contradict law. A decontextualized reading of the Constitution of Canada would lead those with no knowledge of practice to assume the Queen or the governor general were supremely powerful actors. One of the lynchpins of Westminster-style parliamentary systems, responsible government, holds that the prime minister and cabinet (the government) must maintain the confidence of a majority of members of the House of Commons. A set of related conventional rules, that the sovereign (formal) power act on the advice of the government, ensures the democratic functioning of the system. Conventions can even render operative law dormant or politically illegitimate. For example, the federal powers of reservation and disallowance are widely regarded as falling into a convention of disuse. At the same time, conventions can be incorporated into law, thus being adapted into judicially enforceable legal rules (Marshall, 1984: 211). In a country like the United Kingdom, this is a straightforward proposition of legislating; in Canada, it is complicated by a partially codified constitution with an amending formula. Codifying any convention with implications for the constitutional text would necessitate the use of the relevant amending procedure.
How conventions arise remains subject to an intense scholarly debate. Conventions are often characterized as emerging from long-standing practice. At some point, the relevant actors consider themselves bound to a rule. The Jennings test, famously applied by the Court in the Patriation Reference (1981), is a three-prong assessment that asks: "first, what are the precedents; secondly, did the actors in the precedents believe that they were bound by a rule; and thirdly, is there a reason for the rule?" (Jennings, 1959: 136). Jennings notes that while many conventions emerge from long-standing practice, mere practice or the existence of precedents are not enough (136). The normative reasons for a rule distinguish conventions from mere customs or tradition (Lagassé, 2019).
Where Jennings saw precedent as a necessary but not necessarily sufficient condition for the emergence of new conventions, some scholars note that precedents are not necessary (Heard, 2014). Indeed, conventions can also emerge as result of explicit agreement of, or declaration by, the relevant actors (Wheare, 1966: 122;Marshall, 1984: 8-9). For example, the Imperial Conference of 1926 and resulting Balfour Declaration established the "principle of equality of status" of the UK and the Dominions (Russell, 2004: 53).
Nonetheless, in many instances it will not be clear that a convention is established until there is a precedent of it being followed or that, over a period of time, successive actors adhere to the rule. Domestic examples of conventions created by agreement or mere declaration are few, and some of those offered by commentators are questionable. Andrew Heard argues that a "prime minister could, for example, create a new convention by unilaterally declaring that future lieutenant governors would be appointed from a list of nominees presented by the relevant provincial premier" (2014: 10). This is highly questionable, and subsequent events at the federal level, specifically the 2017 decision of Prime Minister Justin Trudeau to abandon the Advisory Committee on Vice-Regal Appointments established by his predecessor in 2012, reveal how successor actors may not feel inclined to follow newly established practice, even if there was a well-articulated principle behind the reform. Absent clear indication of bipartisan support (and public articulation of such support around conventions is rare, as Heard notes) or absent a record of successive governments committing to a rule, it is questionable whether a binding convention can be created with immediacy via the unilateral pronouncement of a single constitutional actor. Instead, a new practice-a rule that is supported by reasons but for which there is no evidence of strong consensus-may need time to mature and solidify into convention, and hence come to enjoy its status as a binding political rule (Lagassé, 2019).
A third process for the creation of conventions, advanced by Heard, suggests they may "arise from accepted constitutional principles, even when precedent, agreement, or declarations are absent" (2014: 6). Heard notes that on occasion novel situations can arise, with no precedent or prior agreement or declaration, yet there might still be a substantive obligation that can require a particular outcome rooted in identifiable principle. One example he points to is former Governor General Michaëlle Jean's decision to renounce her French citizenship to alleviate perceptions of divided loyalty (12). Heard's approach may address deficiencies in the Jennings test and the explicit agreement approach, but key challenges remain, including: whether the broad unwritten principles identified by the courts are precise enough to ably support specific conventions in the absence of clear agreement or precedent; how conflicting principles ought to be addressed; and perhaps most importantly for the present analysis, whether certain conventions are clear enough to warrant judicial enforcement.

Constitutional Conventions and the Constitutional Architecture
Courts have long dealt with conventions in the course of issuing decisions. Marshall notes that this occurs in various ways, including when conventions are part of the material incorporated into law, by recognizing practices that "elucidate the background against which legislation took place" or by "constituting a practice or set of facts that fell under an existing legal doctrine" (1984: 15). The Canadian Court's most significant engagement with convention came in the Patriation Reference (1981), where in a complex set of four decisions and a dual set of majorities, it concluded that the federal government, by way of resolutions of the two houses of Parliament, faced no legal restrictions in making a unilateral request to the UK to pass the constitutional amendment package but by convention required a substantial degree of provincial consent before proceeding.
