You have arrived at one instrument of the Objective Observer Initiative, on its observatory node. This experiment is closed: its record is sealed and the controls replay it.Start with the plain accountSee the map
The discipline. The quadrant-staircase reading computed THE GRADIENT from the Edge-Zone ladder’s own landed rows: the backward-heavy (taxed) share decays 60.2% → 25.3% → 15.6% → 9.3% → the 7.5% interior plateau by steps from the segment’s end, the tie share rising inversely, the two-step contraction ≈ 3.9 ≈ 4 — the operator’s own scale rule appearing in the decay. That reading was POST-HOC and marked as such; his word opened the door. This instrument registers THE GRADIENT’S SHAPE ON FRESH GROUND, and it is the programme’s first FULLY DETERMINISTIC knife: the observable is a sign census — no null ensembles, no draws, NO RNG ANYWHERE — a run is pure computation, exactly reproducible from its registration alone. THE GROUND: three waves × 256 fresh segments of ONE length (L = 32 — one length means no length-mix confound CAN arise, the sibling audit’s lesson made a design choice), past the program’s deepest cell (6,246,415) to the final cell 10,968,079; 768 windows per offset pooled, 17,664 deep windows. THE SHARED-WALK BUILDER, value-identity PROVEN in the selftest and re-proven on archived ground by HG1: one forward walk per group of 32 segments, one truncated backward walk per segment (the youth algebra) — minutes per wave, the lightest instrument in the family. THE KNIVES (pooled at wave 3; waves 1–2 RECORDED; every verdict an exact rational vs the rational α 0.01): HG2A/B/C THE MONOTONE DESCENT (offset 0 > 1 > 2 > 3, one-sided exact Fisher per step); HG3 THE TWO-STEP BAND — r₀/r₂ bounded in (2, 8), one factor of two either side of the conjectured 4, never the sample’s own edge, a PASS actively bounding the ratio; HG4 THE CONTRAST (r₀ > 4× the deep plateau; archived 8.0×); HG5 THE TIE MIRROR (the boundary suppresses the majority). ALL LANDED EITHER WAY. BEFORE ANY LOOK: no computation on fresh ground exists before registration. The labels: HG1–HG5 are experiment-local hypothesis tags of a computational experiment, ordinal labels only, nothing chemical.THE FIREWALL binds every panel: no claim on any Millennium Prize problem, ever; no claim about π’s normality; no claim over the prime numbers’ own laws; no claim over machine learning’s own results, quantum computing, physical cosmology, quantum mechanics, consciousness, or phenomenal experience, ever. Acts, never eyes. The kin pages: /dlx · /dpx · /dfx · /ovx · /csx · /djx · /adx · /eqx · /apx · /ssx · /msx · /blx · /fcx · /ezx.
The gate — honest acceptance
The observable, the protocol, the calibration witness (the value
identity aboard), and the ladder’s standing — derived from the engine’s own
acceptance, never asserted here.
The ladder — registrations, runs, and verdicts
One rung per WAVE (256 fresh L32 segments each; the census only —
no draws; minutes per wave). Wave 3 pools all 768 under the frozen knives and
refuses until waves 1 and 2 stand. The acts are the operator’s; the
engine refuses out-of-order acts. The fail-safe: if a run’s response
times out in the browser, the run continues on the server under its claim
file — re-read the ladder rather than re-running (a second run refuses
while the claim stands, and refuses forever once results exist). The D-sign
is read against each segment’s OWN vmax, exactly the Deep Join’s frozen
convention — per segment, never per group; the value-identity proof is
exactly this sensitivity, held.
the ladder loads on page open; the button re-reads
The external anchor
A wave’s records to the immutable EXTERNAL record (mldata CKAN when
configured; the dummy-node fallback otherwise) — the archive record is the
citable object, never this host.