The Ignition Watch
How the lab will decide that a structure has ignited, and later that it has become sovereign, written down before anything has been seen.
Nothing on the lattice has been tested for ignition yet. This page exists now for exactly that reason. A test written after a promising structure appears will be fitted to it, however honest the intent. So the lab writes its tests first: what it will check, how, and what result would count against the claim. The tests come from the kernel's four sovereignty conditions. None of them depends on the chemistry that is still to be built.
Two things to watch for
The lab watches for two separate moments, as set out on the metrics page.
| Moment | What it means | What must hold | What follows |
|---|---|---|---|
| Ignition | the loop closes; true of any attractor, base, coordination or sovereign | Condition 1 (recursion lock), with the structure resident in its basin | a running attractor; no metrics yet |
| I‑Pop | the structure becomes sovereign | all four conditions at once | the bootstrapping interval opens and the metrics apply |
A structure that is sovereign from the start passes both at the same tik. A base or coordination attractor passes the first and may never pass the second. The watch records both, with the tik of each, and records the gap between them when there is one.
First, name the structure
Every test below is run on a named structure, called X: a stated set of lattice cells and the atoms in them, recorded before the test begins. The kernel is explicit that this comes first, because moving the boundary moves the answer. A loop that fails a test as drawn may pass it once the operators that supply it are drawn inside X, and that is a different candidate, which needs its own four tests.
The four tests
1. Recursion lock
Does the loop keep closing with nothing outside it carrying a step?
List every operator the loop uses: each atom or structure that performs a conversion. Each one must be made or maintained inside X. The kernel's wording: where each conversion “is carried by an operator supplied from outside X and unaffected by X, the cycle is state conversion under external orchestration,” and the condition fails however long the cycle runs.
Test. Stop the lab from placing any operator. Passes if the loop continues on raw supply alone. Fails if it stops once placement stops. Passing this test, with the structure resident in its basin, is ignition.
2. Persistence under a knock
Knocked within tolerance, does it come back as the same structure?
The kernel requires at least one error-correction or basin-recovery capacity: deviations inside the tolerance are absorbed and identity survives.
Test. Disturb a stated number of cells of X at random, repeatedly, and watch for a stated number of tiks. Passes if X returns with its boundary unbroken. Fails if it does not return, or if the boundary breaks (that is rupture, and the structure is gone). The disturbance size and the tik window are fixed in the run plan before the run.
3. Boundary made by the structure itself
Could the inside rebuild its whole boundary from raw supply?
This is the kernel's reconstitution criterion. X produces its own boundary if it “retains the recursion to remake” the boundary’s parts from its own activity on raw, unstructured supply, were those parts to turn over completely. What it happens to import does not decide it; what it could remake does.
Test. Replace the supply with raw material only, with no ready-made boundary parts anywhere in the medium. Remove boundary parts in stages until all of them have been replaced. Passes if X rebuilds its boundary from the raw supply. Fails if it can only keep a boundary while ready-made parts are available. This test is rungs 8 and 9 of the ladder.
4. Maintenance at its own cost
Does staying together cost it something, all the time?
The kernel allows no regime in which a sovereign structure persists for free. The Modeled Economy already keeps a ledger in which nothing is created or lost, so the cost can be read directly.
Test. Read X’s upkeep from the ledger while it runs, then cut its feed. Passes if upkeep shows as a continuing debit and X declines when fed nothing. Fails if X persists unchanged on no intake, which means it is not paying for itself and something is holding it up.
What the four results mean together
The kernel reads the conditions in order, and each failure has a fixed meaning.
| Result | Reading |
|---|---|
| Fails test 1 | not a recursion at all: an attractlet or a non-recursive process |
| Passes 1, fails 2 | ignited; a base or coordination attractor, not sovereign |
| Passes 1 and 2, fails 3 | ignited; at most a coordination attractor. The kernel calls this the decisive failure |
| Fails 4 | not admissible as sovereign: either it has stopped, or it persists without paying, which the kernel forbids |
| Passes all four | I‑Pop. The first sovereign structure on this site |
Controls that must be run alongside
- Remove the supply. An attractlet relaxes to an inert state when its outside support is taken away. Every candidate is run once with the lab’s own supports withdrawn.
- Turn the time dial below u = 1. The substrate page proposes that a structure which persists only near u = 1 is being held up by the shared clock, and one that survives the asynchronous regime is producing its own order. That sweep has not been run yet; the watch adopts it as a control once it has been.
- Run a structure that cannot be sovereign through the same tests. The Modeled Economy food chain models no boundary at all, so it must fail test 3. If it passes, the test is wrong, and no candidate is scored until it is fixed.
What does not count as evidence
Rules for a run
- The run plan is written and dated before the run: the boundary of
X, the supply, the disturbance sizes, the tik windows. - Every run is reported, including the ones that fail. A failed test is a result.
- A claim of I‑Pop needs all four tests passed on the same named structure, and the three controls run beside it.
- Nothing a run shows changes the kernel. The lab is strictly downstream.