The Patriation Court endorsed a core distinction between recognizing a convention and enforcing one, underscoring that conventions are "not enforced by the courts" (Patriation Reference, 1981: 880), a point on which both the majority and dissenting justices agreed. Many commentators were deeply critical of the Court's willingness to formally recognize a convention, referring to the "questionable jurisprudence" of the reference (Russell, 1982). Peter Hogg argues the Court should have confined its reasoning to the question of law, in large part because conventions pertain to political accountability, something the Court itself lacks (2003: 22). Adam Dodek considers the reference "more political than jurisprudential" and argues that the "distinction between 'recognizing' and 'enforcing' conventions is artificial and untenable" (2011: 127). He excoriates the Court for "creating constitutional danger" with a "very broad, sweeping and unnecessary approach" to conventions (123). Even scholars more sanguine about judicial recognition, and even enforcement, of conventions take issue with the way the Court interpreted and applied precedents to arrive at its conclusion (Heard, 2014: 174-75). Some interpret the decision not as identifying an existing convention but as having changed convention (Barry et al., 2019).
In the years following the Patriation Reference, courts affirmed the "traditional supremacy of law over convention" (Heard, 2014: 112). The 1991 case Osborne v. Canada asserted the law and politics distinction and addressed in a more specific way the relationship between conventions, statutory law and constitutional amendment. The case involved a provision of the Public Service Employment Act prohibiting public servants from engaging in partisan activities. Since the legislation was premised on the convention of political neutrality in the public service, the government argued the law was "a part" of the Constitution and could not therefore be inconsistent with the Constitution (in this case, the Charter of Rights and Freedoms). In response to this assertion, the Court noted that "while conventions form part of the Constitution of this country in the broader political sense, i.e., the democratic principles underlying our political system and the elements which constitute the relationships between the various levels and organs of government, they are not enforceable in a court of law unless they are incorporated into legislation" (Osborne, 1991: 87). The Court also articulated an important understanding of the relationship between conventions and constitutional amendment, noting that "statutes embodying constitutional conventions do not automatically become entrenched to become part of the constitutional law, but retain their ordinary status. If that were not the case, any legislation which may be said to embrace a constitutional convention would have the effect of an amendment to the Constitution which would have escaped the rigorous requirements of the constitutional amending process" (87).
This distinction between constitutional law and conventions is one that the Court in the Senate Reform Reference appears, to some observers, to abandon via its use of the architecture concept. A central question in the reference was whether Parliament could unilaterally implement consultative elections for Senate nominees. The formal appointment of senators, as set out in the Constitution Act, 1867, is made by the governor general. As the Court noted, by convention, the governor general follows the recommendation of the prime minister when making appointments. The question the Court faced was whether consultative elections would amount to a change in the "method of selecting Senators" under section 42(1)(b) of the amending formula in Part V of the Constitution Act, 1982, which specifies that such changes require recourse to the general procedure, necessitating changes approved by the House of Commons and the Senate and at least seven provinces representing at least 50 per cent of the population.
The federal government argued that because the reform involved no changes to the constitutional text and because consultative elections preserved the discretion of the prime minister in deciding whether to appoint the winners of Senate elections, it was not an amendment to the Constitution. The Court dismissed this argument as "privilege[ing] form over substance. It reduces the notion of constitutional amendment to a matter of whether or not the letter of the constitutional text is modified" (Reference re Senate Reform, 2014: para. 52).
The Court further determined that the "words 'the method of selecting Senators' include more than the formal appointment of Senators by the Governor General. . . . By employing this language, the framers of the Constitution Act, 1982 extended the constitutional protection provided by the general amending procedure to the entire process by which Senators are 'selected'" (Reference re Senate Reform, 2014: para. 65). The justices might have relied on this for their conclusion, but a key factor, in their view, was that consultation elections for the Senate would amount to a fundamental alteration of the "architecture" of the Constitution (para. 54). The framers, the Court asserted, had the intention of making "the Senate a thoroughly independent body" and "to remove Senators from a partisan political arena that required remitting consideration of short-term political objectives" (para. 57). It also ensured that the Senate would be a "complementary" rather than a competitive body to the lower house, as senators would not enjoy the expectations and legitimacy that come from popular elections (para. 58).
Moreover, the fact that the prime minister would technically retain discretion to refuse to appoint the winners of elections did not satisfy the Court: It is true that, in theory, prime ministers could ignore the election results and rarely, or indeed never, recommend to the Governor General the winners of consultative elections. However, the purpose of the bills is clear: to bring about a Senate with a popular mandate. We cannot assume that future prime ministers will defeat this purpose by ignoring the results of costly and hardfought consultative elections. (Reference re Senate Reform, 2014: para. 62) As several scholars have recognized, in effect, the Court determined that legislation to enact consultative elections had the intention of creating a binding practice of appointing elected senators. Richard Albert describes this as "constitutional amendment by stealth," an effort by political actors to "consciously establish a new political practice whose repetition is intended to compel successors to conform their conduct to that practice. Over time, this practice matures into an unwritten constitutional convention, though without the popular legitimacy we commonly associate with a constitutional amendment" (2015: 678). In his view, the Court recognized this attempt. The fact that it was an attempt to circumvent the constitutional text by creating a convention and the fact that it was done so through legislation are what amount to a change to the constitutional architecture.
