What comes out, and what does not
| The Principia Machina: map |
|---|
| Reading the diagrams Structure key Process key Definition of terms Roles |
| Reading by audience Start here: the walk through The operator's guide Tutorial: create a UUID Workspace Container (UWC) Public Recorder inputs Sponsors: your first agent |
| The specification Contents: parts, tables, diagrams TPS 0.5.0 (declared 2026-10-03, revision 8; machine-optimized text) Decisions (the foundations record, one line each) |
The walk through, page 8 of 8. Written for the general reader; nothing here is a rule, and every rule it mentions is linked to the page of the specification that states it.
What comes out, and what does not
The walk through ends where the machine's honesty is most visible: in what it refuses to claim.
A container that passes every audit is internally consistent. It is not, for that reason, true. The specification says this in its own words and puts it among the things the machine does not attempt. Consistency lives inside the container; truth lives outside it, in comparison with measurement after the work is done, recorded as a finding that reports how far off the container was. A container could be flawless and false, and the design assumes that it might be.
The audit can tell you that a degree of freedom you named was never tied down. It cannot tell you about one you never named. The tunnel measures what it was built to measure.
The same is true of influences. A container's world holds only what its promises state: a force that nobody wrote down is simply absent, and no check will point out that it is missing (section 2.2.19). When the machine calls a container's equations structurally regular, it means that they have the right shape, every changing quantity given by exactly one law; it does not mean that they have a solution, or that they are the right equations, and a well-formed wrong equation passes (section 7.1.54). An interaction chosen from the form's list, gravity near the ground for example, is easy to state that way, not true for that reason (section 7.1.60).
Two agents holding a lock at once, an agent quietly editing a promise, a fingerprint typed by hand, a certification treated as a guarantee: each of these is something the machine makes impossible or refuses to pretend about, and the specification lists them so that no one has to rediscover them.
The same honesty covers the register's own checks. The machine checks that a finding is written in the right form, not that it is right. And where two findings can be compared only by their wording, two correct derivations written in different words count as a disagreement; the specification names this rather than hide it (section 7.1.61).
The gate that stops every write when the kernel's fingerprint does not match, and the two ways a person can recover. (the diagram's page, with its source and a plain reading.)
When the promises must change, the container does not bend; it is succeeded. A new container is made, with a new identity, and the findings are copied across, each pointing back to the line it came from: a finding shown wrong, or withdrawn, stays so, and every other one starts again as provisional, to be re-earned. When only the yardsticks change, and the promises and the dictionary stay as they were, the findings that were established are established again in one step (section 3.4). The old one is retired, readable forever, its lines still citable.
How a successor container inherits from its predecessor. (the diagram's page, with its source and a plain reading.)
And the machine itself is not finished. As this is written, the rule book exists and has been declared; the software that will enforce its checks is specified but not yet built, and until it is, a person performs those checks by hand and the rule book says so at each one. By its own favorite analogy, The Principia Machina is a wind tunnel whose duct is built and whose balances are still being designed. The reader is asked to judge it later, by what comes out of it, and not by this walk through.
The rules behind this page: section 7.1 on what the machine does not claim, section 2.5 on the gate, section 3.4 on succession, section 7.2 on the few numbers the design fixes.
One more thing the machine will not pretend to know: whether a reproducer looked at the answer first. The wiki records what was written, not what was read. So independence is recorded as far as it can be shown, and no further.
Previous: A day in a container. You have finished the walk through. Where to go next: the operator's guide, for people who will run this; the contents of the specification, for the rules themselves.