To do

This page is the wiki's working list: where it stands, page by page, and what to do next. It is rebuilt with the site.

Each entry gets a rough maturity score out of 100. It adds up whether a person wrote or has read it and whether it has been promoted, whether it lists its sources, its length, a description that fits a search result, a picture, how few machine-prose tells it has, and whether other pages link to it. The score is a way to sort, not a grade.

The open items are the project's own list, todo.md, shown as it stands in the repository.

One list for the project, kept current. Decisions and reasoning live in devlog/; this is only what is open and who it waits on. Cited from README.md.

Anselm's

• Read the ten use-case pages and nine guides for the first patron's ear; promote or send back.

• Promote entries by section: engines first, then data sources, books, twins.

• Write the four bylined articles (seeds in the survey's appendices D to G, archive/v1/seed/Digital-Twins-Industry-Survey.md): what is a digital twin; are there whole systems models yet; the limits of reason; the ecological frontier.

• Decide the chrome: the site name, the twins label, the home text.

• Decide whether to drop the staging badge on digitaltwins.wiki. (Indexing: opened 2026-10-23 at Anselm's word; the machine copies and /todo/ stay noindex.)

• A town for the town-twin experiment (devlog/20261023-town-twin-experiment.md); Anselm will circle back with one.

• Climate Shift Index: how a regional, daily attribution figure belongs in the wiki (a Vitals layer, a twin layer, or only the Communication entry).

• Search Console and structured data (schema.org) now that indexing is open; semantic search later, if plain Pagefind proves too literal.

• Wikipedia's references, triaged in inbox/: Gemini Principles, Virtual Bradford, the building reviews, a UK energy twin, Helsinki and Kyoto. Which become entries.

• Forests 2030: send the link to Kormos's LinkedIn post if its paragraph (updates through 2030, PRIMA statistics) should return; removed 2026-10-24 for want of a source.

• Novel bamboo uses (inbox/), fuel for the bamboo scenario: entries or not.

• Law (shelf since 2026-10-24, devlog/20261024-stretching-and-tearing.md): six US sources so far. Which non-US layers come next (England's Planning Data platform is named in the Rules layer term).

• Tearing: Vitals now draws the tipping-element thresholds beside today's warming, and the Tipping points term covers societies (Regime Shifts DataBase, IPC phases). Still open: a page on how to show a tear in a country or bioregion (Hormuz's dependence-and-stress axes as one worked approach), as Anselm's essay or a wiki entry; and an essay of his on "isolated pools", if he wants one.

• Bioregional tearing (Anselm, 2026-10-24: "that may be where we actually need to run real models"). Pick a bioregion with a known threshold from the Scenarios shelf (peat fires, San Joaquin groundwater, the Delta, Alwar) and run a model that looks for the runs that flip; a sibling of the town-twin experiment, one scale up.

• Ambassadors: real people to stand for a patron path, if any.

• Export the ChatGPT guide site (sign-in only) before deleting it, if there is anything in it beyond the survey.

• Many authors, later (Anselm, 2026-10-23): the wiki will have many authors and people wrangling change histories; the GitHub commit log is where disputes will be shown, visually. A process for that is still to design: review roles, how a disputed claim is marked, how the log is surfaced on a page.

• Inline citations (Anselm, 2026-10-23: "aggressively"): [^n] markers against each entry's sources, rendered and checked since 2026-10-23. Done: Scenarios, Communication, Everyday questions, Engines, Data, the history article; Articles and Notable twins on 2026-10-24. Left, about 100 agent-written entries: Media (27), Sites (19), Use cases (12), Terms (11), Ontologies (10), Guides (9), Organizations (9), Presentation (8), and a handful of events, mandates and about pages. Entries of human or survey origin are skipped. Brief: devlog/20261024-wrap-up.md.

• The maturity score on /todo/ barely moved with citations (mean 30): sources were reweighted down when citations were added. Decide whether cited coverage should count for more.

• Wikipedia's Digital twin article as a map (Anselm: "a lot of great detail ... we may want to scavenge"). Its text is CC BY-SA, so copying would oblige attribution and the same licence; instead read its ~50 references, open the primary ones, and write from those (history of the term, NASA and Grieves' early uses, the industry examples that touch places), the same rule as for Mike's syntheses.

