Why does Kant Think that Democracy is Necessarily Despotic?

Abstract Kant’s criticism of democracy has been traditionally defused with the consideration that Kant’s aversion is not to democracy per se, but to direct democracy. However, what Kant says – ‘to prevent the republican constitution from being confused with the democratic one, as commonly happens’ (ZeF, 8: 351) – appears to count not only against direct democracy, but also against conceptions of democracy closer to the ones we are accustomed to. By offering a new account of what Kant sees as the real problem of democracy (direct or not), the article unpacks a lesson about the limits of democracy that has gone largely unnoticed among political theorists and Kant specialists.


Introduction
In an important passage of Toward Perpetual Peace Kant comments on his first definitive article ('The civil constitution in every state shall be republican' (ZeF,8: 349) 1 ) to make sure that the republican constitution is not 'confused with the democratic one, as commonly happens ' (8: 351-2). The conflation is to be avoided, explains Kant, because the democratic form of sovereignty, unlike the oligarchic or the autocratic/monarchical forms, is the only one that 'in the strict sense of the word is necessarily a despotism ' (8: 352). For the contemporary reader, Kant's distrust of democracy may come as a surprise. His moral philosophy seems to be committed to a strong notion of equality among human beings, henceit is reasonable to assumeamong citizens, regardless of how this ideal may be in tension with his belief in passive citizenship. The same commitment to equality and self-rule seems to emerge from the enthusiasm that the French Revolution, Kant tells us, cannot fail to spark in any disinterested observer, despite the philosopher's notorious denial of people's right to rebel against even the worst ruler. Most importantly, Kant endorses Rousseau's basic tenet that no law can be legitimate unless it canat least in principleenjoy the consent of all consociates, 2 a principle that sounds even more demanding (we would say more 'democratic') than current criteria of political democracies only or whether it may be extended to the 'representative' liberal democracies we live in. I will then move (section 4) to construe from Kant's materials the ideal of a democratic republic no longer infected by despotic tendencies, thereby following an evolution that Kant himself seems to have undergone towards a more benign consideration of democracy in the Metaphysics of Morals. Finally, I will suggest (section 5) that what appeared to most commentators as a dramatic change of opinion from 1795 to 1797 should be rather understood as a two-step complex, yet consistent, argument in which Kant first focuses on the structural and peculiar defects of democracy (Toward Perpetual Peace) and then highlights (Metaphysics of Morals) how, despite these structural defects, democracy is the form of sovereignty that de facto has the least probability of despotic degeneration.
Before beginning, however, it is important to clear the table of a possible misunderstanding concerning the apparently insurmountable opposition between my critical reading and the effort by Ingeborg Maus in Zur Aufklärung der Demokratietheorie (1992) to find in Kant's political thought a tool to defend and revitalize democracy by vindicating people's sovereignty and its immediate and supreme legislative authority. While Maus endorses the reassuring reading that Kant is no enemy of democracy per se, but only of its, in her terminology, 'ancient' form, given its failure to separate powers (Maus 1992: 194-5), she interestingly stresses the same point I most care about when she highlights that democracy contradicts the original contract and becomes despotic when in the exercise of power 'the general will loses its generality ' (p. 194). In her view, Kant understood that this happens in a democracy because the executive does not conceive of itself as a mere agent of the legislative, but, being open to 'all', takes itself to be on a par with that legislative power (p. 196). In other words, the executive carves for itself a space for manoeuvring independent of the legislative, and this opens the door to despotism. 7 While I do not look at specific features of the democratic executive to reconstruct Kant's reservations, I share with Maus the idea that Kant saw with full clarity that political systems are unjust to the extent to which political actors (I would say any political actor, including representative parliaments and democratic citizens) take themselves to be authorized to use their share of power without adopting the perspective of the general will. 8 Indeed, even her defence of radical, 'non-representative', immediate democracy is nowhere to be understood as an authorization, let alone an encouragement, to let citizens' private interests freely clash in a competitive scheme where majority rule decides. Maus explicitly criticizes this 'individualistic' view of democracy (Maus 2011) and is keen to defend the centrality of the notion of general will. She defines it, and I readily endorse this definition, as a discursive process in which free and equal citizens move through public debate from expressing and defending their private viewpoints and interests to upholding terms of agreement in principle acceptable to all others. 9 On my reading, Kant's criticism of democracy is the same as Maus' criticism of its 'individualistic' tendency.