Robert Hawkins (2010) refers to the same process as a "constitutional workaround." Hawkins does not view such workarounds as automatically establishing conventions, and he writes approvingly that they "help prevent the constitution from becoming ossified" and can be a legitimate path to amendment in the face of an onerous written formula (2010: 517). Hawkins is also skeptical that successor prime ministers would feel compelled to appoint senatorial election winners, noting that a new government that "found itself in opposition in the Senate, might well have passed over the winner of a Senate election who was from an opposition party. The democratic claim of the Senate election winner would be offset by the democratic mandate of the prime minister obtained in the recent general election" (525). Hawkins is also skeptical of claims that legislating such options is any less constitutionally legitimate, noting that "surely what is constitutionally permissible as an exercise of prime ministerial discretion does not become impermissible when authorized by ordinary legislation" (525). At any rate, even if such appointments became bound by convention over time, Hawkins notes, such a process "is a legitimate path to amendment. The fact that a non-binding change might one day evolve into a convention is no reason to say that the non-binding change is void at the outset because of its potential to amend the constitution" (525-26).
Despite this disagreement, the Court's ruling stands, and because the Court did not explicitly invoke conventions as part of the constitutional architecture, it is not clear what to make of it. Indeed, the architecture concept itself has been subject to critical scrutiny as amorphous and heavily dependent on the justices' ability to accurately define the animating features of the institutions and processes that compose the broader constitution (Macfarlane, 2015). Kate Glover Berger notes that the decision suggests "that constitutional conventions are generally subject to Part V" and that Canada's constitutional rigidity might be even worse than thought as a result (Glover, 2014: 250-51). Michael Pal notes that the way the Court employs the architecture concept creates uncertainty with potential implications for changes for a host of issues, from conventions to electoral reform (2016: 383-85). Elsewhere I note the lack of clarity over the issue but add that the "Court's constitutional architecture logic, if pushed to its natural end, might also mean that abandoning certain conventions may be interpreted as an unconstitutional attempt at constitutional change" (Macfarlane, 2021a: 155). It is not clear, for example, whether a prime minister's decision to depart from the convention on regional representation for Supreme Court appointments would now be regarded as a breach of the architecture.
Two scholars have engaged in depth with this interpretation of the architecture concept. Christa Scholtz writes that the Patriation majority recognized the Constitution as "a complicated apparatus with various components accreted through time" and as something that "defies any pretense of visionary industrial design. In this apparatus, political conventions relieve the undeniable frictions and inconsistencies between its component parts" (2018: 662). She argues that the Court has now abandoned this vision and that the metaphor of a constitutional architecture, first invoked in the Secession Reference (1998) and then elaborated on in the Senate Reform Reference, "presumes incoherence away, and with it, a clear role for conventions to achieve constitutional durability. Conventions then become suspect means that allow political actors to fracture and sabotage, rather than renovate, the Canadian constitutional structure" (662-63).
Like Hawkins, Scholtz would justify the potential alteration of conventions by permitting consultative Senate elections to evolve into practice as something that "is hardly unexpected, as informal and consequential change is precisely what our core conventions allow us to do" (2018: 665). She points out that the Court's concern about legitimation of the Senate, transforming it from a complementary to potentially competitive body, raises a question about "how the prospect of the Senate's increased use of its legal powers becomes contrary to the rule of law rather than its exercise" (665-66). In Scholtz's view, the Court has relegated conventions to the sidelines, and its caution against privileging form over substance "marks a remarkable shift in the account of what conventions can be allowed to do in Canadian constitutionalism" (666).
Where Scholtz depicts the Court as marginalizing conventions, Léonid Sirota argues "the Court's opinion has the effect-perhaps the deliberate effect-of entrenching at least some, although possibly not all, of the conventions of the Canadian Constitution, erasing the bright line that Dicey sought to draw between convention and law" (2020: 318). Sirota suggests the Court was not willing to explicitly depart from the Patriation Reference and subsequent cases upholding the law/convention distinction, but that "conventions are strategically located between the lines of the Court's opinion." In fact, "conventions are well-nigh all there is to 'constitutional architecture' understood as the constitutionally protected set of relationships between institutions" (334). The result of potential enforcement of conventions, Sirota notes, is "tantamount to their erasure," but entrenchment may be partial or finite, and we do not know which conventions are entrenched (334-35).
Unlike Scholtz, Sirota finds this prospect encouraging, even suggesting that the incorporation of conventions into the architecture concept makes it "less amorphous and threatening than Professors Macfarlane and Pal, among others, have suggested" (2020: 327). For reasons described in the next section, I argue the opposite: incorporation of conventions into the architecture is a grave threat to the coherence of the Constitution and for the prospects of future constitutional evolution. In spite of framing the place of conventions within the architecture concept quite differently, Scholtz and Sirota ultimately come to the same conclusion about the effect of the Court's decision: the legalization of conventions and new limitations on how constitutional change might legitimately come about.

