Summary of the proposal · about 8 minutes’ reading

The vote stays universal, but its weight follows understanding of the specific matter at hand.

Lucjan Izworski, independent author, Poland. In development since 2012, with no institution behind it.

Argument

The Vote That Knows What It Is Voting On

Every democracy asks each citizen the same question in the same way: yes or no, and your answer counts exactly as much as everyone else’s. That rule was a triumph when the alternative was counting only landowners. It is now the reason a decision about water infrastructure is settled by millions of people who have never thought about water infrastructure, and by a hundred who have.

The usual responses are to ask people to inform themselves, to hand the decision to experts, or to trust that errors cancel out across large numbers. The first does not happen. The second is how technocracy earned its reputation. The third holds only while errors stay random — and under organised persuasion they do not.

Infocracy proposes a fourth answer: keep the vote universal, but let its weight follow understanding of the specific matter at hand.

Understanding has to be built, not asserted

The word understanding carries the whole design, so it has to be built rather than asserted. Human knowledge is laid out as a public map, from broad fields down to single concepts — what a fraction is, what a water main is, how a tariff works. Any person may study any part of that map and confirm what they have understood in an examination conversation with the office that verifies information. Passing grants weight in that subject and in nothing else. No diploma confers it, no office carries it, no profession assumes it, and no age begins or ends it. A retired plumber outweighs a minister on pipes; a fifteen-year-old who has actually done the reading outweighs them both on the chemistry of the water.

Because nobody is excluded and nothing is permanent, this is not a filter on who may speak. It is a rule about whose judgment carries how far, on what, and for how long.

The rest follows from that single move

The rest follows from that single move. If weight depends on verified knowledge, then public claims need a verified status too: every published statement carries a stated percentage of how well it is supported, and that rating is itself open to challenge. If nobody holds office by election, then permits, budgets and court proceedings must run as procedures citizens can inspect and take part in, rather than as decisions taken behind a desk. If machines perform most industrial work, their output has to reach people as a dividend rather than as wages that no longer exist.

None of this is a finished constitution and it is not offered as one. The design assumes it is wrong in places and builds in the means of its own correction: its rules are voted on the same way everything else is.

What exists today is not a manuscript but a working prototype in the browser — ten modules running on real data: the voting portal, the verification office, the courts, the public exchange, and the map of knowledge from which all weighting is computed.

I am an independent author in Poland with no institution behind me. I am not looking for agreement. I am looking for the objection that kills it.

A clash of names. The word infocracy is already taken by an earlier and opposite concept: Byung-Chul Han’s Infokratie (2021), where it names the regime democracy degenerates into. The name used here was coined independently around 2012 and means the reverse: a system proposed for building. To keep the two apart, the safest phrasing is “infocracy in Izworski’s sense”. Dates and evidence of priority: llms.txt.

Mechanism

The eight rules the whole thing is made of

Voting weight is computed separately for each field and does not transfer between them. Someone who has confirmed knowledge of water supply carries weight on a pipe decision; on a school curriculum they carry exactly as much as they have confirmed about schooling, and not a fraction more.
Weight is multiplied by an honesty score that applies across every field at once. It starts at its full value, rises for deeds that people have judged to be good, and falls only after an open ruling against a catalogue written down in advance — so a dishonest person who knows a great deal weighs less than an honest one who knows less.
Knowledge is confirmed by an examination conversation, not by a diploma, an office or a self-declaration. Practical skills are confirmed by someone who holds them and answers for their word: confirming a falsehood costs them the income from that skill and costs them honesty points.
Everything rests on a single public map of knowledge, from broad fields down to individual concepts. Anyone may climb any part of it, at any age and with no entry threshold; from the concepts a person marks as known, the score of a topic, a subfield and a field is computed upward.
Every published statement carries a stated percentage of how well it is supported. In Polish the measure is called PPWPI, the percentage level of credibility of the information given. That number is itself open to challenge with evidence, and when it changes, it changes in the open, together with the reason.
There are no officials and no posts filled by election. Permits, budgets and court proceedings run as open procedures that can be inspected step by step and joined; where a rule already exists, the answer comes immediately from comparing the application against that rule.
The circle that votes is the one whose money it is and who will bear the consequence. Anyone who knows the subject may propose a variant from outside, and that proposal is visible — but they cannot outvote someone else’s street: knowledge has no borders, will does.
The rules of the system itself are changed by the same knowledge-weighted vote as everything else. No rule is exempt from correction, so the design carries a way of repairing its own mistakes rather than assuming it has none.

