The white paper source is in this directory. Community review is open.
Challenge a claim — Open a GitHub issue titled [§X.Y] Claim: <short description>. Describe the claim, your challenge, and evidence or reasoning. The author will respond.
Propose a factual correction — Missing citation, incorrect date, miscounted metric, broken reference. Open a PR with the fix. Maintainer reviews and merges.
Attempt replication — Clone the repository, run the Rx experiment (see root README), compare your results to the committed evidence. Open an issue if your results differ.
Dispute the methodology — Describe your objection in a GitHub issue mapped to the relevant section (§S3, §S10 in the supplement for experimental methodology). The author engages in the issue thread.
Merge substantive changes to the paper’s argument without author review. The paper’s claims are the author’s claims. Community PRs are for factual corrections only. The author retains editorial control.
| Section | Topic |
|---|---|
| §1 | Chomsky/Backus parallel; drift framing |
| §2 | Abstraction ladder |
| §3 | Context-sensitive gap; prosody analogy |
| §4.1 | GS mechanism; economic consequence |
| §4.2 | Paradigm taxonomy |
| §4.3 | Seven properties: Self-describing, Bounded, Verifiable, Defended, Auditable, Composable, Executable |
| §5 | Related work: Gordon (2024), Thirolf (2025) |
| §6 | Artifact grammar; ForgeCraft |
| §7 | Case studies; adversarial experiment (AX); replication (RX); practitioner study (DX) |
| §8 | Implications |
| §9 | Discussion; I(S) formula |
| §10 | Conclusion; community convergence theorem |
| Supplement §S1–§S14 | Experiment pre-registration, conditions, rubric, results, audit transcripts |
The supplement (§S1) documents the pre-registration commits. To verify that the rubric and design were committed before any experimental run:
git log --oneline --all -- experiments/ax/
The commit timestamps are the evidence. Clone this repository and inspect them directly.