Unworkable Architecture
If the Court has indeed read at least some conventions into the Constitution with the full legal status that implies, it creates an immediate and arguably intractable problem. Conventions are by definition political rules that are considered binding obligations on relevant political actors, enforced by politics. But it is not enough to point to this conceptual definition, lest one risks invoking simple tautology. Instead, we need to look to the underlying reasons for the distinction between conventions and the law. This goes to the role of conventions, which historically emerged to evolve a monarchical system of government into a functioning representative democracy and constitutional monarchy. Conventions developed as a non-legal part of modern Westminster constitutions. In a system like Canada's, which adapted a partially codified constitution that was nonetheless premised on its similarity in principle to that of the United Kingdom, conventions can operate to alter or even contradict the constitutional text in order to ensure the system's democratic functioning. Their sudden entrenchment as legally enforceable aspects of the constitutional architecture would result not just in confusion or incoherence but outright paradox.
Sirota (2020) acknowledges this issue, but he argues it would be "too quick" to suggest that this amounts to an irreconcilable problem. He suggests that it may "plausibly be argued that the conventions that nullify some of the Constitution's textual provisions are among the underlying assumptions that explain why this text was not amended more than it was in 1982 or at some other point" (341). The conventions themselves, in other words, can make the formal repeal of superfluous textual provisions unnecessary. This logic may work for conventions surrounding the disuse of particular textual provisions, but it is not clear how it operates with respect to conventions that shift practical power to actors away from the formal authorities (usually the Queen or governor general) referred to in the constitutional text. Moreover, Sirota notes that the Court itself is not clear on whether conventions "become law" under the architecture or whether it has confined its use of the concept to attempts at legislative workarounds that alter or create conventions. But if the Court's approach does imply conventions are now law, this would seem antithetical to the architectural concept itself, for it means that the formal law of the Constitution has been altered without recourse to the amending formula in Part V, itself a fundamental element of the Constitution.
Another issue raised by the legalization of conventions would be the impact on actual practice and institutional behaviour. The entrenchment of conventions would presumably render departures from convention unconstitutional (in the legal sense) generally, or at least justiciable. Sirota notes that for the Court to take conventions into account properly in the Senate Reform Reference, it would have looked to conventions "to find that the 'powers of the Senate' were circumscribed by a duty of deference to the elected House of Commons, and that the 'method of selecting Senators' entrenched by paragraph 42(1)(b) involved prime ministerial discretion rather than democratic choice. Both these features are thus protected from unilateral amendment by Parliament" (2020: 343). If conventions are now law, and the Court were to recognize a convention requiring deference on the part of the Senate, then any action by the upper house to exercise its full powers as enumerated in the constitutional text would somehow be contrary to the Constitution. The Senate has asserted itself as a competitor to the House of Commons on occasion throughout Canadian history (Macfarlane, 2021a).
For his part, Sirota advocates for courts to rely on conventions only insofar as they are relevant to the interpretation of specific constitutional provisions, and in his view, the clear constitutional text should prevail. Thus, for example, the federal powers of reservation and disallowance "could not 'be interpreted away by invoking conventions '" (2020: 345-46). This seems to imply that not all conventions are entrenched in the sense of making them tantamount to constitutional law or at least not to the extent of permitting them to conflict with unambiguous constitutional text. Sirota also notes that "while the Court set out a broad reading of Part V, mere governmental practice, or even the development of an opinio juris making practice obligatory, are beyond its reach" (356). As with Albert's amendment by stealth, the act of legislating with intent to create a binding convention is illegitimate, whereas practice that eventually comes to be regarded as convention is legitimate, for the purposes of Part V.
Yet if conventions are considered law under the architecture of the Constitution, this distinction is difficult to reconcile. For some reason, new conventions can emerge and become law, absent Part V, so long as it is through practice rather than legislation. But presumably once established, the same conventions cannot be altered without recourse to Part V, even in the case of departures from convention absent legislation. Sirota even says that alterations to mere political practice are permissible and points to the recent innovation of a merit-based, non-partisan appointments process to the Senate as an example of the "mere governmental practice" beyond the reach of Part V. But he also seems to suggest that if and when those practices crystallize into convention, they become immutable under the architecture. On this particular reading of the Senate Reform Reference, new conventions may come to be part of the architecture of the Constitution (how and when, we do not know-presumably until the Court tells us), and it is not clear whether a political actor could depart from them as a matter of law. Sirota does not explain this other than to say "it is appropriate to treat practice and legislation as distinct mechanisms for constitutional change, as the Senate Reform Reference does" (2020: 358), but this distinction seems more consistent with the traditional understanding of conventions as political rather than legal obligations.
What results from this particular understanding of the place of conventions in the constitutional architecture is that it allows additions to the Constitution without recourse to Part V but not changes to the Constitution. This presents another fundamental problem, for it would only further exacerbate the constitutional stasis Canada suffers from, stemming from an onerous formal amending formula, a political culture antagonistic to formal constitutional change, and jurisprudence that has made institutional reform more difficult (Albert, 2019; Macfarlane, 2016, 2019).
Entrenching constitutional conventions could effectively freeze the last major element that gives the Canadian Constitution a degree of flexibility. It would also make Canadian constitutional development all the more dependent on the will, ability and discretion of the courts, wresting democratic control over the fundamental rules for society from resolution by politics and handing them over for resolution by judges.
Many of these challenges and problems arise in the context of justiciability of conventions generally. Given that there seems to be growing receptivity to the justiciability of conventions, it is an issue worth exploring further.

