# Scientific review guide

Read the PDF for the argument and the LaTeX source for exact mathematics. This guide points to existing manuscript sections; it adds no scientific results.

| Review question | Entry point |
|---|---|
| What does the Rayo connection actually supply? | Sections 1–2 |
| Does the old grammar include the incumbent's real capabilities, sharing, and installed modules? | Section 3; operational adequacy condition |
| Does an extension preserve the same meaning and contract? | Section 4 |
| Are a valid old lower bound and a complete new upper bound compared? | Section 5; productive boundary escape theorem |
| Are actual work and allocated budgets distinguished? | Section 6 |
| Which assumptions permit repeated savings, and which remain unproved? | Section 7 |
| What is the proposed bounded implementation loop? | Section 8 |
| What is preserved through architectural change? | Section 9 |
| Which baselines and causal ablations distinguish the proposal from prior approaches? | Section 10 |
| What finite optimization model, counterexamples, ledger, and dependencies are supplied? | Appendices A–E |

The key open question is whether class-wide obstruction certificates and useful extensions can be discovered cheaply enough, and transferred widely enough, to repay their complete cost in later improvement episodes.

The proposed multi-generation evaluation is a future protocol. No executed experiments, benchmark scores, autonomous system, or mechanized certificate artifacts accompany this release. JSONL records preserve supplied arguments and stated status; their successful parsing or reconstruction does not establish mathematical correctness.