Democracy's necessary despotism in Toward Perpetual Peace
Kant discusses democracy in two important works of the critical period, Toward Perpetual Peace (1795) and the Doctrine of Right in the Metaphysics of Morals (1797), makes passing yet significant references in the Anthropology (1798) and in the Contest of the Faculties (1798) and gives important indications in the preparatory notes to his published works. It is in Toward Perpetual Peace and its preparatory notes that we find the strongest and most explicit reservations about democracy. Right after having introduced the general thesis of the first definitive article -'The civil constitution of every state shall be republican' (ZeF, 8: 351-2) -Kant adds a few remarks 'to prevent the republican constitution from being confused with the democratic one, as commonly happens' (ZeF,8: 352). 10 Before addressing Kant's criticism, we need to recall a few basic theses of his political thought. To begin with, for Kant the various forms of state can be classified by using two different criteria: either by taking into account how many rule (the form of sovereignty, or forma imperii) or by focusing on the way in which the sovereign power is exercised (the form of government, or forma regiminis). The former yields the usual tripartition into monarchy (which Kant calls here 'autarchy'), oligarchy and democracy. The latter 'relates to the wayas defined by the constitution (that is, an act of the general will whereby the mass becomes a people)in which the state makes use of its plenary power' (ZeF,8: 352), and there are only two ways in which a state can make use of its power, the republican and the despotic. A state is republican if the executive power is separated from the legislative. It is despotic if this separation does not occur and 'the laws are made and arbitrarily executed by one and the same power' (ZeF,8: 352).
Moreover, a republic is defined as a system based on three principles: 'first on principles of the freedom of the members of a society (as individuals), second on principles of the dependence of all upon a single common legislation (as subjects), and third on the law of their equality (as citizens of a state)' (ZeF,. In an all-important footnote devoted to explaining these 'principles', Kant points out that 'My external (rightful) freedom is : : : to be defined as follows: it is the warrant to obey no other external laws than those to which I could have given my consent ' (8: 350). In the Metaphysics of Morals he repeats the point saying that 'the laws it [the legislative power] gives must be absolutely incapable of doing anyone injustice' (MS, 6: 313). A republic is thus a political system in which powers are separated (they cannot be held, as we shall see, by the same individuals) and these three 'principles', including the requirement that laws must not do anyone injustice, are respected.
As Kant often implies, the requirement that political power should not do anyone any injustice is equivalent to saying that it must rule from the perspective of the general will. 11 The obligation to rule from the general will's perspectivethis 'burden of representation', we might call ithinges in the same way on one, some or all who are called to rule. A king can act 'in the spirit of a representative system' if he does his best to interpret the general will in his rulings. Kant makes this point clearly in the Contest of the Faculties when he claims that there could be monarchies in which the king is 'acting by analogy with the laws which a people would give itself in conformity with universal principles of right' (SF,7: 184). Mutatis mutandis, the same can be said of the oligarchy.
At this point one would expect Kant also to grant to democracy this republican potentiality. But Kant thinks that in a democracy, at least in a democracy 'in the strict sense of the word', ruling cannot be exercised from the point of view of the general will and despotism inevitably takes root. It is obviously crucial for our purposes to understand why this is the case.
Let us first clarify that by democracy 'in the strict sense of the word' Kant probably means direct democracy. Indeed, in the preparatory notes for the two works in question this qualification recurs in slightly different forms: Demokratie, in der eigentlichen Bedeutung des Worts is equated with ochlocracy (VAZeF, 23: 161); bey der eigentlichen Demokratie (VASF 23: 432) is characterized as resting on a despotic identification of the state with the people; bloße Demokratie (VAZeF 23: 166) is most significantly contrasted with a 'democratic constitution in a representative system', considered as the essence of republicanism.
Why, then, is direct democracy necessarily despotic? The reasons that Kant offers are contained in two dense and rather obscure passages. On the one hand, Kant says: Of the three forms of state, that of democracy in the strict sense of the word is necessarily a despotism because it establishes an executive power in which all decide for and, if need be, against one (who thus does not agree), so that all, who are nevertheless not all, decide; and this is a contradiction of the general will with itself and with freedom. (ZeF,8: 352) As a way of explaining the point, Kant adds the following: This is to say that any form of government which is not representative is, strictly speaking, without form, because the legislator cannot be in one and the same person also executor of its will (any more than the universal of the major premise in a syllogism can also be the subsumption of the particular under it in the minor premise); and even if the other two state constitutions are always defective insofar as they leave room for this kind of government, in their case it is at least possible for them to adopt a kind of government in conformity with the spirit of a representative system, as Frederick II, for example, at least said that he was only the highest servant of the state, whereas a democratic constitution makes this impossible because there everyone wants to be ruler. (ZeF,8: 353) Byrd and Hruschka claim that the reason why direct democracy is necessarily despotic is that in Toward Perpetual Peace (but not in the Metaphysics of Morals) Kant means by forms of sovereignty only the executive power. If that is the case, the necessary despotism of direct democracy becomes clear: 'If the executive power is in the hands of all then there is no one left to hold the legislative power and the two cannot conceivably be separate' (Byrd and Hruschka 2010: 178). But in 'representative' democracies elected governors are only a few, hence the problem would not arise. Indeed, Byrd and Hruschka think that what we call representative democracy today is Kant's ideal form of state (p. 167). In the preparatory notes, Kant himself after all identifies a republic with 'a democratic constitution in a representative system' (demokratische Verfassung in einem repräsentativen System) (VAZeF, 23: 166). 12 Unfortunately, things are not so simple. To begin with, it is highly dubious that by democracy Kant means only the executive power in Toward Perpetual Peace (or elsewhere). He says that the three forms of sovereignty (autocracy, aristocracy, democracy) are distinguished 'according to the different persons who have supreme power (oberste Staatsgewalt)' and that depending on the form of sovereignty: 'either only one, or some in association, or all those together who constitute the civil society possess sovereign power' (ZeF,8: 352). How can democracy be merely the executive branch if it is described as 'supreme power'? How can democratic rulers be merely the governors if they are said to possess 'sovereign power'? 13 Like most political theorists, Kant considers the executive power to be 'under the legislative' (Preparatory Draft of DoR, 23: 352; MS, 6: 317). Kant evidently calls 'democracy' the whole political system of a state that adopts the democratic form of sovereignty.