Dangers of Justiciability
As noted above, courts in the United Kingdom and in Canada have taken notice of conventions in the course of interpreting law, most commonly to recognize them as crucial background material to contextualize governmental functioning. In India, the Supreme Court has gone even further and has "read in" conventions to rule that the chief justice enjoys "primacy" in judicial appointments such that their consent is required, despite constitutional text that only requires consultation by the executive (Ahmed et al., 2019a: 800). Farrah Ahmed, Richard Albert and Adam Perry argue that courts might legitimately treat conventions as justiciable when dealing with "power-shifting" conventions-those rules that "transfer power from those who have legal power to those who can legitimately wield it" (2019b: 1155). For example, they assert that courts could enforce the convention that the monarch must follow ministerial guidance.
Yet it is not clear that even the example of a rogue governor general cannot be dealt with as a political matter in the Canadian context. A governor general who acts in a manner fundamentally at odds with the role can be dismissed, or the issues core to a dispute between a prime minister and a governor general might be best settled democratically, via an election. Moreover, the idea that courts are well equipped to deal with power-shifting conventions is highly debatable. As Eugene Forsey (1984) notes, there are serious capacity issues with regard to the courts that arise in the context of conventions. Dodek writes that "conventions do not lend themselves to the definitive answers that courts are good at providing in binary litigation " (2011: 134). Even scholars open to the idea of the justiciability of conventions have been deeply critical of the courts' performance when they engage with them (Heard, 2014).
Ahmed, Albert and Perry's proposition does not heed the architectural problem presented by the enforcement of conventions. They assert, for example, that courts could appropriately invalidate an attempted use of the federal powers of reservation and disallowance on the basis of the convention of disuse (Ahmed et al., 2019b(Ahmed et al., : 1163, but this would plainly conflict with the amendment requirements in Part V. Indeed, it would dramatically shift an even greater degree of amendment authority to the courts, who would be empowered to determine if and when conventions somehow crystallize into law-something the authors themselves note the Indian Court did, even though it "offered no clear reasons for concluding that it ought to enforce them. It assumed, rather than justified, its role as constitutional guardian" (1155).
Heard also writes favourably about the potential for the justiciability of conventions, noting that the Court's "embracing" of unwritten constitutional principles "raises the justiciability of at least some conventional rules that both rely on and give life to those same principles" (2014: 206). In Heard's view, courts might recognize or enforce the most fundamental or important ("semi-rigid") conventions, at least in part to avoid blind enforcement of archaic or antiquated rules "divorced from constitutional reality" (230). Heard is correct that if courts were to blindly apply the written text of the Constitution without any appreciation for actual practice, then judicial outcomes could border on the absurd. Yet it is not clear, for the reasons already explained, that formal recognition or enforcement of conventions is necessary to avoid that absurdity. Courts have relied on practice as background context for constitutional interpretation in myriad ways, including in the context of federalism jurisprudence (Burningham, 2021). This falls well short of formal judicial recognition or declaration of certain practices as convention, let alone enforcement.
Nor is it clear that fundamental conventions are as well suited to judicial protection as Heard suggests. In any extreme example, such as a government losing a clear, formal vote of confidence in the House and attempting to stay in powersomething that would, in clear circumstances, at least, amount to an effective coup d'état in Canada's system-there is little reason to think actors taking such drastic, anti-democratic action would heed a court opinion. In less extreme, but still severe, breaches of a fundamental convention, such as a federal government's decision to disallow provincial legislation, there is every reason to think that the political mechanisms that exist to enforce conventions are the better measures to resolve the resulting conflict.
The 2008 prorogation affair is an excellent example of the ways politics is better suited to resolving conflict over convention. Five weeks after the 2008 federal election, a controversial fiscal update by the Conservative minority government prompted the leaders of the Liberals and the New Democratic Party (NDP), with the support of the Bloc Québécois, to unveil a formal written accord to form their own minority coalition (Russell, 2015). In a bid to stave off a confidence vote, Prime Minister Stephen Harper asked the governor general to prorogue. After some deliberation and consultation with advisors, the governor general agreed, albeit on the condition that a new session would begin only 10 days after the normal holiday recess and that the government submit to an immediate confidence vote when it did. The coalition collapsed over the period of the break, in part because of the weak position of the Liberal Party, whose leader, Stéphane Dion, had already signalled his intention to step down, and in part due to the government withdrawing some of the more controversial elements of the budget.
As Russell (2015) notes, most constitutional scholars agreed there might be hypothetical circumstances under which a governor general could deny a prorogation request, but there was sharp disagreement over whether doing so would have been appropriate here. The dominant view was that the governor general made the right decision (Heard, 2014: 68), and the way in which things played out seem to support it. Heard cautions that we ought not view the coalition's fortunes as evidence about how it might have fared had it formed government, arguing that Harper would not have survived and that, in his view, the governor general should have refused (70). Yet others correctly point out that a refusal could have sparked a full-blown constitutional crisis (Monahan, 2012) and that "the opposition parties had failed to establish sufficient democratic legitimacy for their own proposal," in part because "the Liberals had vigorously denied, during the election campaign, that they would consider a coalition" (Webber, 2015: 94).
