Sīdus · E1 RC4 · working research record
Several useful monitors.
One finite budget.
Scarcity, allocation, and total reliance under a shared rate constraint
The question
When several useful monitoring channels must share limited information capacity, what gets lost, what receives the budget, and which messages does the system actually use?
All four registered claim families confirmed. The archived decision reproduces from the final assembled data.
Working report: independent human review and experimental replication remain pending.
Supported here
Local scarcity, value-sensitive displacement, primary high-capacity total reliance and reference-v2 allocation advantage in the declared synthetic fixture.
Not established
Strict metarepresentation, biological cognition, architecture-general laws, a metarepresentational ecology, or independent experimental validation.
Open provenance
An unplanned distribution-verifier refusal and later restart. Current verification passes; the cause and remediation of that interruption remain unresolved.
Editable manuscript
Original RC4 statistical labels and frozen records preserved.
An evidence companion, not a simulated rerun
See where the budget goes.
Follow actual rate shares, inspect any of the 501 recorded seeds, replace a message conceptually, and compare the frozen allocation predictions. The formula lab distinguishes measured arithmetic from invented teaching examples.
Mean per-seed shares at C = 0.8. Gold: source 1 · blue: source 2 · rose: source 3. These are recorded endpoints, not an interpolated response curve.
Open the interactive E1 companion →What the registered analysis found
| Claim | Result | Qualification |
|---|---|---|
| T1 · scarcity | Base 0.423863; shift 0.354734 | Normalized composite losses; margin 0.05 in both. |
| T2 · displacement | Source 2 share +0.497778 | Paired base → shift; margin 0.05. |
| U · total reliance | All six cells confirmed | Primary at C = 4.5; margin 0.02. |
| V · allocation prediction | All eleven comparisons confirmed | Reference-v2 advantage; Holm control; margin 0.05. |
Each primary mean includes all 501 seeds. Invalid components contribute exactly zero. All 3,006 conditions completed, but not every scientific gate passed: T1 had 381 valid pairs in base and 429 in shift. Completion and validity are different claims.
The interruption is part of the record
On 22 October at 20:43 Vancouver time, a retained control record reports an environment-distribution verification refusal; a later record reports the driver restarting at 20:54. The author recalls an unplanned stop but does not know what happened.
The final numerical result reproduces, and the current environment verifier passes. Neither fact establishes what changed during the interruption or how it was restored. The report makes no claim of uninterrupted execution, zero historical drift or fully verified recovery.
Inspect the incident record →Evidence and reproducibility
- Retrospective numerical audit: 481 comparisons with no mismatches within stated tolerances. Source-aware reconstruction sharing SciPy with the original; not an independent experiment.
- Primary results · condition gates · assembled final data (8 MB) · archived adjudication.
- Frozen contract · protocol · selected evidence identities and display derivations.
- Canonical claim freeze · freeze manifest · independent replication charter.
- Version and provenance guide. The selected website evidence is not the complete execution archive. Private custody and sealed payloads are excluded; the full local audit is retained separately.
Verification boundary: local identity checks and arithmetic reproduction are supported. Fresh checkpoint-derived measurement reconstruction and reference-certificate regeneration were not performed in this audit. Earlier technical reviews were AI reviews. Commitment timestamps are self-recorded.
Publication status: working report; no peer-review, independent-replication or DOI claim. Final author/contribution, release and unresolved-provenance decisions remain open. R1 planning is not authorized for execution.