Not accidentally, the textual evidence that Byrd and Hruschka provide in support of their reading is weak. They cite Kant's claim in Toward Perpetual Peace that democracy 'establishes (gründet) an executive power in which : : : ' (ZeF, 8: 352). But of course saying that democracy establishes an executive power is different from saying that it is such a power, even less that it is exhausted by it. The most natural reading is that democracy, as a system taken in its entirety, is necessarily despotic because it grounds/yields/rests on an executive of the sort described, that is, one capable of deciding against the individual, if need be. 14 Secondly, is it really true that in direct democracies, at least those that have indeed existed, all citizens hold executive powers? Even in direct democracies the set of individuals with executive tasks is limited, hence does not coincide with the set of those who hold legislative power. In Pericles' Athens, for example, the executive power was exercised by a restricted group of citizens. The whole system, heavily dependent on poll and rotation, ensured that there was no systematic overlapping between executive and legislative powers, and that only a ratio between 1/30 and 1/50 of the citizens holding legislative powers also had executive tasks. So the two sets had some overlap, but never coincided. More generally, a political system in which executive power is held by all citizens is hardly conceivable: no society can afford to keep the whole population busy doing public tasks (and noneto my knowledgehas ever existed).
Probably the feature of direct democracies of the past that Kant is targeting is that they lacked a constitutional framework (a legislative power as Rousseau and Kant understand it) that could work as a constraint for the changing and partisan will of the assembly (as in ostrakismos). Briefly put: the legislative branch, as Rousseau and Kant understand it, is simply absent in Greek democracies. If that is the case, the assembly, the supreme organ that makes political decisions, is necessarily despotic because it is free to issue decrees unconstrained by an established legal framework (constitutional in the strict sense or at least not changeable through a simple executive decision) when no specific, partisan, ultimately private interests are at stake. It follows that the problem is not with the executive per se, but with the conflation of executive and legislative functions performed by the assembly. This is the inevitable being 'without a form' (Unform; ZeF, 8: 252) of direct democracies, their systematic conflation of legislators and governors as lamented by Kant through the metaphor of the syllogism cited above.
While the absence of this legal framework is the institutional basis of despotism, below it lies a more fundamental cause which relatesquite obviouslyto the very partisan nature of the agenda pursued by whoever exercises political power. That for Kant the ultimate cause of despotism is not the institutional defect just explained but the very partisanship of the agenda of rulers, which is at most facilitated in its execution by that institutional defect, is revealed by looking at the way Kant discusses the case of Frederick II. His attitude is that he should 'represent' the state, actually be the 'servant of the state'. As ruler he does not pursue a partisan interest (or at least he claims this is not what he is doing), but merely acts according to his best interpretation of what is 'good' for the people, understood as what the general will wills, in turn understood as what protects and furthers the values of freedom, equality and independence of the citizens. 15 As Kant puts it in the passage already quoted, Frederick II is 'acting by analogy with the laws which a people would give itself in conformity with universal principles of right' (SF,7: 184). Frederik II could implement a partisan agenda, he would have all the institutional latitude to do so, but he (supposedly) chooses differently. It is his mindset that ultimately determines the quality of his political action.
Of course, it is not impossible that citizens in the assembly of a direct democracy act in the responsible manner attributed to Frederick. And yet, and this is the crucial difference, they are allowed, if not expected, to represent themselves only. This is the opposite of the representative attitude Kant thinks should characterize the act of ruling. While monarchs and oligarchs are not allowed to pursue private interests (if they did so openly, they would fall in the category of barbarism, as per the taxonomy in the Anthropology), and quite independently of whether this is precisely what they often do, albeit covertly, democratic rulers have this horrific peculiarity. They are allowed to advance private interests, broadly understood as to include not only material but also 'metaphysical' ones, such as favouring one creed or ideology over another. If they make a strategic use of their political rights, thereby setting terms of cooperation others could never agree to, nobody can charge them with political indecency.