My point here is not to rehash the debate over the governor general's decision but to note that had judicial review been sought to settle the constitutional debate, it would have risked bringing the courts into a deeply contested and explicitly partisan affair, even raising the spectre of the Supreme Court ultimately determining which party or parties form government. Had this occurred in a context where there was another constitutional actor-the governor general-properly playing the role of constitutional umpire, and where the political resolution, embedded in the core convention of responsible government itself, ultimately settled the matter, it would have been the judicialization of politics at its most extreme. Far beyond influence as a policy maker, it would mean the courts effectively take over an explicit and core executive function. This would be a significant violation of the separation of powers.
Dennis Baker reminds us that the conventional view is that there is no separation of powers in Canada, a result of the perceived fusion of the legislative and executive branches (2010: 8-16). It is a view that belies the formal and functional distinction in the roles and powers of those two branches, and Baker argues persuasively for recognizing the "partial agency"-some degree of mixing of functional powers across branches-that imbues a Canadian separation of powers doctrine with proper recognition of the nuance and complexity at play (11). As Baker explains, where the executive does indeed dominate the legislature in practice, by setting the agenda and by originating money bills, it nonetheless must maintain the confidence of the legislature, which can in practice, particularly in minority government situations, give the legislature a significant degree of control. Moreover, even the conventional view recognizes a pronounced separation between the judicial branch and the executive and legislative branches, such that courts, protected by judicial independence, are widely viewed as providing an (perhaps the) essential "checks and balances" in Canada's separation of powers (although that term does not tend to appear in judicial discussions of the independence of their branch).
Importantly, however, Baker's discussion of separation of powers occurs in the context of constitutional interpretation-indeed, the idea that partial agency necessarily invites coordinate interpretation of the Constitution by the various branches in order for each of the them to fulfill their proper legislating, implementing and enforcement roles. Courts expound on the law. Conventions are political rules underpinned by democratic principles. The core conventions especially ensure that Canada's formal constitutional monarchy operates as a functional democratic constitutional system in practice, by locating the exercise of power in the appropriate actors who are themselves subject to democratic mechanisms and whose choices are constrained or informed by subsidiary conventions. The conventions surrounding responsible government also ensure that the guardians of responsible government are already designated-in some contexts that may mean the governor general, in others the legislature, and in others still the electorate. Judicial enforcement would not merely intrude on the functions of the executive or legislative branches but would fundamentally transport the courts away from the law and into the nucleus of executive and legislative power relationships, which are properly governed by politics. It would, in other words, upend any coherent separation of powers that animates Canada's governing system.
Indeed, judicial review of a prorogation decision would arguably precipitate a bigger constitutional crisis than if the governor general had refused a prime minister's request. In the 2008 affair, politics resolved the matter. Had the coalition been determined to form government, it was not prevented from doing so at the first opportunity, a confidence vote then scheduled for 10 days after the normal recess. One of the reasons the coalition crumbled was that the threat of no confidence compelled the government to significantly alter its budget plan, including scrapping a controversial proposal to eliminate the per vote subsidy awarded to political parties (Banfield, 2015). There is little doubt the opposition parties also took stock of public opinion during the break. In short, no matter what one might think about the unprincipled use of prorogation in delaying a potential no confidence vote (not to mention the government casting doubt on the constitutional legitimacy, rather than merely the political legitimacy, of the coalition), responsible government was preserved, and the government ultimately backed down from the policies that provoked the situation.
No litigation arose from the 2008 prorogation affair, but Canadian courts have dealt with misguided legal challenges based on purported (perhaps imaginary) conventions. One significant example is litigation launched over the federal fixed-date election legislation and whether the Harper government violated a new convention when it requested dissolution in 2008. Enacted in 2007, the law established that each general election be held on the third Monday of October in the fourth calendar year following polling day for the last general election. However, the law was enacted with a provision preserving the powers of the governor general, "including the power to dissolve Parliament at the Governor General's discretion" (section 56.1 (1)). This was because any changes to the powers of the governor general would implicate section 41(a) of the amending formula, requiring the unanimous consent of the provinces for changes to the office.
Despite this explicit proviso, which means the law's practical effect is to establish a maximum duration for a Parliament no later than the fixed date, litigation was launched after the Harper government dissolved Parliament in 2008, only a year after the legislation's enactment. The legal claim raised a number of issues, including whether the law created a new constitutional convention limiting the prime minister's ability to request a dissolution and whether the prime minister's advice was contrary to the law. The Federal Court (Conacher v. Canada, 2009) andFederal Court of Appeal (Conacher v. Canada, 2010) both made short work of these questions, finding that the applicants failed to establish the existence of a new convention and that the broad language employed in the legislation makes no note of specific situations under which the governor general was not free to dissolve Parliament or when the prime minister may or may not bring forward a request.
