"""BH applicability scorecard — synthesizes the research sweep into a ranked map. Data: the 20-domain research sweep (one web-grounded agent per domain; see SURVEY_SOURCES.md for the citations). Scores are 0-3 per criterion. The composite is computed here with DECLARED weights so it is transparent and reproducible. composite = (0.30*substrate_dominance + 0.20*multiplicity + 0.20*selective_read + 0.15*structural_access + 0.15*adversarial) / 3 * 100 (0-100) Verdict (from the research, captures novelty vs already-SOTA): BUILD = strong fit AND the union is not yet a mainstream tool ANCHOR = strong BH shape BUT store-once + selective-read already mature SOTA DELEGATE = dense signal — BH should call a specialist codec, not compete NO = no real fit (substrate does not dominate / payload is the point) Run: X:/miniconda3/python.exe X:/bitH/applicability/scorecard.py """ from __future__ import annotations from pathlib import Path import matplotlib matplotlib.use("Agg") import matplotlib.pyplot as plt OUT = Path(__file__).resolve().parent W = {"sd": 0.30, "mult": 0.20, "sr": 0.20, "sa": 0.15, "adv": 0.15} # area, substrate_dominance, multiplicity, adversarial, selective_read, # structural_access, verdict, built, scale, note ROWS = [ ("Knowledge graphs (multi-ontology)", 3, 3, 3, 3, 3, "ANCHOR", False, "16B triples", "named graphs + RDF-star already SOTA"), ("Medical imaging (multi-reader)", 3, 3, 3, 3, 2, "ANCHOR", False, "WSI 1-100 GB", "DICOM SEG references source — mature"), ("Earth obs / satellite", 3, 3, 3, 3, 2, "ANCHOR", False, "10+ PB", "COG + STAC already do it"), ("Legal eDiscovery", 3, 3, 3, 3, 2, "ANCHOR", False, "TB / matter", "Relativity store-once + coding fields"), ("Video archives / MAM", 3, 3, 3, 3, 2, "ANCHOR", False, "PB masters", "MAM timecoded tracks — mature"), ("Dataset versioning / data lakes", 3, 3, 3, 3, 2, "ANCHOR", False, "PB", "lakeFS/DVC zero-copy IS the win"), ("Data labeling / annotation", 3, 3, 3, 2, 3, "ANCHOR", True, "COCO 20 GB", "CVAT/Label Studio store-once — mature"), ("Genomics (variant calling)", 3, 2, 3, 3, 3, "ANCHOR", False, "ref 3 Gbp", "CRAM + tabix selective — mature"), ("Geospatial map tiles", 3, 3, 2, 3, 2, "ANCHOR", False, "PB", "COG/PMTiles/vector tiles — mature"), ("CAD / BIM", 2, 3, 3, 3, 3, "BUILD", False, "tens of GB", "federation duplicates; rival overlays UNSERVED"), ("Model checkpoints / MoE", 3, 3, 1, 3, 2, "ANCHOR", True, "base GB-TB", "S-LoRA / Punica already SOTA"), ("Agent memory", 1, 3, 3, 3, 2, "ANCHOR", True, "KB-MB text", "substrate LIGHT; Mem0/Zep mature"), ("Scientific sim / HPC ensembles", 1, 3, 3, 3, 2, "DELEGATE", False, "CMIP6 30 PB", "per-member dense -> Zarr/zfp"), ("Autonomous driving datasets", 3, 3, 3, 3, 2, "DELEGATE", False, "Waymo 2 TB", "sensor signal -> codecs; sidecars mature"), ("Time-series / IoT", 2, 2, 2, 3, 1, "DELEGATE", False, "1-50 TB/day", "Gorilla codec owns the signal"), ("RLHF / preference data", 1, 3, 3, 1, 2, "NO", False, "MB-GB text", "substrate = label order; payload IS the point"), ("Distributed tracing", 1, 2, 1, 3, 2, "ANCHOR", True, "TB/day", "many distinct traces; Tempo mature"), ("3D scenes / glTF (LOD)", 1, 1, 1, 2, 2, "DELEGATE", False, "MB-TB", "geometry -> Draco/Nanite; LOD = dup copies"), ("Vector DBs / embeddings *", 0, 2, 2, 1, 0, "DELEGATE", False, "TB", "control: vectors ARE payload; PQ/HNSW own it"), ("Image / audio codecs *", 0, 1, 1, 2, 0, "DELEGATE", False, "exabytes", "control: signal IS payload; codecs irreducible"), ] VCOLOR = {"BUILD": "#2e7d32", "ANCHOR": "#1565c0", "DELEGATE": "#ef6c00", "NO": "#c62828"} def composite(sd, mult, adv, sr, sa) -> float: return (W["sd"] * sd + W["mult"] * mult + W["sr"] * sr + W["sa"] * sa + W["adv"] * adv) / 3 * 100 def main() -> None: scored = [] for (area, sd, mult, adv, sr, sa, verdict, built, scale, note) in ROWS: scored.append({ "area": area, "c": round(composite(sd, mult, adv, sr, sa), 1), "verdict": verdict, "built": built, "scale": scale, "note": note, "sd": sd, "mult": mult, "adv": adv, "sr": sr, "sa": sa, }) scored.sort(key=lambda r: r["c"], reverse=True) # ---- chart ---- fig, ax = plt.subplots(figsize=(11, 8)) ys = range(len(scored)) labels = [f"{'★ ' if r['built'] else ''}{r['area']}" for r in scored] ax.barh(list(ys), [r["c"] for r in scored], color=[VCOLOR[r["verdict"]] for r in scored], height=.7) for y, r in zip(ys, scored): ax.text(r["c"] + 0.8, y, f"{r['c']:.0f} · {r['verdict']}", va="center", fontsize=8.5, fontweight="bold", color=VCOLOR[r["verdict"]]) ax.set_yticks(list(ys)); ax.set_yticklabels(labels, fontsize=9) ax.invert_yaxis(); ax.set_xlim(0, 118) ax.set_xlabel("BH-shape composite score (0-100) · weighted: substrate-dominance .30, " "multiplicity .20, selective-read .20, structural .15, adversarial .15", fontsize=8.5) ax.set_title("BH applicability sweep — 20 domains, research-grounded\n" "composite = how BH-shaped · color = verdict (novelty vs already-SOTA)", fontweight="bold", fontsize=12, pad=12) handles = [plt.Rectangle((0, 0), 1, 1, color=c) for c in