Paper Classification — against the two-axis model
Companion to docs/paper-taxonomy.md. Every shipped and planned paper plotted
on the two axes: velocity (fast/slow data → refresh cadence) and window
type (closed = finishable / open = perpetual standing monitor). Built
2026-06-09 from a full sweep of web/.../articles, the workspaces, and the
three planning docs + the GH research backlog.
Headline finding
All 14 shipped papers are closed-window. The open / standing-monitor quadrant is 100% empty in shipped work. Every published paper fences itself to a fixed past interval (an event, an outbreak window, a historical span). None is a living instrument that refreshes forever. The model predicted this gap; the data confirms it exactly. The 60-second-forever pipeline was built for the empty quadrant, and we haven't shipped into it once.
Shipped papers (14) — all CLOSED window
Velocity sub-split (matters only for the "finalized" caveat, since closed + finalized = truly done):
Closed + Fast data — closed, but check revisions have settled
| slug | window | data | finalized? |
|---|---|---|---|
| fireball-surge-2026 | Q1 2026 | CNEOS fireballs | Not yet — CNEOS revises; 2026 not closed until ~Q3 |
| sanriku-m66-may2026 | event, May 2026 | USGS quakes | Settling — USGS revises mag for weeks |
| kermadec-doublet-apr2026 | events, Apr 2026 | USGS quakes | Settling |
| cascadia-swarm-apr2026 | Apr 8–12 2026 | USGS quakes | Settled |
| reno-earthquake-swarm | Apr 12–13 2026 | USGS quakes | Settled |
| solar-watch-april-2026 | Apr 2–5 2026 | GOES/DSCOVR/DONKI/NMDB | Settled |
| wspr-aircraft-detection | 24h Apr 12–13 2026 | WSPR + ADS-B | Settled (pre-registered null) |
| wspr-tornado-precursor (v1) | Apr 9–19 2026 | WSPR + SPC | Settled |
| wspr-tornado-v2-historical | 7 outbreaks 2011–2023 | WSPR + SPC | Settled |
| wspr-tornado-v3-general | 12 outbreaks 2021–2024 | WSPR + SPC | Settled |
| wspr-tornado-v4-decoupled | 12 events 2021–2024 | WSPR + GLM + SPC | Settled |
Closed + Slow data — closed and stable, done-done
| slug | window | data |
|---|---|---|
| earthquake-lights-104-years | 1910–2014 | USGS + NUFORC historical catalogs |
| wspr-solar-cycle-bug | solar cycles 24–25 (2008–2026) | WSPR monthly aggregates + SILSO |
| skinwalker-ranch-seven-layers | 10 filming days 2020–2023 | 7 mixed layers (Kp fast … Black Marble slow) |
Reading: 11 fast / 3 slow. The only live maintenance flag on the whole shipped corpus is the handful of recent-2026 fast papers whose source data is still being revised (fireball-surge most of all). Everything else is frozen.
Planned papers — where the open quadrant finally appears
The backlog is still mostly closed-window historical tests, but it contains a real cluster of open / standing-monitor candidates. These are the ones that would populate the empty quadrant and actually exploit the live pipeline.
OPEN / standing-monitor candidates (the quadrant to deliberately fill)
| id | topic | data | why it's open |
|---|---|---|---|
| GH #105 | Starlink reentry watch | CelesTrak TLEs | Explicitly a watch — events keep arriving, never finishes |
| paper-ideas #8 / DP-seed-7 | "Was 2024 weird?" → annual weirdness / climate report | all 113 metrics | Recurring by design — re-runs every year, the change is the point |
| GH #75 | SPC convective-outlook verification | SPC forecasts × NWS reports | Grows as forecasts/observations accrue; living scorecard |
| GH #207 / DP-seed-3 | NWS-alerts living climatology | NWS 713K-record archive | "System evolved over time" — a climatology that should refresh |
| DP-seed-4 | Solar-max 2024–25 retrospective synthesis | WSPR/GLM/IGRA/indices | Could run as a rolling solar-cycle monitor |
CLOSED / finishable — the rest (majority of the backlog)
Bounded historical tests; ship once and (once finalized) archive. Grouped by velocity:
- Closed + Slow (stable, shelve on ship): magnetic-pole drift (paper-ideas #3 / DP-006 / GH #206), Carrington tail-risk (#4), methane × quakes (#42), tidal × GVP eruptions (#43), ERA5 urban-heat (DP-seed-8), Sentinel-1 SAR coverage (#78).
- Closed + Fast (perishable inputs, but the fence freezes them): DP-001 IGRA×SPC tornado, DP-002 Tornado-Alley drift, DP-004 Earth normal-modes in WSPR, DP-005 RF Anthropocene, DP-007 solar×upper-air, DP-008 tidal earthquake triggering, DP-009 WSPR×TEC, plus the WSPR/space-weather GH issues (#201, #200, #132, #120, #102, #76, #68, #69), AQI case studies (#203, #71), Safecast (#202, #17), hydrology (#16), volcano-quake (#23), GW-null (paper-ideas #6).
- Hybrid (closeable now, open if you keep watching): magnetic-pole drift and Tornado-Alley centroid drift both have a closed "here's the trajectory so far" framing AND an open "track it forward" framing. Scoping choice decides which.
Velocity note
The scouts tagged most sources FAST. True at the source level, but for closed papers velocity only governs shelf life, not whether the paper finishes. It's the OPEN candidates where fast velocity earns its keep: a standing monitor on fast data (Starlink reentries, SPC verification) is high-value precisely because it reads different every refresh. A standing monitor on slow data would barely move.
Takeaways
- Shipped = 14/14 closed-window. We are exclusively a snapshot shop so far.
- One maintenance flag: recent-2026 fast papers (fireball-surge especially) sit on data the source is still revising — "closed" but not yet "finalized."
- The empty quadrant has ~5 ready tenants. Starlink reentry watch (#105), the annual weirdness/climate report (paper-ideas #8 / DP-seed-7), SPC outlook verification (#75), and the NWS living climatology (#207) are the cleanest first standing monitors — and they're the only backlog items that use the live pipeline for what it's for.
- Duplicate clusters to resolve before queueing: magnetic-pole drift exists three times (paper-ideas #3 ≈ DP-006 ≈ GH #206); WSPR-centric pitches overlap four ways (DP-004 / DP-009 / #132 / #200); solar-cycle omnibus twice (paper-ideas #11 ≈ DP-007).