Agent work, when asked

• Mary's tools from their own pages: i-Tree, ENVI-met, PALM-4U, UMEP; and A/B Street for Ben.
• The "there is no one tool" guide.
• Ontologies shelf: more papers and observations on how models cut up a place (Anselm, 2026-10-19, expects to add them); each from a page read, bar as in devlog/20261018-patrons.md. Whether engines get a layer field, and whether twins name their four parts, is in devlog/20261019-stack.md.
• ontology.monster is drawing itself an SVG icon (Anselm, 2026-10-19); when it exists, fetch it beside the entry as depiction. Cards without an image now show an empty box so lists align.
• Reading list of 2026-10-19 (inbox/2026-10-19-reading-list.md): retry OpenGridWorks and the IOP wildfire paper with a headless browser; the fourteen X-only items need a maker page before they can be entries.
• Things Anselm wants to model (devlog/20261020-things-to-model.md): the post is written (articles/places-with-a-question). His call whether local / bioregional / macro becomes a field, and which of the twelve get a question page. Harvest2Market (NASA Harvest) could be a tool entry; the Amazon soy pact page (phys.org) needs a browser fetch.
• Forests 2030 (articles/forests-2030, devlog/20261020-forests-and-destine.md): published at primaryforestsandclimate.org on 2026-10-21 and the entry links there. PRIMA, the mapping app the post promises, has no page yet. forestregimemap.net was unreachable; retry for a data source entry. The Rogers et al. 2022 paper on ecosystem integrity could be its own article entry.
• Organizations (Anselm, 2026-10-20): only nine entries, but 122 distinct maker sites stand behind the tools, data sources and twins. The made_by predicate exists and is unused. His call on a landscape view of makers sized by the entries behind them; see the devlog entry above.
• Inbox inbox/2026-10-20-links.md: Google and Working Trees silvopasture in Brazil, read as fuel for the land-use-change use case, not an entry.
• kepler.gl: the home page is a script shell; find a readable page (docs or the repository README) and rewrite the body.
• Retry the eight YouTube channel feeds on a quiet day.
• Move tools/import-linkedin.js and tools/fetch-vitals.js onto store.put.
• Prune or extend the search vocabulary in data/sites/index.md.
• Scan-only cron, and a daily vitals refetch and redeploy.
• Fuel for later, noted by Anselm 2026-10-18: a great many people are trying to explain digital twins with layered pictures, which is what inspired the Bay Layers scene on the home page. An image search for "several layers digital twin" shows the genre: https://www.google.com/search?udm=2&q=several+layers+digital+twin Possible uses: a terminology entry on the layered-diagram convention, or credited examples beside the Bay Layers entry.

Done

See the tail of devlog/.

Prose (2026-10-20, from the adversarial reading; devlog/20261020-prose-tells.md)

• Anselm's call: a site-wide rewrite pass, template residue first (status stamps, shelf-filing closers, absence sentences), then read-then-refute per entry with node tools/tells.js --file as the brief.
• Gate every new draft on tools/tells.js and the heckler inside the drafting workflow.
• Site: the page prints the description as a dek and 42 bodies open by repeating it; either the dek goes or the opener must start where the dek stopped.
• Unopened technical terms in 164 files: no regex; needs a reader pass with the non-technical patron in mind.

Mike's Niantic and NVIDIA syntheses (2026-10-21, devlog/20261021-mike-niantic-nvidia.md)

• Niantic: Singapore Land Authority national VPS map; Cross-View Splatter paper. Two drafts done (twins/rancho-cordova, ontologies/spz).
• NVIDIA: read NVIDIA's own pages for Earth-2's users (Taiwan CWA, Southwest Power Pool) and fold into engines/weathernext-nvidia-earth-2; PALM and Cesium microclimate claim; Aerial Omniverse Digital Twin.

Coverage sweep against outside lists (2026-10-21)

TESSERA was missing because the wiki grew from the survey, the LinkedIn scrape, the inbox and Mike's maps, and nobody checked against a list made by someone else. Tested today: the curated GitHub lists have it (Jack-bo1220/Awesome-Remote-Sensing-Foundation-Models, satellite-image-deep-learning/techniques); sacridini/Awesome-Geospatial and acgeospatial/awesome-earthobservation-code do not; Wikipedia has no TESSERA article and no Category:Digital_twins, so it is weak for this. Idea: a tool that reads such a list and diffs it against our entries, giving a gap list for Anselm to triage. Entries still written only from the projects' own pages.