Tag: semantic-search

Population genetics has Mitochondrial Eve: the ancestor every living human's mitochondrial DNA traces back to. Fine print has an equivalent question. Legal text is almost never written fresh — it descends. This is a technical orientation on measuring that relatedness: content hashes for identity, winnowing fingerprints for inherited wording, on-prem SLM embeddings for kinship of meaning, and a dated derivation graph for descent — across 6,738 documents and 125,295 clauses. The search for the common ancestor turns up something real: the warranty disclaimer, carried by nearly every license measured.

A word count tells you a terms-of-service document got longer. It does not tell you what the document covers. This is a different measurement: a fixed list of 34 consumer issues, checked against real terms-of-service and privacy documents, with every clause scored three ways — how closely it matches the issue, how much text it spends on it, and how harshly it reads. It starts with three documents read closely and ends with an interactive leaderboard of sixty-four, ranked by what they actually address.