vizfootprint, on public data

Four dashboards over data nobody here authored. Every interaction is a commit you can travel back to, every proposal an agent makes is judged before it lands, and every empty cell keeps its kind — a withheld number, a bound and a real zero are never the same silence.

Every chip on the cards below was read off the surface it names — off the built definition, off the one recorded walk, or off the one table a person had to write — and each card says which. Nothing here was typed into this page. The CDC demo is two cards because it is two builds: the desk published here, with the disease co-occurrence graph, and the story page, built without it, which carries the only recorded walk this repository has. That page is its own build (npm run story:page) and is not published on this site.

What this build is, and is not

These pages are static. There is no server: each desk fetches the repository's own committed CSV files over http, runs the same shaping code the served version runs, and builds its dashboard in your browser. The first load of either desk therefore carries the whole snapshot, and it is not small.

Because nothing runs on the other end, three things are missing, and each desk says so on its own page:

the analyst — it needs a model key and a process to hold the conversation;
a durable log — your commits live in the tab and are gone when you reload;
a refresh — the files are what the repository committed, and re-reading them answers the same bytes.

Everything else is the real thing: selections, acts, undo, named paths, bookmarks, compare, the Sheet, the commit log and the branch map.

Where the data comes from

Disease counts: NNDSS Weekly Data on data.cdc.gov — a work of the United States Government, public domain. Using this material does not mean the CDC, the agency that published these tables, the Department of Health and Human Services or the United States government endorses this site or anything said on it. The same tables are available from the CDC itself, free of charge, at that link; the copy here is a convenience slice, not a source of record.

Grid data: Source: U.S. Energy Information Administration, Hourly Electric Grid Monitor, six-month file for January–June 2025 (retrieved September 2026). Public domain; EIA's logo and seal are trademarks and are not reproduced here.

Exoplanets: This research has made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program. Two tables are used: the Planetary Systems Table (10.26133/NEA12) and the Planetary Systems Composite Parameters Table (10.26133/NEA13). The archive publishes no licence for them; it asks for the acknowledgement above, and for work using a specific literature reference to cite that reference directly. The slice here is a convenience snapshot for a demo, not a source of record.

Protein structure: one entry from the RCSB Protein Data Bank's copy of the wwPDB archive (1AY7, files.rcsb.org) — the committed one, and the desk's example; its search box opens any other entry in the archive, fetched in the browser from the same archive and credited the same way, out of that file's own header records. wwPDB releases its data files into the public domain under the CC0 1.0 Universal dedication, which asks for nothing; the archive asks that a structure's depositors and its primary citation be credited, and that desk prints both, read out of the file's own header records. The 3D picture is drawn by Mol* (MIT), which is somebody else's code joined to this interaction grammar through the library's renderer contract — not a chart this project wrote.