There are two formulas that repeatedly pull the conversation about democracy in opposite directions.
One is old, almost sacred: Vox populi, vox Dei – "the voice of the people is the voice of God." In this view, the people possess an immanent wisdom: however much individuals may err, society "in aggregate" is said to sense truth and justice. If that is so, the prescription seems obvious – simply give people more direct voice, and the system will begin to heal itself.
The second formula is almost the opposite of sacred, but no less widespread in contemporary explanations of politics: rational ignorance. The idea is not insulting and not contemptuous – it is a matter of cost logic. Understanding politics is expensive: it takes time, attention, nerve, and carries risk; and the personal influence of any single citizen is often minimal. For most people, therefore, "not knowing" and "not engaging" becomes a rational strategy. From here it is only one step to another conclusion: if that is so, then mass participation inevitably produces populism, emotional waves, and vulnerability to manipulation – and therefore quality filters are needed, and decisions should be delegated to "the competent."
The trouble is that both frameworks are too simple. They do not break against a rival ideology – they break against the physics of life.
Politics unfolds under conditions of scarcity: scarcity of time, attention, information, and energy. If we accept this as given – not as a moral verdict on people – the answer becomes a different one. Not "who is better," the people or the elite. Not "which faith is correct," direct democracy or meritocracy. But a very concrete question: what rules and channels do we build so that accountability functions under these constraints.
This essay is about two positions that habitually quarrel with each other – and about how they might be reconciled without producing either a utopia, a caste, or yet another machine for manipulation.
Most people cannot be experts in everything. This is neither a flaw nor an insult – it is the economics of life. A person has work, a family, health, a household, and their own risks to manage. The political process, meanwhile, produces vast quantities of documents, procedures, reports, laws, and budget lines. Even a highly motivated person is physically incapable of reading "everything" and monitoring "everyone" continuously.
To feel the scale, no conspiracy theory is needed – an ordinary day will suffice. A bill appears running to several hundred pages, with dozens of amendments buried across different sections. In parallel, a budget is released with thousands of line items, where consequential decisions are concealed in footnotes and classification codes. Formally everything is transparent; in practice, control belongs to whoever has the time and the staff to read it every day.
This is precisely why complexity so often functions as a weapon – not necessarily through malicious intent, sometimes through the sheer inertia of the system, but with the same effect either way. When the rules become opaque, control migrates to a narrow group. In the vacuum left by understandable channels of participation, those who can most easily mobilise and sustain focus prevail: organised interests, money, manipulators, bureaucracy. The dispersed majority loses – not because it is unintelligent, but because it is dispersed.
The key question, therefore, is a very concrete one: where exactly in the system do we place the mechanism of quality – and how do we protect it from capture.
One temptation is the quality filter. The initial intuition here is sober: if everyone can influence equally, the most organised or the most emotional may break the process. Therefore requirements are needed – competence, demonstrated merit, discipline, accountability. In different versions this is formulated differently, but the logic is the same: influence should be tied to a proven level of quality.
This approach has real advantages. It genuinely reduces the risk that emotional waves will translate directly into decisions. It raises discipline. It is more convenient for accountability – because it is easier to name the responsible persons, the criteria, and the sanctions.
But there is a trap that must not be glossed over. A filter created "for quality" very easily becomes a mechanism for self-perpetuation. The "competent" begin to select one another, evaluate one another, and justify one another. This is not necessarily malicious – it is the banal logic of any group that has gained access to the levers of power. Imagine that expert roles are introduced "to raise quality": methodological moderators, auditors, arbiters of appeals. At first this genuinely reduces chaos – standards and accountable persons appear. Then the rules for entry into these roles begin to be written by those who already hold them. And the boundary between "a quality standard" and "a closed club" becomes very thin indeed.
The other temptation is the romanticism of mass participation – the belief that "in aggregate" the people are always right. There is a strong truth in this: the system must remain "about us" rather than "about them"; participation cannot be severed without losing legitimacy; society has the right to a voice not only on election day.
But the moment we move from slogans into reality, a symmetrical trap appears. Without protections, mass participation easily becomes noise, pressure campaigns, fraud, manipulation, emotional waves, and influencer-driven mobilisations. Imagine the inverse scenario: two days before an important decision, a wave is launched across social networks in which one or two figures set the tone, followed by coordinated artificial amplification. In a system without protections, this looks like "the will of the people" – though in reality it is well-organised noise.
