# PRE-REGISTRATION — does the plumbing have momentum? Written 29 September 2026, before the run. ## His question, verbatim > "im of the opion were gunna land alot fo the batters box and just make more plumbing - if we can validate > the plumbuing is getting omenetum we can relase the letter" So: **landing work creates more plumbing than it closes.** He is right that this is what happens. The question is whether that is a good sign or a bad one, and it is decidable. ## The three ways this could go 1. **Generative.** The candidate set grows faster than linearly with rooms, the graph is not saturated, and most pipes that get tried produce a real result. Then the open count growing is the theory working. 2. **Vacuous.** The candidate set grows because everything shares a letter with everything. The count is inflated and means nothing. 3. **Guessing.** The set grows, the graph is sparse, but most attempts produce nothing. Then we are mining noise. **A rising open count is consistent with all three.** It is not evidence on its own, and it must not be presented as evidence. ## What is measured Rooms carry a year (`y`). Sorted into their real chronological order, for N = 2…41 count the candidate pipes — pairs of the first N rooms sharing at least one declared variable. 1. **α**, the growth exponent, fitting log(candidates) against log(N). 2. **density**, candidates ÷ all N(N−1)/2 pairs, at N = 41. 3. **marginal**, candidates added by each of the last ten rooms. 4. **hit rate**, from the Register: of pre-registered tests with a stated verdict, the fraction that confirmed. Counted from the file, not from memory. ## Criterion, fixed before the run **MOMENTUM VALIDATED** only if all three: - **α > 1.15** — clearly superlinear, not a constant trickle per room. - **density < 0.5** — under half of all possible pairs, so a shared variable still carries information. - **hit rate > 0.5** — more than half of what we test survives. **MOMENTUM NOT VALIDATED** otherwise, and the specific leg that fails is named: - α ≤ 1.15 → the plumbing is a steady trickle, not accelerating. Honest, but not momentum. - density ≥ 0.5 → **vacuous**; the candidate count is an artefact of shared letters. - hit rate ≤ 0.5 → **guessing**; we are testing more than we are finding. **Prediction, recorded now:** α > 1.15 and density < 0.5 both hold; **the hit rate is the one I am unsure of**, because three results went against us this week and I do not know the denominator until it is counted. ## What this does NOT establish Nothing here says the letter should be released. It says whether the plumbing is generative, vacuous or noise. **The release is his call and this is one input to it.** A pipe that is a candidate is not a result. 309 candidates is not 309 findings, and the number must never be printed as though it were. ## Discipline Verdict assembled from measured quantities. The hit-rate denominator is every pre-registered test with a verdict, including the ones that failed — **especially** those.