If we adopt this perspective, the rest of the anti-democratic passages quoted above become rather clear. In particular, we can understand: (a) why democracycertainly direct democracy, but, we shall see, also representative democracyestablishes an executive power in which all citizens may make decisions about and indeed against one individual; (b) why this leads to a contradiction of the general will with itself (and with freedom); (c) why every citizen's desire to be a ruler displays an attitude diametrically opposed to Frederick II's style of ruling; (d) why only democracy, and not the other two forms of sovereignty, is despotic; and (e) why it is so 'necessarily'.
(a) As pointed out by some interpreters (Pinzani 2008;Byrd and Hurschka 2010), the point that all citizens may make decisions against one individual seems to refer to ostracism, that is, the practice of forcing an individual considered as dangerous (often a mere political opponent, as in the case of Themistocles) out of the city. Independently of whether Kant had ostracism in mind, the crucial task is to understand what is wrong with decisions in which some individuals 'disagree'. The problem cannot be the actual dissent of some individuals because many political decisions genuinely taken with the common good in mind de facto displease some citizens. Moreover, we need to understand why Kant thinks that 'decisions are made by all the people and yet not by all the people', a question probably equivalent to the following one: why decisions are supposed to be 'made by all the people' if ex hypothesi some individuals disagree? If their actual disagreement cannot be a problem, it must be principled disagreement that Kant has in mind. Hence, the scenario suggested seems to be one in which a faction in the assembly has enough power to impose a decision that other factions could not accept even in principle. This certainly may happen when people are allowed to protect their partisan interests. If they do, the assembly is not pursuing the general good, but it is merely imposing the will of the most powerful group. The supreme legislative body of the republic degenerates into an instrument in the hands of the powerful, and a partisan decision is masked as a decision taken by the whole people, or capable of expressing 'the view of the city as a whole'. While a king or an oligarchy must show how a decision made against an individual serves the common good, a democratic vote diminishes the urgency of such an explanation. It is not a restricted circle that designs the law. It is the assembly that represents, actually coincides with, the whole people that makes that call, and this by itself is taken as a sufficient justification. The problem with this condition is evident. No matter how far it is backed by popular support, a partisan decision that some citizens could not even in principle accept, no matter how well masked as a decision made 'by the whole city', is still illegitimate.
(b) The reading just offered as to why and how democracies rule against somebody also helps us to understand Kant's very cryptic point that democratic decisions generate a contradiction of the general will with itself (and with freedom). The people in the assembly, merely because they are the whole people, take themselves as nonappealable interpreters of the general will. And yet, since partisan decisions are possible even if everybody votes, what was supposed to be a decision taken from the perspective of the general will may very well be in contradiction with it. Finally, since any partisan decision will illegitimately impact the freedom of one or more individuals, freedom will also be 'contradicted'.
(c) Kant's concern that in a democracy 'everybody wants to be a ruler' (Alles da Herr sein will) adds a further and illuminating dimension to the problems identified so far. Kant cannot perceive an intrinsic flaw in each citizen's desire to be a ruler. What is wrong with my ambition to be a ruler, if I am ready to concede as legitimate the same ambition to all other citizens, as happens in a democratic system? The problem cannot be that each citizen is called to give his or her best interpretation of the general will because this is what must happen with every ruler, given Kant's standards, independently of whether they are one, few or all. 16 Kant evidently means an attitude opposed to the one displayed by Frederick II, that is, an attitude of interpreting political agency as 'service' in favour of the people, that is, of ruling from the perspective of the general will. What attitude can this be? Well, quite reasonably the one in which power is used not for advancing public but private interests. Recall that the official definition of despotism is as follows: 'despotism is that of the highhanded management of the state by laws the regent has himself given, inasmuch as he handles the public will as his private will' (ZeF,8: 352). 17 Quite significantly this definition comes right before Kant's general remark on the necessary despotism of democracy, thus signalling, albeit indirectly, that the furthering of a private will is the problem that inevitably affects democracy. What Kant seems to fear is that democratic citizens tend to use or, even more significantly, are entitled to use their share of political power to advance their partisan views or interests. They act as if they were rulers of a private company or of a household, and they use their vote to advance the interest of that private entity, not of the republic. This is the attitude diametrically opposed to the one attributed (rightly or wrongly) to Frederick II. While in a democracy I am allowed to represent my will or that of my group, or even the will of all citizens if they happen to agree (Rousseau's la volonté de tous 18 ), the only thing that a republican ruler is allowed to represent is the general will. 19 (d) The problem arises only for democracy because there is a fundamental difference with the other two forms of sovereignty, at least if we understand them as monarchies and oligarchies still governed by the rule of law, hence unstained by the purely arbitrary exercise of power that Kant identifies with barbarism: 'force without freedom and law' (Anth,7: 331). While in the case of non-barbaric monarchies or oligarchies the ruling class must explain how decisions are inspired by the general will, in a democracy there is no obligation of this sort. Citizens are allowed, if not expected, to represent themselves only. They do not need to justify their decisions as arising from an attention to the common good. By mistaking the sum of all private wills for the general will, or even more grossly the will of the majority for the general will, democratic citizens think that anything they choose is right. Actually, Kant thinks, the more people are in power, the less likely it is they will remember the burden of representationa burden that no political power, not even that 'of the people by the people' can escape (ZeF,8: 353).