Strip away the flags, and it becomes clear that both sides perceive the same threat – only from different directions. Democracy breaks when people are cut off from influence under the pretext of "quality," and it also breaks when influence becomes cheap and defenceless against manipulation. It breaks, too, wherever complexity defeats attention.
The hybrid solution begins with a simple thought: participation and quality do not have to live in the same switch. They can be separated so that each performs its own function – and so that the weakness of one is compensated by the strength of the other.
The starting point is a universal right to a basic signal – not because everyone is an expert, but because mass experience and mass sense of injustice are also data. The question is not whether people are "worthy" to speak, but how to make that signal comparable and durable.
The next step is to fix a boundary: the public result must not be "a profile of a person." The public result must be an aggregated summary – generalised indicators, indices, and assessments of official roles measured in clear units such as office+period. And if we want this not to become a backdoor for reverse-engineering individual identities, we need minimum thresholds before any result is published, rounding and calibrated noise so that aggregates cannot be "unwound" to a specific person, and anti-fraud mechanisms to detect artificial amplification.
To make this concrete rather than abstract: instead of "people's assessments," what is published is a card for a specific role in a specific period. It shows how the aggregated level of trust has moved, where anomalies appeared, which thresholds were triggered, and which signals were filtered as fraud. If someone believes the data have been corrupted, they do not "shout into the void" – they enter an appeals procedure with clearly defined grounds.
After this, the quality filter can be returned to its natural place – not as the right to exclude, but as a set of roles and responsibilities with elevated accountability, access to which opens through contribution and demonstrated track record. There are things that must not be decided "by feel": auditing and fraud detection, handling appeals, governing methodological changes, managing anomalies, protecting against coordinated attacks and organised amplification. Here competence requirements are appropriate – not as an ideology, but as a form of accountability.
A separate word on artificial intelligence. It can genuinely change the game, but in the role of an assistant: reading large document sets, finding contradictions, tracing connections between decisions and their consequences, identifying patterns in budgets and contracts. It must not "rule" in place of people. Its function is to reduce the cost of understanding complexity – not to replace procedure.
And finally, the principal test of this whole construction is capture. Can an organised group, money, bureaucracy, or a media machine take control of it? If so, the symbiosis turns out to be a facade. This is precisely why checks and balances matter here, along with separation of roles, transparent rules for changing the rules, and real – not merely decorative – appeals processes.
One clarification, because it is easy to confuse: this is not about a social credit system for citizens. Not about personal political profiling or targeting. Not about a black box that decides on its own. It is about an aggregated public signal regarding official roles and institutions – and about the rules that make that signal useful, durable, and safe.
A model like this faces two honest questions: does it function in ordinary conditions, and does it hold under stress?
The simplest criterion is whether the system produces a stable signal, or whether every information surge destabilises it. Then there is fraud: how easily can results be artificially inflated, and how quickly is it detected. Then coordinated attacks: can a group buy or organise influence over a short period.
The field test is straightforward: does the system survive an information storm? Suppose that after a widely reported story, three times the usual volume of signals arrives within twelve hours – accompanied by a visible attempt at coordinated amplification. If the instrument either collapses under the load or publishes raw noise without thresholds or flags, that is a failure.
There are also less obvious points where things tend to break. The capture of the "role-based" part of the system: do the people responsible for methodology and appeals gradually become a closed club? Privacy: does a path emerge for working backwards from aggregates to specific individuals? The fairness of procedures: is there a real possibility of correcting errors, or do appeals exist only on paper?
If these tests are not passed, the construction is not ready. If they are – the debate between "people" and "merit" becomes less toxic, because a shared language of quality has appeared.
The first framework is useful because it teaches sobriety: mass participation without protections genuinely can produce chaos. The second is useful because it reminds us of legitimacy: without broad participation, any "quality" easily becomes the power of a narrow group. The mature position is to accept both truths and to make neither of them a faith.
The symbiosis then looks like this: a basic signal for everyone; aggregates instead of profiles; competence as accountability rather than as exclusion; artificial intelligence as a way of reading complexity rather than replacing procedure; and continuous testing against capture.
This is not a promise of a perfect system. It is a way of making abuses more costly, accountability more real, and civic participation less naive and more productive.