Critics of the Federal Court and Federal Court of Appeal decisions argue that the purpose of the legislation was to prevent prime ministers from taking political advantage of the power to dissolve Parliament and calling snap elections. This maps onto Heard's argument that conventions can emerge from the explicit agreement and even declaration of relevant actors, rather than necessarily relying on past practice (2014: 7-8). The fixed-date election law highlights the uncertainty that arises in this context. The claim that a single prime minister or government can, in effect, bind its successors to a rule by making a political promise and enacting non-binding legislation to bring it into effect, and the result is a new, judicially enforceable convention, is at best dubious. It mirrors precisely the sort of "amendment by stealth" legislation the Supreme Court rejected in the Senate Reform Reference. As Rainer Knopff writes, this "poses a conundrum for anyone who believes that other workarounds in the form of 'ordinary legislation' are unconstitutional to the extent that the conventions they inaugurate rapidly overwhelm their formal loopholes" (2016: 141).
Controversy arose again in 2021, when NDP leader Jagmeet Singh publicly called on the governor general to deny a dissolution request from the prime minister two years into a minority government (Levitz, 2021). The fact that the controversy has repeated itself does not support the idea of judicial review so much as it further demonstrates that the fixed-date election law has not established a new convention. At the federal level we now have a record of prime ministers in two successive governments, of different partisan stripes, exercising discretion to request dissolution consistent with long-established practice. Provincial fixed-date election laws have also been repeatedly ignored in this sense (that is, early election calls absent non-confidence votes) by eight of the nine provinces that have such a law on the books. 1 The idea that these laws establish new conventions preventing such practice should be dismissed on this fact alone.
Criticisms about judicial reliance on the Jennings test remain valid, as there may be circumstances where conventions exist absent precedent, but this does not translate into the idea that any political promise from a constitutional actor, even if grounded in a principled reason, automatically constitutes a convention. Peter Aucoin, Mark Jarvis and Lori Turnbull suggest one problem emerging from the judicial decisions in the fixed-date election case is that it gives the prime minister veto rights, and that "in this instance, the prime minister was even able to veto a change in the convention after he had both proposed it as a campaign promise and then endorsed it as government legislation" (2011: 85). Except it is not clear why the inverse is not a concern: that a single government can, in effect, establish a binding convention that imposes obligations on its successors, especially in a context where the non-binding nature of the rule was explicitly written into the legislation. One might instead conclude that a convention that is "broken" on the very first occasion in which it would have regulated the behaviour it purportedly existed to regulate was never a convention to begin with.
This raises another issue in the context of justiciability: the risk of an inflationary or overly broad view of enforceable conventions. The controversy over the fixeddate election law conflated mere practice (a political rule, supported by reasons, but for which there is no strong consensus) with convention. Similarly, Philippe Lagassé argues that much of the debate over the 2008 prorogation affair confused norms for conventions. Norms (also distinct from practice), like conventions, "are morally-binding rules of proper conduct" and can "define what is acceptable behaviour in a constitutional context, including which actions are necessary to uphold fundamental principles" (Lagassé, 2019: 3). They differ from conventions in that they "refer to actors' willingness to respect both the letter and the spirit of the Constitution. In this sense, norms are a type of meta-rule that ensure laws and other rules are followed properly" (3). In the prorogation context, the convention is that the governor general follows the advice of the prime minister, and thus convention was followed. The problem, as Lagassé states it, is that the prime minister arguably failed to act fairly and honourably, while the opposition parties were acting contrary to another norm, "namely that coalitions should not be formed by parties lacking a plurality of seats unless their intent to form a coalition was clearly stated during a general election" (4). Justiciability of conventions not only risks transforming binding political rules into enforceable law but risks transforming practices and norms into enforceable law as well. There would be scant room left for political disagreement and discretion if all of these things became the purview of judicial determination.
Finally, it is worth noting that the principle underlying the purported convention in the fixed-date election law context is itself open to debate. It is true that there are valid concerns about first ministers' capacity to call snap elections. 2 But it is equally valid to wonder why the unwritten constitutional principle of democracy does not weigh in favour of allowing the electorate to decide whether to reward or punish governing parties that are perceived as taking advantage of that power. To the extent that scholars are open to judicial enforcement of conventions on the basis of those underlying principles, it is worth emphasizing that the principles themselves are contested and often in conflict-another reason to avoid judicial involvement in explicitly political and partisan matters and instead rely on the political and democratic accountability mechanisms conventions exist to support and facilitate. As Jeremy Webber notes: Constitutional conventions regulate the very core of the democratic process. If they were enforced by courts, judges would be making decisions that must be the province of other actors-in the case of responsible government, who gets to form the government. Moreover, in true conventions, there are sufficient mechanisms to enforce the rule without the intervention of courts, generally through the exercise of the Governor-General's authority, through appeal to public opinion either directly or at a future election, or by government simply becoming unworkable. These recourses are awkward and sometimes cataclysmic and they are generally dependent on the engagement of the citizenry but they are preferable to the intervention of judges.
