Dammit. Again, I need to admonish myself for reading. This time, Cioran is to blame, and now I’ve spent half the night writing this – not to mention half of another on the Sorcerer's Apprentice now waiting in the wings. I’ve been reading his History and Utopia, which is the sort of book one reads slowly, partly for the prose and partly because reading it quickly would be bad for the nerves. Cioran’s suspicion isn’t that people occasionally tolerate tyrants. It’s more that they want them, and that the wanting is not an aberration of the species but one of its more reliable outputs.
Full disclosure: I don’t support Democracy or the so-called and self-proclaimed Age of Enlightenment for countless reasons. Feel free to take this into account whilst reading. See my Anti-Enlightenment corpus for more details.
The liberal account of tyranny is more flattering, which is presumably why it has done so well. In that account the tyrant arrives from outside: institutions decay, civic virtue thins, a demagogue slips through a gap in the fence. Democracy is the victim of a burglary. Everyone gets to be sad about it at conferences.
Plato is less accommodating. In Republic VIII–IX, tyranny is not democracy’s burglar but its offspring. The democratic subject wants freedom without limit; desires proliferate and no longer submit to ranking; hierarchy comes to look like an insult; and the resulting disorder produces an appetite for somebody who will impose coherence. The tyrant doesn’t defeat the popular will. S/he’s one of the things it produces when it gets what it asked for.
The prosthesis
What interests me is the mechanism, which isn’t really submission at all. People needn’t desire to be ruled instead of ruling. They can desire a ruler experienced as an extension of their own ruling:
He will do what I would do, if only I had the power.
This single sentence does an extraordinary amount of work. It converts surrender into agency. I hand over power to X, but because X is me, enlarged, I experience his exercise of it as my own. The sovereign becomes a prosthesis of the citizen. The strongman is a wearable device.
This is the manoeuvre I traced in The Will to Be Ruled: submission can be grammatically recoded as self-determination, so that “I choose to obey” reads as an exercise of autonomy rather than a contradiction of it. Democracy provides the most elegant possible disguise for it, because the tyrant can claim an electoral genealogy. You chose me; therefore my coercion is your sovereignty. It’s the same double grammar I set out in the work on democratic coercion – procedural validity doing the authorising whilst a borrowed moral vocabulary does the flattering – only now compressed into a single body, which is far more efficient and photographs better.
The dimensionality problem
Here is where the fantasy meets arithmetic. There’s no single political self available to be proxied. Not even one of you is one of you. An individual holds competing preferences that are unstable over a weekend, let alone a parliament. Now multiply that by millions of people across taxation, migration, religion, reproduction, war, property, sexuality, welfare, speech, policing, environment, class, region, foreign policy and an unbounded set of symbolic loyalties that nobody will admit to holding but everybody votes on.
Before anyone can embody ‘the people’, the people must first be squashed into something small enough to embody. This is the argument of Rational Ghosts: aggregation has a ceiling, and it’s not a ceiling we can raise with better civics teaching. Arrow wasn’t describing a bug in mass democracy under adverse conditions. He was describing what happens under ideal ones. Heterogeneous, inconsistent, incommensurable preferences don’t resolve into an object deserving of a singular noun.
The tyrant doesn’t solve this problem. S/he performs a more violent version of the same compression, and hides the seam.
Instead of:
millions of preferences → electoral aggregation → policy
you get:
millions of preferences → charismatic identification → ruler
The second feels cleaner, because the impossible step has vanished from view. One bloke now incarnates the contradiction. Humanity has discovered a magnificently economical solution to multidimensional preference aggregation: put a face on it.
The bundled licence
S/he can’t proxy everybody. This is unremarkable. What’s more interesting is that s/he can’t reliably proxy even his own supporters.
A voter backs them over immigration and discovers they also govern abortion, tax, schooling, policing, the courts and the disposition of the armed forces. Political allegiance is a bundled software licence nobody has read. One salient preference purchases an entire ontology, and there’s no partial refund. You wanted the border; you also bought the theology.
So every tyrant must eventually betray their constituency, and – this is the part I want to insist on – they needn’t change a single position in order to do it. The betrayal is not a corruption narrative because it’s baked into the initial identification, which was underdetermined from the start.
One supporter sees a defender of tradition. Another sees a destroyer of elites. Another, an economic nationalist. Another, a religious restorationist. Another, more honestly, sees some geezer who will hurt the people they despise.
The successful candidate functions as a semantically underspecified political signifier – a Contestable wearing an empty suit. Their usefulness lies precisely in remaining unresolved. Everyone can see enough of themselves in them, which is only possible whilst there isn’t very much to see anyway.
Then s/he governs. Ambiguity has to become decisions, and decisions collapse possibility like a wave function. Every act of rule is an act of disambiguation, and disambiguation is where the projections start failing one by one, in a queue.
Not rankings – inventories
There is a deeper version of this failure, and it’s the one I have been working on under the heading of ontological non-alignment. Political conflict is usually modelled as different rankings within a shared space: we all want the same sorts of things and merely order them differently, so a sufficiently clever mechanism could split the difference. But people are frequently not disagreeing about ranking. They’re working with different inventories of what exists – which entities have standing, which injuries count as injuries, which authorities are authorities at all. Evidence denominated in one ontology doesn’t convert into another, and no amount of it accumulates into agreement.1
No sovereign can instantiate incompatible inventories simultaneously. He can only rule one of them and render the others institutionally unintelligible, which is not a failure of the tyrant but a description of the office.