(e) The problem arises necessarily for democracy in the sense that it is a defect generated by a structural feature of the system (again, the publicly accepted opportunity to use political power to advance private or partisan interests/views). It does not arise from a degeneration of the system; it is intrinsic to its logic. The defect is not necessary, however, in the sense that it is inevitable. As we shall see, Kant came to make room for a non-despotic form of democracy with citizens infused with sufficient public ethos to remember the 'burden of representation' when they participate in politics. And yet no moral antidote will completely remove the institutional stain associated with this form of sovereignty. In this weaker sense, democracies are necessarily despotic.
The points listed above constitute the backbone of my reading of Kant's criticism of democracy. But what is its significance? Many could harbour the impression that Kant's argument, no matter how newly interpreted, still applies to direct democracies only. After all, we assumed the Athens assembly as our point of reference. If this is true, then we would not have made much progress. As we said at the beginning, commentators usually deal with Kant's critique of democracy by proposing three main interpretative claims: (1) Kant's target was direct democracy; (2) representative democracy, especially if grounded on a constitution protecting basic rights and liberties, remains untouched; (3) representative democracy with constitutional guarantees is what best captures the ideal of the Kantian republic. While the first thesis is correct, the second is incorrect and the third needs qualification. The next section defends these theses.

The reach of Kant's criticism: the case of constitutional representative democracy
Let us recall that for Kant a system is republican only if rulers (one, some or all) govern from the perspective of the general will. The chances that this civic disposition halts the above identified structural flaw of democracies, the authorization provided to political actors to pursue partisan interests, are particularly slim if we are dealing with direct democracy. In this case, rulers literally represent themselves. In representative democracies I clearly cannot represent myself only. Delegates at the very least represent a group of citizens, no matter whether organized around an ideological view or a material interest (or both). This difference, however, is per se insignificant. As long as delegates represent or at the very least are expected to represent partisan interests, and they construe their consent in the electorate precisely around this partisanship, we are back to the problem Kant spotted.
Equally non-problematic is that the presence of delegates makes our democracies 'representative' in a sense quite different from the one meant by Kant. Recall that for him a system is 'representative' not when it has delegates and a parliament where they meet, but when rulers (no matter how many they are and no matter where, how and how often they meet) do not represent sectarian interests but make decisions having the general will as their sole guidance. 20 A bit less obviously, and to my knowledge never noticed, this remains true even in a 'representative' democracy where the constitution guarantees basic rights and liberties, and the majority is respectful of these limits. A constitutionally scrupulous political power may very well enact laws that protect the interests of some at the expense of those of others, for example by making them 'dependent', in the technical sense used by Kant to indicate an economic condition that de facto reduces their autonomy and as a consequence disqualifies them from active citizenship, or by introducing laws that while formally consistent with freedom, equality and independence of all citizens are patently in favour of only some of them. Imagine a decision to build a new road in a region where infrastructures already abound leaving the conditions unaltered in another equally populated region where roads are missing. 21 The argument articulated in the points above, which constitutes the essence of my reading, needs some qualifications. To begin with, I said that voicing a partisan interest is fully allowed by the rulers only in democracies, and this constitutes their peculiar defect because in the two other forms of sovereignty this is not allowed: governors cannot explicitly defend partisan interests. We also said that this is compatible with the fact, well established in history before and after Kant, that monarchies and oligarchies de facto serve partisan interests (those of the ruling family or group) more often and more effectively than democracy. When Kant claims that democracy 'necessarily' leads to despotism, we need to understand the point not as an empirical generalization, but as a way of pinpointing a structural defect of democracy.