VCOLOR.values()] ax.legend(handles, [f"{k} — {('greenfield' if k=='BUILD' else 'already SOTA' if k=='ANCHOR' else 'call a codec' if k=='DELEGATE' else 'no fit')}" for k in VCOLOR], loc="lower right", fontsize=8.5, framealpha=.95) ax.text(0.5, -0.085, "★ = prototype already built in this repo · * = control domain (should sit low)", transform=ax.transAxes, ha="center", fontsize=8, color="#555") for s in ("top", "right"): ax.spines[s].set_visible(False) fig.tight_layout() fig.savefig(OUT / "applicability_map.png", dpi=130, bbox_inches="tight") plt.close(fig) # ---- report ---- L = ["# BH Applicability Scorecard — research-grounded sweep of 20 domains\n"] L.append("> Method: one web-grounded research agent per domain gathered real data " "volumes, the tools used today, and whether interpretations conflict; each " "scored 0-3 on five criteria. The **composite** below is computed from those " "scores with declared weights (transparent, reproducible in `scorecard.py`). " "Scores are research-informed estimates, not market data.\n") L.append("![map](applicability_map.png)\n") L.append("**Read it in two axes:** the *composite* says how **BH-shaped** a domain is " "(substrate + many layers + selective read). The *verdict* says whether there " "is **novel work to do** — because being BH-shaped is not enough if the " "store-once + selective-read pattern is already mature SOTA.\n") L.append("| # | domain | composite | verdict | scale | why |") L.append("|---|---|---|---|---|---|") for i, r in enumerate(scored, 1): star = " ★" if r["built"] else "" L.append(f"| {i} | {r['area']}{star} | **{r['c']:.0f}** | {r['verdict']} | " f"{r['scale']} | {r['note']} |") L.append("\n★ = prototype already built in this repo (bhmem / bhtrace / bhckpt / bhanno).\n") L.append("## The honest finding\n") L.append("1. **Most domains score HIGH (90-100) — and most are `ANCHOR`.** The BH shape " "is everywhere, but *store-the-substrate-once + selective-read* is already " "mature production SOTA almost everywhere it matters: DICOM (medical), " "COG+STAC (satellite/GIS), lakeFS/DVC (data lakes), CRAM+tabix (genomics), " "MAM (video), Relativity (legal), named graphs (KGs), S-LoRA (checkpoints). " "BH earns **credibility by analogy** there, not a novel build.") L.append("2. **Among the 20 domains surveyed, CAD/BIM (90) was the only one classified " "`BUILD`.** The research found no mainstream tool that stores ONE canonical " "building substrate once with many additive AND rival discipline/version " "overlays co-registered as first-class layers with selective branch/region " "reads — today federation *duplicates* and treats clashes as ad-hoc. That " "union is the gap. (A claim about this sample, not about every possible domain.)") L.append("3. **The recurring under-served slice is the *rival* layer, not the *shared* " "one.** Across annotation, MAM, eDiscovery, genomics and KGs, existing tools " "store the substrate once but treat **conflicting interpretations as noise to " "adjudicate into one ground truth** — not as first-class, queryable, " "co-registered rival layers. That is precisely what `bhanno` models. So the " "sweep *suggests* BH's principal **still-under-explored** contribution is " "**treating rival interpretations as first-class entities** — narrower and " "sharper than 'a universal format'. Stated as the differential observed so " "far, not a final reduction of what BH can be.") L.append("4. **Our four built prototypes all land `ANCHOR`.** They proved the " "*generalization* (one envelope, many domains) honestly — but the sweep shows " "their economic win is largely already-solved. That is the method working: it " "refuses to let us overclaim.") L.append("5. **Controls behaved.** Image/audio codecs (25) and vector DBs (30) sit at " "the bottom as `DELEGATE` — the score discriminates dense-signal from " "structure-dominant.") L.append("\n## Recommendation\n") L.append("Two moves, in order. **First, formalize the principle** (see " "`BH_PRINCIPLE.md`): the sweep shifts BH from *a file format* to *a " "representation model* — multiple concurrent, possibly contradictory " "interpretations sharing one immutable substrate and remaining queryable " "without forced adjudication. That definition is what separates BH from the " "`ANCHOR` systems; it is the real contribution this sweep surfaced.") L.append("**Then, if a measured test is wanted**, the natural candidate in this sample " "is **CAD/BIM** — the only domain here classified `BUILD` — a `.bh` where the " "building object-graph is the substrate, with rival discipline/version " "overlays + clash annotations co-registered, selective branch reads, and the " "deferred-adjudication face.") (OUT / "APPLICABILITY_SCORECARD.md").write_text("\n".join(L) + "\n", encoding="utf-8") print("ranked:") for i, r in enumerate(scored, 1): print(f" {i:2d}. {r['c']:5.1f} {r['verdict']:8s} {'*' if r['built'] else ' '} {r['area']}") print(f"\nwrote APPLICABILITY_SCORECARD.md + applicability_map.png") if __name__ == "__main__": main()