A research-grounded sweep of 20 data domains, each scored against the Hierarchical Bits law and ranked. The honest answer to “what is BH good for?”
APPLICABILITY_SCORECARD.md — the ranked
table + the findings + the recommendation.applicability_map.png — the visual map (composite score × verdict).SURVEY_SOURCES.md — the citations behind the scores.scorecard.py — recomputes the composite from the raw
scores (transparent weights) and regenerates the chart + report.The BH shape (a heavy shared substrate + many co-registered layers + selective
read) is everywhere — but in almost every big-data domain the
store-once + selective-read pattern is already mature SOTA (DICOM, COG/STAC,
lakeFS, CRAM/tabix, MAM, S-LoRA…), so BH lands as ANCHOR (credibility, not
novelty). The sweep suggests the still-under-explored contribution is
narrower and sharper than “a universal format”: treating rival /
conflicting interpretations as first-class entities (what existing tools treat
as noise to adjudicate) — exactly what bhanno models. Among the 20 domains
surveyed, the only one classified BUILD was CAD/BIM, where federation
duplicates and no tool unifies substrate-once + rival overlays + selective branch
reads. See BH_PRINCIPLE.md for the formalization the
sweep points to.
Run: python scorecard.py -> APPLICABILITY_SCORECARD.md + applicability_map.png