Conventions are yet another example of the fact that the constitution is not and cannot be the exclusive province of the courts. It is sustained by the practices of all constitutional actors. If it is to remain democratic, there is therefore no substitute for the surveillance of an engaged, knowledgeable and demanding citizenry. (2015: 98) This point cannot be emphasized enough. Some might respond that it puts too much faith in the idea of an active and knowledgeable citizenry. Yet the judicialization of politics that judicial enforcement of convention would invite should be deeply concerning: the transportation of conventions to the realm of law and the judiciary not only undermines their purpose and the correlating values of democracy and accountability that support them but also risks further atrophy of the broader political culture around the Constitution and the role of elected officials and other non-judicial actors in supporting, defending and animating constitutional principles. There are valid concerns about the state of democratic accountability in Canada, but it is difficult to see how that situation improves if we abandon the idea that political actors and citizens have any enforcement role to play at all. It also goes to the heart of what Christopher Manfredi describes as the paradox of modern liberal constitutionalism: when "political power in its judicial guise is limited only by a constitution whose meaning courts alone define, then judicial power is no longer itself constrained by constitutional limits" (2001: 22). Justiciability of conventions would extend this paradox beyond the legal constitution and into the heart of the political constitution, undermining democracy in a manner far more troubling than debates about judicial review of federalism or rights have thus far contemplated.

Slamming the Door Shut: Concluding Thoughts
The constitutional architecture, as described by the Court in the Senate Reform Reference, threatens to untether conventions as mechanisms of political flexibility and accountability and instead crystallize them into the amber of constitutional law. If this is what the Court intends, it would amount to the amputation of central limbs from the Constitution's living tree. The very purpose of conventions is tied to their method of enforcement: politics. From Forsey to Russell to the courts themselves, it has long been recognized that conventions should be enforced by public opinion and other democratic measures (Russell, 2004: 129;Aucoin et al., 2011: 77). As Dodek writes, political morality changes and evolves, and litigation is not something that can keep up, making justiciability "destined to freeze conventions at a certain point of time " (2011: 132). Even those open to the prospects of justiciability recognize that conventions should, in many instances, be "used to resist a legal conclusion rather than to support one" in order to ensure political accountability remains robust rather than overcome by the introduction of unnecessary legal constraints (Ahmed et al., 2019a: 791).
This article has explored the incoherence and outright paradox of embedding conventions as law within the constitutional architecture. The idea raises the spectre of not only further constitutional stasis but also unknown and unintended consequences for political practice and constitutional change. It also undermines the concept of a coherent architecture, pitting the amending formula against judicial amendment of the Constitution via the legalization of conventions (Macfarlane, 2021b).
The Court will no doubt have a future opportunity to clarify its intent with regard to the place of conventions in the architecture and to articulate an understanding of the Senate Reform Reference in a way that does not entrench them as Sirota, Scholtz, and others contend. The place of conventions in the Court's decision is best understood by Mark Walters (2013), whose article on the issue was cited by the Court. Rather than legalizing conventions, the Court viewed the legislation at stake in the reference as an attempt to bind practice in a way that could not be done formally, because of the amending formula. The issue for Walters is not that conventions might change but that Parliament attempts, through legislation, "to inform, supplement or modify conventional practice in ways that conflict with, or are different from, the Constitution of Canada" (2013: 56). Walters recognizes the potential contradiction discussed above: "political actors in Canada may be permitted to do things according to convention that federal and provincial legislatures in Canada may be constitutionally prohibited from supporting, supplementing or modifying through legislation" (56). Yet for Walters, the distinction is not so clean as one between legislation and practice. He notes, for example, that legislatures are free to modify conventions with law. The limits of doing so arise when fundamental structures in the Constitution are changed by ordinary law.
Walters provides the example of legislation that modifies practices of patronage and partisanship in the Senate (something that would later be accomplished by Trudeau via a new appointments process) as likely "constitutionally sound" because patronage and partisanship are not essential to the Senate's constitutional role (2013: 57). Thus the problem with the consultative elections proposal at issue in the reference was not "because it conflicts with constitutional convention or because it interferes with the Governor General's legal power to appoint senators, but because its purpose and effect is to transform the upper legislative chamber from the essential form it is given by the law of the Constitution of Canada into something very different" (57).
It is this understanding that is directly cited by the Court. The result is that certain changes to convention may not be permitted, not because conventions are entrenched as law in the architecture but because attempts to impose changes in law can threaten illegitimate changes to the architecture. Such a reading preserves conventions in their proper place as political rules of accountability rather than legal objects of enforcement by courts. It is also consistent with fundamental precedents on the legal status of conventions from Patriation to Osborne. Much of the uncertainty over the Court's depiction of the architecture concept might have been avoided if the justices had limited themselves to the more straightforward and parsimonious conclusion that consultative elections were a change to the "method of selection" in section 42(b) of the amending formula. In other words, it should have been enough to conclude that the formal imposition of a new process (consultative elections), required by law, into the overall method of selection is an attempt at constitutional change, rather than engaging in speculation over the impact such a change would have on constitutional practice (specifically, prime ministerial discretion). The Court's invocation of the nebulous architecture concept has spurred considerable speculation about the status of conventions and renewed debate over their justiciability. The dangers of opening this door are too great, and the Court would be well advised to slam it shut the first chance it gets.