DemocraciaBR

Vision document · press material

Returning government to the people

DemocraciaBR does not want to swap who rules. It wants to take the power to govern away from the hereditary oligarchies and the bureaucratic distance that captured Brazilian politics — and return it to the people by raising the population's own capacity for self-rule. This document explains why that is possible, what political theory already knows about it, and how each part of the bet becomes architecture.

The problem isn't who manages — it's delegation

In the 1920s, Walter Lippmann argued that the public is incapable of governing and that the answer lay in specialized technocrats — the intellectual defense of modern bureaucracy.John Dewey replied, in The Public and Its Problems (1927), with the opposite thesis: the public's incompetence is not natural, it is produced by the lack of participation. Democracy is a way of life learned by practice; the more you delegate, the less capable the public becomes — a vicious circle broken only by more participation, not less.

That is the root of our bet. Bureaucracy and oligarchy are not fixed by swapping the managers; they are fixed by giving the people real, practiced power.

A whole lineage has already thought this through

The thesis has a tradition. Carole Pateman (1970) formalized theeducative function of participation: taking part produces more competent citizens, who in turn sustain more participation. Benjamin Barber (Strong Democracy, 1984) proposed concrete institutions against merely representative democracy. Hannah Arendt saw the way out in popular councils; Cornelius Castoriadisradicalized it in the concept of autonomy — the society that consciously institutes itself, against bureaucratic heteronomy.

The strongest empirical rebuttal comes from Elinor Ostrom (Governing the Commons, 1990; Nobel 2009): communities manage complex resources through polycentric self-governance, without state bureaucracy or privatization, when the rules are created and policed by those affected. In Brazil, Paulo Freire is unavoidable — conscientization as the precondition for the people to stop being an object and become a political subject — and the Porto Alegre Participatory Budget (1989) became the most studied case in the world of de-bureaucratization through participation. On the digital front, Hélène Landemore, David Van Reybrouck andJames Fishkin defend sortition and informed deliberation; platforms like Decidim (Barcelona) and the vTaiwan process (Audrey Tang) try to operationalize this at scale.

The elitist critique, taken seriously

The honest objection (Schumpeter, Lippmann himself, today Jason Brennan in Against Democracy) says: mass participation doesn't scale, the voter's rational ignorance is structural, and enlarged deliberation can be captured by whoever has the most time and rhetoric. The best way to fight it is to take it seriously — each objection has an institutional-design mechanism that neutralizes it. The literature no longer answers with faith in participation, but with engineering.

  1. "It doesn't scale" → polycentrism + representative sortition. You don't scale the assembly, you scale the system: each decision at the smallest scale where those affected can deliberate (subsidiarity), and the scales federate. When the national scale is unavoidable, you use sampling — a sortition-selected, representative minipublic deliberates on behalf of the population, as with the Irish Citizens' Assembly and the French Climate Convention, whose decisions survived referendums.
  2. "Rational ignorance" → change the calculus, not the voter. It is an argument about incentives, not about capacity. Real, local decision power makes information worth acquiring (whoever deliberates about their own street informs themselves). The cost of information is socialized (balanced briefings, paid time — the jury model). Rational ignorance is, in large part, an artifact of delegation itself.
  3. "Capture by the professionals" → sortition, rotation and speech design.Stratified sortition instead of self-selection removes the bias of those who "want to be seen" (Van Reybrouck: election and volunteering select those who seek power; sortition selects anyone). Short, non-renewable, rotating mandates don't attract careerists. Facilitation and speaking rules distribute time; Pol.is/vTaiwan attacks rhetoric computationally — shouting louder does not move the consensus map.

Two points of honesty that strengthen the position. First: the epistocrat has the same problem, worse — any competence test for voting is defined by someone, and whoever defines it captures the system (Estlund's objection). The adverse selection ascribed to bureaucracy applies in full to epistocracy. Second: none of this works as a standalone part — an assembly without power becomes legitimizing theater; participatory budgeting without funds withers, as in Porto Alegre itself after 2004. The critique is right about badly designed participation and has no data against well-designed participation — which demands continuous institutional commitment, a political choice, not a technical problem.

From institutional design to technical design

The tool's design must derive from the institutional design, never the other way around. Each mechanism above maps onto a component of the architecture:

  1. Stratified sortition → unique verifiable identity, without doxxing. Sortition only works with uniqueness (anti-Sybil) and stratification by region/age/income without exposing who is who — including the undocumented population. The piece on top is the zero-knowledge attribute proof ("I reside in municipality X, bracket Y") and apublicly verifiable draw (VRF/beacon with commit-reveal): whoever controls the RNG controls the assembly.
  2. Polycentrism → federation, not a single platform. One instance per scope (neighborhood, municipality, state), federated by protocol (ActivityPub). It solves scale, sovereignty and resilience at once — capturing the system requires capturing every node. A centralized participatory-democracy platform is a performative contradiction and a single political point of failure.
  3. Anti-rhetoric deliberation → opinion clustering, not threads. Threads reward volume and rhetoric. The Pol.is/vTaiwan design inverts it: short statements, voting (agree/disagree/pass), embeddings mapping clusters, and an interface highlighting consensus across opposing groups. The clustering algorithm itself becomes a point of power — it must be open, deterministic and re-runnable by anyone from the public data, otherwise you merely swap the bureaucrat for the model.
  4. Real power → coupling with mandate and consequence. Deliberation's output cannot be a PDF; it has to be a structured, traceable commitment — approved proposal → mandate commitment → follow-up → scorecard. Deliberation without the consequence subsystem is half a product (the ill Decidim suffers from).
  5. Balanced information → adversarial, versioned curation. Each question gets a briefing with designated advocates for each side, cited sources and an auditable public history. Signed, reproducible summarization; AI helps to summarize, never to decide what gets in.
  6. Anti-careerism → rotation in the protocol. The "member of assembly X" credential expires, and the same identifier becomes ineligible for the next pool — with the caveat that a rule too rigid is hard to amend democratically: the meta-process of changing the rules must also exist.
  7. Everything auditable → immutable ledger and sovereign infrastructure. Each vote, draw and briefing edit as a public event (anonymized where needed), running on cooperatively-run national infrastructure — the physical substrate of democratic deliberation is a matter of sovereignty.

What the tool does NOT solve

Three limits best admitted up front:

  • Digital exclusion. In Brazil, any purely digital system stratifies wrongly by definition. The design needs points of assisted in-person participation (CRAS, libraries, schools) as first-class citizens of the protocol, not a hack.
  • Privacy × verifiability. There is no perfect solution, only trade-offs — better to document them publicly than to pretend cryptography solves everything.
  • Software does not create political commitment. Technology can make capture expensive and participation cheap, but transferring real power remains a political conquest. What the tool does is remove the technical excuse that "it doesn't scale."

Next step

Specify the verifiable sortition subsystem as an RFC — the piece no existing civic-tech project (Decidim, Consul, Pol.is) has, and the one that answers the elitist critique directly. That is where the ambition stops being rhetoric and becomes engineering.

Vision document — DemocraciaBR press material. The capabilities described include both what is already live and the horizon under development; the current state of each part is in thepress kit.