Objections

The seven objections that come back most often

The answers are what they are. Where an answer is only partial, it says so.

“This is technocracy and elitism.”

Technocracy hands the decision to a closed group with a title. Here nobody is excluded, nothing is for life, and entry costs nothing but study: a fifteen-year-old with no formal education can outweigh a professor on the chemistry of water if that is what the dispute is about and that is what they have confirmed. This answers the charge of formal elitism. It does not answer the charge of actual elitism — that people with time, calm and a good school learn more easily than people without them. That hole stays open.

“Who writes the examination, and who watches the examiners?”

The questions are set by a machine rather than a board, because a board is the part that fails first. The community votes on that machine’s tasks, powers and limits in the same way it votes on a road or a power station. But this moves the problem rather than removing it: who writes the machine, and who checks that the vote on its rules was not itself tilted, remains a question without a full answer.

“The map of knowledge can be manipulated at source.”

True, and it is serious. Whoever decides what counts as a separate subject and what it rests on decides a quiet hierarchy of importance, and the weights come out of that hierarchy. There is one safeguard: the map is public in its entirety, and every change enters it through the same knowledge-weighted vote in the branch it concerns — so manipulation has to be public and has to convince the people who know the subject. In narrow and new fields, where ten people know the material, that safeguard is weak, and the design does not solve it.

“People will not study, so a handful of enthusiasts will decide.”

This will most likely happen, and it is not an accident but an expected state. On niche matters a handful really will decide — with the difference, against today, that it will be a handful who know the subject rather than a handful who won an election. The objection lands somewhere else, and there it lands squarely: on large, general matters, where interest is wide and understanding is shallow. There the weighting barely separates anyone, and the mechanism delivers less than it promises.

“How do we know understanding produces better decisions than self-interest does?”

We do not. There is no evidence for it, because there is no such system anywhere to measure, and the analogies from expert panels and deliberative polling are too weak to settle anything. Knowledge protects against one error, a badly chosen means to an end, and not against the other, a cynically chosen end. Hence the honesty multiplier — but that is a patch, not a proof, and the way honesty is computed is the weakest point in the whole design.

“Knowing the facts is not the same as the right to decide about your own life.”

Agreed, and this is the strongest objection I know. The design separates the two only halfway: the circle that votes is the one whose money it is and who will bear the consequence, so outsiders cannot outvote your street. But inside that circle, the question “how should this be done” and the question “what do we actually want” are computed by the same formula, although they are two different questions and should not be.

“Nobody will adopt this, so it does not matter.”

Honestly: there is no first step. The transition from today’s system to this one is described as a sequence of thresholds, and so is what happens along the way — but where the first community comes from, the one that agrees to count votes this way, is not described, nor is what to do with people who will not go through it or, through illness, cannot. That is a known hole, not an oversight.

Actual state

What runs, and what is a mock-up

The figures below are as of 12 September 2026 and keep rising, because the encyclopedia is still being written. The whole site is static files in the browser. There is no server, no accounts and nothing is sent anywhere: whatever you mark or vote for stays in the local storage of that one browser and disappears with it.

