# art_nft_history — "From Photography to Blockchain"

Working paper arguing that digital/blockchain-native art is a major movement comparable to
photography's entry into the canon. Stance and scope chosen 2026-08-01: **major movement, full
claim; full article (8,000–12,000 words)**.

## Workflow (same pattern as `compression paper/`)

1. `research_notes.md` — all sourced facts from the 2026-08-01 research sweep; UNVERIFIED items flagged.
2. `data.py` — every number used in figures/text, with source comments.
3. `generate_figures.py` — renders `figures/fig01…fig06` (matplotlib, house style).
4. `build_paper_v1.py` — builds `from_photography_to_blockchain_v1_draft.pdf` (reportlab).

Rebuild everything:

```
python generate_figures.py && python build_paper_v1.py
```

## Draft v2.0 (current)

`build_paper_v2.py` → `from_photography_to_blockchain_v2_draft.pdf` (23 pp, ~10,100 words).
Changes from v1: figures/tables renumbered in reading order (cadence = Fig 1, timelines = Fig 2,
…); five illustration **Plates I–V** added (daguerreotype camera + PD "Cameraman" photo;
1928 steamboat film-frame; Ether/Bitcoin marks + chain schematic; Black Square / "0,10"
recreation; Impression Sunrise homage); Beeple "third-highest living artist price" now
qualified "then / after Koons and Hockney"; royalty-collapse figure labeled sector-wide.
All 15 headline numbers independently re-verified 2026-08-01 (all confirmed).

## Getting the REAL images (v3 pipeline)

The cloud sandbox cannot reach Wikimedia Commons (proxy 403 — also blocks WebFetch and
npm), so v2's plates use copyright-safe stand-ins. The fix runs on the local PC:

1. **Double-click `GET_REAL_IMAGES.bat`** (runs `fetch_real_images.ps1` — plain Windows
   PowerShell, no Python needed). It resolves each work via the Commons API (exact title,
   then search fallback), downloads 1200px public-domain scans of: Niépce heliograph,
   Daguerre's Boulevard du Temple, Stieglitz's Steerage, an 1839 daguerreotype camera,
   Monet's Impression Sunrise, the Steamboat Willie poster, Black Square, the 0,10
   installation photo, Nadar's Baudelaire, and the ETH/BTC marks into
   `figures\open_source\` (with a `_sources.tsv` license log).
2. Tell Claude "images downloaded" → it re-stages the files, re-runs
   `generate_figures.py` (whose `plates_real_overrides()` automatically replaces the
   drawn plates with the real scans, and adds a Plate VI photography triptych), and
   builds `build_paper_v3.py` — whose captions and optional Plate VI adapt at runtime to
   whichever real images are present.

Fallback behavior is automatic: with no downloads, v3 renders identically to v2.

## Draft v1.0 contents

9 sections + abstract, 6 figures, 2 tables, 53 references. ~22 pp, ~9,900 words total
(~8,300 body). Key corrections vs. the original prompt notes: the Met's 1938 Disney cel is
"The Vultures" (obj. 38.154), not a Vain Hag cel; MoMA acquired Anadol's *Unsupervised* in
Oct 2023 (not 2024); Pompidou's 2023 acquisition is 18 NFTs by 13 artists; Dataland opened
June 2026 (slipped from 2025); no MoMA Disney show in 1939 (it was *Bambi*, 1942).

## Next steps (v2 candidates)

- Pull the digital-art chapter of the Art Basel/UBS 2026 report PDF to replace the two
  UNVERIFIED share framings (online-sales share; report's own NFT figures).
- Attack pass (ATTACK_NOTES.md) + defense pass, per compression-paper workflow.
- Consider adding a case-study box on a single work's provenance chain (e.g., Fidenza #313).
- Optional: shorter conference cut (~4,500 words).