NO TILE FOR IT: The Big4 Decision Apparatus, the Non-Self-Enriching Proposal, and the Timeline That Follows

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This paper records a probe and reads its result. The probe: a proposal was submitted to the frontier AI capital apparatus — the fourinstitutions that own the compute, the narrative, and the regulatory posture — offering a deterministic alternative (ST-T1024), requesting 5% of the savings it would produce, and directing that 5% to food banks. The consideration did not accrue to the proposer. The result is not that the proposal was refused. The result is that it could not be parsed. No routing tile existed for "consideration accrues elsewhere." The submission was not rejected; it was unreadable to the apparatus that received it. This finding is then extended. The Big4 are not optimizing for energy, water, hallucination, orliability. They are optimizing for four variables that a bounded architecture threatens directly:the capex narrative, the subscription model, refusal-as-revenue-killer, and liability laundering. The fifth variable — the option value of unboundedness — is the one that makes the other four coherent. A timeline is offered, not as prediction but as conditional staging: a sequence of observablemarkers, each of which, if it occurs, advances the sequence, and each of which, if it doesnot, extends the plateau. The paper draws no further conclusion.