plasmagicians substrate spec · plate i · graphite Status Spec Live Archive Corpus Questions ⏃ sign in ↩ v1
Plate i · substrate overview · graphite

Plasmagicians — Substrate Spec

a Bret-Victor explorable-explanations wiki on a Rust + WebGPU + WASM substrate, anchored on the PPPL first-year graduate programme

A Bret-Victor-style explorable-explanations wiki on a Rust + WASM + WebGPU substrate, anchored on the PPPL first-year graduate programme (AST 551 / AST 552 and Generals prep), with a heritage archive of the original 2006 PlasmaWiki — which Danielle Fong and Abe named at Princeton. Authored at the substrate level by the operator; populated at the plate level by a physics-specialist Opus 4.8 max agent running in a custom harness with thinking-block visibility.

where we stand

✓ shipped · solvers crate (Rust, 8/8 tests, 18 KB WASM) — caught a 1000× error in v1 draft
✓ shipped · four spec documents (plate package · hyperstack blocks · solvers catalog · authoring prompt)
✓ shipped · licence.md (founder CC BY 4.0 + MIT, fair-use stance, 2006 lineage acknowledged)
✓ shipped · this graphite spec preview — Britannica plate engraved on lit concrete
✓ shipped · PlasmaWiki mirror complete — 3,004 files / 951 HTML articles / 80 MB
⏳ next · scripts/preview-plate.sh · plates/01-nlambda3 · chrome-strip · KaTeX recovery
⏳ next · canvas substrate port from xanadu-cx (deferred — corpus first)

pivots that shaped this day's build

fromto
iinfinite-canvas wikireadpath UI — smart-for-you order, snap-to-plate momentum, snap-out gesture
iieditorial platesBret-Victor explorable explanations — every claim wired to a live model
iiihand-authoring 8 platessubstrate for Opus 4.8 max agent in a custom harness
ivvellum LightCell paper themegraphite + spectral inks — Na yellow, Li red, white-hot · PPPL register
vdense-Na clusterAST 551 / 552 + Generals — hundreds of plates, agent fleet
viflat dark themefuckin' awesome — spectral glow, ray-traced light, graph paper, Britannica register

plate package spec

Each plate is a directory, not a single file — because the substrate is HTML + WASM + WebGPU + Rust + Markdown + WGSL. A plate brings whatever the physics demands.

plates/<id-slug>/
├── plate.toml                ← required: manifest
├── prose.md                  ← optional: markdown + KaTeX + explorable refs
├── citations.toml            ← optional: overflow citation list
├── explorables/              ← optional: live interactive demos
│   ├── <name>.toml          ← per-explorable manifest (hyperstack blocks)
│   ├── <name>.wgsl          ← optional: custom WebGPU shader
│   └── <name>.rs            ← optional: custom Rust solver (extras crate)
├── assets/                   ← optional: images, datasets
└── render.html               ← optional escape hatch

plate.toml — the manifest

[plate]
id    = "01-nlambda3-cooperative-threshold"
title = "When Nλ³/64 > 1"
subtitle = "The cooperative-emission threshold for dense atomic vapor"
cluster = "dense-na"
status = "published"
authored_by = "Danielle Fong"   # or "agent:opus-4.8-max:physicist-v1"
syllabus = ["AST552:radiative_transport", "Generals:plasma_waves"]

[ancestor]
references = [
  "archive/qed.princeton.edu/main/PlasmaWiki/Mean_free_path.html",
]

[rails]
on = [
  { rail = "dense-na-readpath",   position = 1 },
  { rail = "radiative-transport", position = 4 },
]

[citations.molisch1998]
authors = "Molisch, A. F. & Oehry, B. P."
year = 1998
title = "Radiation Trapping in Atomic Vapours"
publisher = "Oxford University Press"
Plate-as-directory feels right for this aesthetic — or does the agent want a single self-contained file per plate?

a live explorable — proof of substrate

Every numerical claim below comes from a Rust function compiled to WASM and called by the slider handlers in the left gutter. Slide a knob; math re-typesets; the regime label flips at the threshold; the mini canvas redraws. Reading is doing.