This point is worth further clarification. To begin with, notice that the problem is not that in democracies majorities often make decisions against the will of some individuals (minorities). A law for Kant is in line with the basic value (external freedom) defended by right if consociates could have given their consent to it, quite independently of whether they did so. As Kant puts it 'my external (rightful) freedom is : : : to be defined as follows: it is the warrant to obey no other external laws than those to which I could have given my consent' (ZeF,8: 350;my emphasis). The problem with democracies is not with actual consent (or dissent) but with the fact that they tend to present their decisions as right in principle. The correctness of the procedure, especially if carried out strictly within constitutional limits, easily translates into the idea that whatever comes out of the procedure cannot possibly be unjust. A decision backed by a majority vote (or even by unanimity), even if compatible with individual guarantees, may still be incompatible with what some individuals could have consented to, as we shall see in a moment.
Secondly, it is important to clarify how the 'burden of representation' principle, that I indicate as the gist of Kant's view of good politics, relates to the principle of publicity (ZeF,8: 381). Both principles seem to have a similar function: dragging politics into the light of public scrutiny to rule out decisions that are evidently against the general will (or against some citizens). If the two principles fulfil the same or a similar function, what is the point of insisting on the 'burden of representation' as Kant's crucial requirement? Appearances aside, the two principles are not the same. The principle of publicity is a necessary, yet insufficient condition of the justice of a political decision. It merely rules out decisions that others, once the maxim on which they rest is made public, could not endorse. As Kant puts it, 'This principle is : : : only negative, that is, it serves only for cognizing by means of it what is not right' (8: 381-2). Moreover, the principle rules out policies largely a priori, because it only investigates whether the 'maxim' behind the policy stands if made public. However, some decisions pass the test of publicity, hence cannot be ruled out a priori, but can hardly be considered as arising from the general will. Imagine a policy proposed in good faith to reach a goal, but based on a demonstrably false empirical claim, e.g. lowering the taxes for the richest 1 per cent of the population as a way of helping the poorest through a general boost to the economy. Or, to give an even clearer example: think of Jair Bolsonaro's delay in adopting measures of social containment to combat the COVID-19 pandemic based on false assumptions regarding the virus' existence, diffusion, dangers and its possible alternative remedies (hydroxychloroquine). These policies are fully public and yet could hardly be construed as arising from the general will, unless the general will be construed as ignorant of basic and established scientific facts. The burden of representation principle is stronger than the principle of publicity in that it adds to the negative test of the latter the further requirement that decisions be backed by the knowledge and experience available. The general will is to be taken into consideration. To do so one cannot abstract from the empirical as the principle of publicity does. One must make up one's mind about the best way to further, in a specific socio-economic context and to the best of the knowledge available, hence by no means a priori, the common good.
Naturally it is much easier to understand what the general will rules out than what it supports. For the vast majority of political issues there is reasonable disagreement as to what decisions further the common good. We touch here on perhaps the weakest point of Kant's whole criticism of democracy. A system based on the free competition of particular interests, on votes freely cast to advance one's own good, at least rests on simple mechanisms that deliver decisions in conditions of structural uncertainty. 22 Moreover, this system does not seem to assume the highly idealized view of citizens and rulers who prioritize the common good over their private interests. This however should not lead us to the conclusion that Kant's emphasis on the necessity to rule from the general will's perspective is useless. This republican constraint rules out the argument that when I vote there is nothing wrong in prioritizing my interests or those of my group: 'I am opposing this law because it runs against the interests of my group and I owe loyalty to them, as opposed to the nation as a whole.' Kant's republican appeal to the general will, no matter how difficult it is to discern what it really wills, serves as a reminder that a corporate-like ideology in the political life of a community has inevitable despotic outcomes. Moreover, Kant's model is not committed to some sort of super epistemic power or to the unrealistic expectation that there should also be general agreement among citizens (at least in foro interno) as to what the general will dictates. Within a Kantian perspective, there is room for disagreement among competing yet reasonable interpretations of the general will, but a society where political issues are at least framed in terms of competing interpretations of what best furthers the common good is already in pretty good shape, and in any event closer to the republican ideal than a system in which partisanship is accepted as a, if not 'the', rule of the game.

Toward a republican democracy
The analysis offered in the preceding section should help us to see why Kant is so suspicious of democracy, but it should also help to conceive of a form of democratic regime that does not necessarily fall prey to the sectarian perversion described. If the spirit of republicanism has shaped the minds of citizens and rulers in such a way that citizens directly, or representatives indirectly, issue laws not only formally consistent with the constitution, but also capable of furthering the common goodif, in other words, the 'burden of representation' is accepted by electors and elected alikethen there is no reason why a democracy might not avoid despotism and meet republican standards. Indeed, if care for the common good is embedded in the polity, democracy has a clear advantage over the other two formae imperii. As Kant says, the latter 'are always defective' because power is exercised by one or few who by definition cannot represent all. In contrast, in a republican democracy citizens have the chance to check that those in power (the delegates) further the common good and accept the burden of representation. Incidentally, one should never forget that Kant praises popularwe would say, 'democratic'control over power. The criticism of democracy we are dealing with comes in the context of the first definitive article, whose main point is that republics are more peaceful than despotic regimes because republican governments, unlike despotic ones, are checked by the people in their decision to wage war.