Tyranny persists whilst tyrants fail
Hence the cycle, which has the loveliness of a well-made machine and the ugliness of what it produces:
plurality → frustration → demand for coherence → proxy sovereign → forced simplification → exclusion → resentment → renewed demand for a different coherence.
Note what survives each turn: The tyrant doesn’t, but tyranny does.
This suggests the usual diagnosis is slightly off. It may be misleading to say people have a propensity for tyrants specifically, but the propensity is rather for outsourced resolution. The tyrant is merely one extraordinarily efficient technology for supplying it – and, being mortal, disappointing and available for blame, arguably a rather good one.
The competitors do the same job with less scandal: Monarchy, party, church, nation, revolutionary vanguard, technocracy, judiciary, market and algorithm all take unbearable plural complexity and convert it into an authoritative decision-point. We don’t currently call the market a tyrant, because it lacks a face, which is precisely what makes it the most durable of the lot. The algorithm is now competing hard on the same advantage.
Cioran’s pessimism therefore looks more structural than psychological. The claim isn’t that humans secretly relish domination, though the historical record has never had to look far. It’s more that plural social existence continuously generates irresolution, whilst humans and their institutions continuously seek machinery that makes irresolution disappear.
Democracy’s distinctive promise is to do this whilst preserving everybody’s sovereignty. That’s its most extravagant fantasy, and it’s not a fantasy about tyranny. It’s a fantasy about representation – the hope that compression can be lossless if the procedure is fair enough.
So Plato’s warning can be radicalised. Tyranny isn’t necessarily democracy’s opposite. It may be one of democracy’s recurring dreams about itself: the moment at which representation gives up on approximating identity and simply declares it.
Because the question being asked at the ballot box isn’t the modest one, who will rule us?
It’s the impossible one:
Who will rule exactly as I would rule, whilst remaining the ruler of everyone else?
Political history consists largely of expensive demonstrations that there’s no answer. We keep asking. Cioran’s contribution is to suggest we ask because there is no answer, and that a mechanism which finally admitted as much would have nothing left to sell us.
This content was produced by Bry Willis, a philosopher with a focus on language. He shares content here on Substack as well as on his Philosophics Blog. Bry uses various LLMs in his production workflow at all stages, from ideation to production, as well as for the production of ancillary content – audio and video summaries, images, PDFs, and possibly other media.
You can also find him on various social media platforms:
For more information, check out my About Me page
A colleague shared a comment on Mastodon, but I only get allocated fewer than 500 characters, so I respond here and hope for the best: https://kolektiva.social/@homohortus/117112659771316059
It starts with this: I find your argument illuminating, but I would restrict it to bourgeois representative democracy rather than democracy as such.
RESPONSE
This objection is useful to a degree, but I think we're talking at different levels. I'm not arguing that bourgeois representative democracy is uniquely defective, nor that direct democracy, federalism, assemblies, councils, or mutualist arrangements cannot improve particular outcomes. They very well may. My narrower claim is that any democracy involving more than one person with divergent preferences must sometimes – arguably most of the time – produce outcomes that are suboptimal for at least some participants.
This problem arises before capitalism, representation, spectacle, class domination or mass society enter the picture. Two people deciding where to eat already encounter it. If one wants burgers and the other Indian food, some cost must be absorbed: one yields, they expend additional resources satisfying both preferences, they alternate, or they choose a compromise neither preferred. None of this is catastrophic. Much of ordinary social life consists precisely in accepting such small losses, so we can pass it off if it involves a partner or friends, but this is not most social situation.
Kenneth Arrow formalises a related problem under almost absurdly favourable conditions. Give agents coherent rankings over a shared set of alternatives, assume semantic agreement about what those alternatives even are, remove misinformation, ideology and ontological disagreement, and collective aggregation can still fail to produce a social ordering satisfying several seemingly modest conditions. The difficulty therefore can't be attributed wholly to bourgeois institutions. Those may intensify it, distort it or distribute its costs unjustly, but they don't create it.
Dunbar then adds a different pressure. At small scales – 150 to 5,000 people/nodes – repeated interaction, reputation, trust and reciprocal bargaining can soften preference conflict. My own experience of households and neighbourhoods is banal evidence of this: people concede things they care little about in return for things they value more. But these mechanisms decay as strangers replace intimates and administrative proxies replace relational knowledge. They certainly don't work on orders of magnitude in the hundreds of thousands or millions.
Then there's the still harder case: ontological non-alignment. Once people differ not merely over preferred outcomes but over what constitutes harm, standing, authority or even the legitimate objects of political concern, the bargaining space itself becomes unstable. At that point, reducing the assembly from 100 million to 1,000 may improve accountability enormously, but it doesn't make incompatible commitments commensurable.
So I would distinguish these questions:
Can some democratic arrangements be less coercive, more participatory or better at containing pluralism than others? Certainly.
Can democracy eliminate the loss generated by aggregating divergent preferences into shared decisions? I see no reason to think so.
And this is where Hobbes enters the room. A society may rationally accept considerable local preference loss because the alternative carries other costs: insecurity, violence, instability, transaction burdens, defence, coordination failure. In that sense, the democratic arrangement could have a positive overall cost-benefit calculus.
But there's still a snag: the costs are not naturally commensurable. How many unwanted dinners equal one avoided civil war? How much political frustration equals physical security? How many infringed preferences equal stable infrastructure? There is no neutral unit in which the ledger balances.
Cheers.