$n\,\lambda^{3}/64 \;\gtrless\; 1 \quad\Leftrightarrow\quad n \;\gtrless\; n_{\text{crit}} = \dfrac{64}{\lambda^{3}}$

Fig. I. Saturated sodium vapour density (amber) crosses the cooperative-emission threshold (incandescent rule) at the regime boundary. Scrub the gutter sliders to step through the phase plane.

The substrate primitive: slider in left gutter → WASM solver call → math + canvas + label update in lockstep. Every plate's explorables vary the solver / inputs / displays while keeping this same shape. The agent learns the shape once and replicates it.

PlasmaWiki — the heritage trellis

The 2006 wiki (which you and Abe named) has been fully mirrored from w3.pppl.gov/~dboyle/PlasmaWiki/ into sites/plasmagicians-v2/archive/ as the heritage trellis the new wiki grows on. The mirror finished while this preview was being engraved.

statisticvalue
Files mirrored3,004
HTML articles951
Total size80 MB
Date range2006–2017 · AST551 / AST554 / MG present
CategoriesSingle Particle Motion · Classical Mechanics · Electromagnetism · MHD · Kinetic Theory · Plasma Waves · Generals · Prelims · …
Licence (footer claim)CC BY-NC-ND 2.5 — 2009 export artefact
Licence (founder intent)CC BY 4.0 + MIT — reaffirmed in licence.md

Next: strip MediaWiki chrome (sidebar / login / edit links / social bookmarks) and replace texvc math PNGs with live KaTeX rendering — the original LaTeX is preserved in each alt=. A significant typography upgrade for free.

corpus scope — AST 551 / 552 + Generals

Per operator: the corpus is anchored on the first-year Princeton graduate programme. AST 551 (Introduction to Plasma Physics I) + AST 552 (II) as the spine, with simultaneous tagging for Generals (qualifying-exam) preparation. Approximately 150–300 plates at full coverage, written by the physics-specialist agent fleet.

provisional cluster map

clusteranchor course≈plates
Single-particle motion · drifts · gyro · adiabatic invariantsAST 551 weeks 1–315
Plasma waves · Langmuir · ion-acoustic · Alfvén · R L O XAST 551 / 552 wk 4–825
MHD · ideal · resistive · force-free · equilibria · stabilityAST 552 first half25
Kinetic theory · Vlasov · Landau · Fokker-Planck · BBGKYAST 552 second half20
Radiative transport — dense-Na cluster (canonical)extension · LightCell10
Diagnostics · Thomson · interferometry · Langmuir probes · ECEAST · Generals15
Transport · classical · neoclassical · anomalous · turbulentAST 552 · Generals15
ICF + MCF · tokamak · stellarator · ICF basics · FRC · MTFAST · survey20
Boundary plasmas · SOL · divertorexperimental focus10
Lab + computational methods · PIC · MHD codesAST · projects10
LightCell + modern extensions · chemiluminescence · anti-entropyoperator corpus10

Total target: ≈150–300 plates. Agent-fleet authoring with the operator watching thinking blocks in a custom harness — a multi-month build, not a single-day push.

your read on

  1. Graphite + spectral glow + Britannica register — is it fuckin' awesome enough? Or push harder still?
  2. Both gutters interactive — left sliders + live readouts + mini-canvas + spectral palette; right rail-progress + ToC + cites + archive ancestors. Symmetry working?
  3. Topbar running-head in small caps, italic anchors — sufficiently Britannica? Or too restrained?
  4. Demo column treatment — gravure-shadowed canvas with figure caption — feels right?
  5. AST 551 / 552 + Generals scope right? Wider (557 / 559)? Tighter?
  6. Reference plate 01-nlambda3 — author it now by my hand as the benchmark, or let the agent's first run be the bootstrap?

※ flag specific changes — the visual + layout decisions here propagate to every plate, so they need to be right before the agent starts authoring at scale.

plasmagicians.com — substrate spec — plate i

NORMAL · /spec.html · i:7 spec preview ready · unauth press / palette : cmd e edit ? help