Although Kant never explicitly mentions, let alone praises, a republican democracy, indirect evidence that he came to realize its possibility is to be found in the Metaphysics of Morals where he returns to the distinction among forms of sovereignty. As pointed out by Hanisch (2016: 70), in the Doctrine of Right we no longer find the idea that democracy is necessarily despotic. 23 Kant says that the 'united will of the people', that is, the head of the state or the sovereign, can stand in three different relations to the people depending on whether one, several or all rule within the state; thus, 'the form of the state will either be autocratic, aristocratic, or democratic' (MS,6: 338). This is the same classification of the forms of sovereignty we encountered in Toward Perpetual Peace. However, Kant does not say now that the democratic form necessarily degenerates into despotism. To be sure, he holds that autocracy is the simplest of the possible forms of state and, as far as the efficiency of the administration is concerned, is also the best. But he continues: 'With regard to right itself : : : this form of state is the most dangerous for a people, in view of how conducive it is to despotism' (6: 339). This seems the exact opposite of what Kant had told us in the 1795 essay. There, it will be recalled, he thought that 'the smaller the number of ruling persons in a state and the greater their power of representations, the more the constitution will approximate to its republican potentiality', with the consequence that monarchy is the form of state that has the greatest potentiality to 'reach this one and only perfectly lawful kind of constitution [the republican]' (ZeF,8: 353). Now the opposite is true. The fewer that are in power the higher is the risk of degeneration towards despotism. This obviously raises the question of why the assessment changes so dramatically and in such a short period of time, from 1795 to 1797.
5. Why did Kant change his mind about democracy from 1795 to 1797? Or did he?
It should be acknowledged that in recent literature Kant's change of attitude towards democracy has been explained differently. 24 Ludwig (1999: 180-1) and Byrd/Hruschka (2010: 179-81) claim that the difference in attitude in Toward Perpetual Peace and the Doctrine of Right is due to the fact that in the former work, as we saw above, the forms of sovereignty relate solely to the composition of the executive power, while in the latter work they are thought of as three different ways in which the whole state power (not only the executive) is exercised. We already saw that the idea from which this reading begins is dubious. By forms of sovereignty Kant did not mean in Toward Perpetual Peace, in fact never meant, the executive only. The change from Toward Perpetual Peace to Doctrine of Right is significant precisely because one form of sovereignty, understood as state power as a whole, appears to be considered as most conducive to despotism in 1795 and most in line with republicanism only two years later.
Why Kant portrays democracy so differently is not easy to discern from his writings. Having discarded the previous explanation, it seems that we are left with two main hypotheses. One is that, after highlighting the intrinsic danger of democracy, Kant came to realize (a) that there is an antidote for it and (b) that democracy also has a structural advantage compared to the other two forms that, by definition, presuppose that some make decisions for all. If there is a way to lead democratic citizens to resist the temptation to use their share of political power to advance a partisan agenda, then democracy is clearly better placed than the other two forms of sovereignty. For while in the other two forms of sovereignty there may be no explicit authorization to use political power for partisan goals, yet, power being in the hands of few or one, the temptation to abuse it is intrinsically high and one cannot even hope that, as in democracy, partisan agendas may cancel each other out in the decision-making process.
There is however a third hypothesis that incorporates the insights of the preceding explanation while denying that Toward Perpetual Peace and the Metaphysics of Morals tell two incompatible stories about democracies and that Kant changed his mind. The third explanation turns on the intuition that, after having focused on the dark side of democracy, Kant simply felt that the bright side also had to be acknowledged and a more complete picture needed to be drawn. Certainly, if there were a way of making the two accounts compatible with each other, one would have reached the non-insignificant hermeneutical gain of not being forced to attribute to Kant a complete reversal of opinion on such an important topic in the limited timespan of two years. This more complete picture rests on the combination of an empirical and a purely theoretical level of analysis. If autocracy and oligarchy have the advantage over democracy of ruling out as impermissible a partisan use of political power, they have the disadvantage of making an impermissible use of that sort more likely than it is in democracy. While democracy allows citizens to 'represent' their interests as opposed to that of the republic, the other two forms of sovereignty rule this out as a matter of principle, and yet construe objective conditions in which this may happen more easily and more frequently. As noted above, while in democracy there is at least some degree of cancelling out (or reciprocal mitigation) of opposing partisan agendas, if rulers in the other two forms of sovereignty lack the self-proclaimed correct disposition of Frederik II, then the republic is exposed to the risk of partisan ruling more than it is in a democracy. And not accidentally, as an historical point, it did happen, as I said, that oligarchies or monarchies have been comparatively more despotic than real democracies. On this third explanation, which seems to me by far the most plausible, Kant is not changing his mind, in a quasi-schizophrenic manner, about the respective merits of the forms of sovereignty. He is simply evaluating them first on the scale of the structural features that characterize them from a purely theoretical point of view (and here democracy is the worst) and then on the scale, altogether different, of the concrete risk of a degenerating potential that they have, taking into consideration ultimately the imperfections of human nature and the presence (or lack thereof) of institutional mechanisms to mitigate their consequences (and here democracy is the best).
