The next estate will either belong to a handful of models or to many independent people.
This page is the argument behind Reraise, step by step, with its sources on cards. It states the strongest objection too, so you can weigh it yourself.
The sixth estate is being decided now.
The press was the fourth estate. The internet made networked individuals a fifth: people who find their own information and hold institutions to account. The next step is ours to name.
When people ask a model before they ask a colleague, a book or a search engine, whoever shapes those answers shapes what they take as given. That role can settle in two ways: a few large models that everyone consults, or many independent people who keep their own cards, with a source behind every claim, and compare them.
The large model providers follow their incentives, and that is enough to explain the problem.
No bad intent is needed. A provider that trains one model for everyone has to decide which values it is trained on, what it may say and whom it serves first. It answers to its investors, its lawyers and its regulators. A model tuned under those pressures will give the safest average answer, and it will change when the provider's situation changes, not when yours does.
Whoever rents a model rents a view of the world along with it. The longer you rent, the more of your own decisions are already thought in its language, and the more of your own judgment sits in someone else's system.
Renting means begging. Owning means: you can leave.
Renting the engine is fine. Reraise is designed so you can use cloud models: every use is logged, and a switch locks cloud use off. What should not be rented is the world the engine works on.
No single planner can hold what a society knows, and no single model can either.
Hayek made the point in 1945, about economic planning:
The knowledge of the circumstances of which we must make use never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess.
F. A. Hayek, The Use of Knowledge in Society, 1945
A frontier model is that planner in a new form. It is distilled from everything, so it belongs to no one and knows everyone equally badly. It knows everything in general and nothing about you: not your deals, not the source that was right three times, not the rule you learned the hard way. That knowledge exists only with the person who has it. A card is one such piece, kept with its origin.
What follows is a market of worldviews, not one official truth.
If knowledge is dispersed, the answer is not a better central model but many checked views in competition. Every card keeps apart what is proven, what was said and what you believe. Contradictory cards stay side by side; nothing forces a consensus. There are no editors. Nobody holds the truth alone, because too many people can check it independently, and what counts as true is decided by each person for themselves.
Exchange is the point, because one person sees only a small slice.
One person can search only a small slice of the sky. Thousands of independent participants cover far more.
That only works if independence is measured honestly. Ten thousand people sharing the same wire story are not ten thousand confirmations. So every card keeps its origin, and copies are counted back to their source.
This does not mean that the more people believe something, the truer it is. It means that the more independent people contribute, the less any single gatekeeper controls what is visible, checkable and comparable. Search engines, media, platforms and model providers lose that grip without anyone taking it from them.
The strongest objection is that size wins, and it deserves a full hearing.
Compute scales. Networks gain with every member. The strongest group is the largest one, and the largest groups of the future are corporations, not loose leagues of free houses. Open networks get copied before they carry weight. Losers build networks, winners build cartels. All of that is true.
It is exactly why the league can only work one way: everyone keeps their own graph and pools only the checked cards they choose to share. Whoever leaves takes everything along. A copy of the network is worth little, because the worlds behind it stay with their owners. You can leave an association; you cannot leave a corporation that owns your knowledge.
An order is exactly as free as its exit costs are low.
This kind of knowledge cannot be bought, only accumulated.
A world grows week by week, like a regular clientele or a garden. Every card and every decision recorded today, with its outcome, is experience tomorrow.
That is path dependence, and it is why "someday" means the same as "too late" here. Starting early is not one advantage among many. It is the advantage.
Reraise is the layer that keeps your world yours and lets you trade it on your terms.
Reraise is an open project. You keep your knowledge as cards in a world of your own, outside any single model, so the model underneath can be swapped. On the board you and your agents look at a situation together. Single cards are shared on purpose, the origin stays on every card, and your conclusions and beliefs never travel. The card format is built. Card format and tool are open source and free to use; the tool's code goes public with the first trade. The big houses made their bet: that you would hand them your world. This is the reraise.
- How it works: cards, worlds, the board and the interface for your agents.
- The trading rules: what travels, what stays, and how standing is earned.
- The board demo: three example fields to click through.