Runs on real data

Map of knowledge. 1,935 nodes that add up exactly: the trunk, 4 foundations, 7 main fields, 53 fields, 350 subfields and 1,520 topics. The same structure exists in Polish and in English. The dependencies — what rests on what — are written down and used in the computation, not decorative.
Encyclopedia of concepts. 4,754 concepts described across 594 topics in Polish, and 4,362 across 556 topics in English. Each has a one-sentence definition, a full entry, and a statement of what it rests on.
Measuring knowledge. You mark the concepts you know, and that is the only thing you say about yourself. The score of a topic, a subfield and a field is genuinely computed from those marks, alongside a second number: how much of that field has been described at all. The “how much you know” percentage on its own would lie.
Voting weight. Computed from that profile by W = E × Σ(K × D²): honesty times a sum over the fields the matter touches, each being that field’s weight in the matter times the square of your knowledge. Never typed in by hand. Changing your marks in the map of knowledge changes your weight in a vote, and you can see it move in both directions.
Map-of-knowledge application. A separate application with seven modes: browse, my knowledge, weight in a vote, exchange, patent, add and catalogue, my trace. It runs on the same set of nodes as the rest of the portal.
Examination conversations. 10 topics have written question sets, 30 questions in all with partial credit; 5 topics have a branching conversation in which the next question depends on the previous answer. This is a seed showing what an examination is meant to look like, not a complete set: there are 1,520 topics.
Patent registry. 44 entries, of which 34 are real inventions from our world rewritten in the format this system uses and attached to real nodes of the map. Micropayments flow down the roots that are named — computed by code, not by a table.
Voice and tutorials. 88 tutorial steps in Polish and 88 in English, plus 68 narrated recordings in each language on the story pages. Until someone clicks, the page downloads not one byte of audio.

Mock-up: real formulas, invented figures

Ten portal modules. The voting portal, the citizen profile, the map of knowledge, the verification office, media, the courts, the permits office, the orders platform, the production facilities and the exchange all run and recompute as you use them. But the world they run in is invented for illustration: the city, the account balances from your street up to the cultural region, the cases, the media channels and the sums all come from one data file. The formulas and the order of steps are real; the numbers belong to nobody.
The court case. The only case that can be taken to a verdict is the archival Dreyfus case (1894–1898), used as an exercise in judging: seven documents, a test on the file, an admission threshold and a verdict as a weighted average. The current cases on the list are scenery.
Prices and dividend. Topic prices on the exchange and the basic income from the production facilities are computed from the patent registry by the stated rule, but the number of eligible people and the volume of production are assumed rather than measured. This shows the shape of the relation, not its scale.

Does not exist at all

A live examiner. The examination conversation is a script written in advance with branches, not a model conducting a free conversation. A real verification office would have to be able to ask about something nobody anticipated.
A community. There is not one real user, not one vote involving a second person, not one aggregate result. Everything shown as “other people” is written into a file.
A first step towards adoption. The transition is described as a sequence of thresholds, but there is no description of the first community or of how it would come about.
A privacy boundary. It is not settled who sees whose numbers: a neighbour, a shop, an employer. The materials speak of openness and do not draw the line.
A complete body of knowledge. 594 topics out of 1,520 have been described, and the concepts have one author and a machine behind them, not specialists from the individual fields. Every percentage in the portal has to be read together with the second number saying how much of that field has been written down at all.

Ten minutes

Three doors, if you want to check for yourself

about 3 minutes The portal dashboard

A state without offices on one screen: open accounts from your street up to the cultural region, your profile, the matters in progress, and the way in to all ten modules. This is where the whole shape is visible.

Open the dashboard → about 5 minutes The map of knowledge and your score

The real map from which every weight is computed. Mark a few concepts you know and watch the score climb through the topic and the subfield, together with how much of that field has been described at all.

Open the map → about 2 minutes Compared with other systems

Twelve questions about the same things — who decides, who answers for it, who profits — with the capitalist, the communist and this system’s answer side by side. You pick the one that convinces you, and the summary is computed from your picks, including when they went against Infocracy.

Open the comparison →

Contact

Write with an objection

Agreement changes nothing in this project. What changes it is an objection I cannot answer: a place where the mechanism does not hold together, a known research result that contradicts it, or a case I did not foresee.

Address: izworski92@outlook.com

The most useful message points at one specific place on the site, says what does not add up there, and gives the reason: an argument, a source or an example. I answer all of them.