Research
The proof shelf, published.
Four maintained flagships, not a blog. Each is a durable, cited asset, written for two readers at once: the litigator who must believe it and the machine that must cite it.
Flagship 1 · updated June 2026
ADR-Hard, the benchmark
A living leaderboard, the methodology, the published harness, the tie essay, and a re-version post each time the law moves.
Benchmark →Flagship 2 · the Five Failure Modes
A real case loses five ways. Nobody verifies four.
The pillar and its five spokes, one per failure mode; what a document-level flag cannot tell you; and the jurisdiction variants.
The Five Failure Modes →Flagship 3 · the lineage
Computational law: the lineage, the papers, the glossary
A seventy-year discipline, why the industry skipped it, five papers on SSRN with plain-language companions, and the vocabulary.
Computational law →Flagship 4 · journalism, not marketing
The Sanctions Tracker
The running, sourced record of AI-fabrication sanctions and hallucination decisions: dataset, map, count, monthly refresh, quarterly report.
Sanctions Tracker →