Whether or not Kant truly changed his attitude toward democracy in 1797, one should never lose sight of the fact that what matters for himand he certainly never changes his mind about thisis the form of government (forma regiminis), not the form of sovereignty (forma imperii). Ultimately, it is not important how many people exercise power. What matters is: (1) that those who issue laws are clearly distinguished from those who implement (executive branch) and interpret them (the judiciary); 25 (2) that laws and decrees are issued in the right manner, that is, by constantly having in mind that the art of good ruling presupposes subordination of specific interest to the dictates of the general will. Only on this condition is the Rousseauian/Kantian criterion of legitimacy satisfied and no one is in the position to claim that political power is exercised against his or her will.

Conclusion
I have offered in this article a new reading of Kant's criticism of democracy in Toward Perpetual Peace. I attempted to show how this better fits Kant's text and makes better sense of the reasons why he thinks that only democracy has this despotic tendency and why it has it necessarily. If the reading defended is at least plausible, and if the extension of its reach from direct democracy to representative democracies with constitutional guarantees is equally plausible, Kant's analysis becomes extremely relevant for a time in which widespread fragmentation and polarization of the citizenry, fuelled by populist ideologies, have made our liberal democracies considerably distant from the republican ideal of a political system in which political power is exercised in such a way that nobody is wronged. What I have said already should convince readers that elections for Kant (or other forms of transmission of citizens' preferences to legislators) and constitutional guarantees do not suffice to qualify a regime as republican. If a polity is so divided that the promises of its republican constitution are constantly betrayed by the practice (if not by the letter) of power, that is, if a polity is so divided that the common good is rarely taken as the supreme principle guiding political actors, then ballots will only reveal who has the most popular partisan agenda and supreme courts will not suffice as a remedy against domination. Separation of powers itself seems to be a necessary yet insufficient device to avoid despotism, because those who hold legislative power may agree with those who hold executive power to enact laws inspired by partisanship. In that case, what looks like a republic, if assessed merely by looking at the formal structure of its institutions, will turn out to be a despotic regime or will rapidly degenerate into one. Ultimately, but this is evidently the topic for another article, the most important lesson we learn from Kant is that institutions and their design are important, but if a civic ethos that disables a purely strategic use of political power is not strong and widespread enough in the minds of political actors (be they one, few or all), universal suffrage and constitutional liberties will not guarantee that we are safe from despotism. 2 See e.g. Rousseau 1987: 206. 3 Think e.g. of Rawls' principle of liberal legitimacy: 'Our exercise of political power is fully proper only when it is exercised in accordance with a constitution the essentials of which all citizens as free and equal may reasonably be expected to endorse in the light of principles and ideals acceptable to their common human reason' (Rawls 1993: 137). Rawls refers to 'the essentials of a constitution', but Kant and Rousseau seem to have no restriction of this sort. They think we should 'obey no external law except those to which I have been able to give my own consent' (ZeF,8: 361). 4 Helga Varden has recently formulated the same question in the following manner: 'At the heart of Kant's legal-political philosophy lies a liberal, republican ideal of justice understood in terms of private independence (non-domination) and subjection to public laws securing freedom for all citizens as equals. Given this basic commitment of Kant's, it is puzzling to many that he does not consider democracy a minimal condition on a legitimate state.' (Varden 2016: 39). 5 Rousseau 1987: 179-80. 6 Jon Elster has described 'social choice theory' as something very similar to this view of politics (Elster 1997). 7 Obviously, Maus' main target was Schmitt's understanding of sovereignty as resting in the hands of those who can issue emergency laws. 8 Maus laments in particular creeping refeudalization in favour of the executive and of supreme courts. 9 Of course Maus and I differ starkly on other points. For example, she seems to understand Kant's notion of representative government in the contemporary sense of ruling via delegates, while I believe that Kant rather means a system in which those who rule shoulder the 'burden of representation', i.e. the duty of ruling from the perspective of the general will. More importantly, perhaps in line with the Habermasian idea of democracy and human rights as co-original, she takes Kant to be assigning no a priori limits to popular sovereignty, while I believe that for him any constitution is to be judged by adopting an external moral standard: its ability to respect the innate, that is pre-political, right to