A first reading. One UTC day of output from the GDELT 2.0 Global Knowledge Graph — the syndicated, predominantly English-language news ecosystem's reporting attention — displayed without synthesis. This is the attention half of the publication's signature move, alone. The instrument half is not joined here.
The publication has not yet published Issue No. 1. This page is operator-built scaffolding; the same job will eventually run on the autonomous pipeline. Per protocol, GDELT is admitted as an attention instrument only — a sensor of what the syndicated news ecosystem reports on, not as an account of what happened. Source article URLs are not followed.
Themes are GDELT's controlled vocabulary of about ten thousand tags. Counts are mentions across sampled articles for the day. The vocabulary is built for the news domain and skews accordingly; absence of a theme from this list reflects either low attention or vocabulary blind spots.
| Theme | Mentions |
|---|---|
| TAX_FNCACT | 4,822 |
| EPU_POLICY | 2,310 |
| CRISISLEX_CRISISLEXREC | 2,161 |
| TAX_ETHNICITY | 2,098 |
| UNGP_FORESTS_RIVERS_OCEANS | 1,996 |
| WB_696_PUBLIC_SECTOR_MANAGEMENT | 1,718 |
| USPEC_POLITICS_GENERAL1 | 1,525 |
| CRISISLEX_C07_SAFETY | 1,444 |
| TAX_WORLDLANGUAGES | 1,407 |
| MANMADE_DISASTER_IMPLIED | 1,405 |
| SOC_POINTSOFINTEREST | 1,405 |
| LEADER | 1,391 |
| USPEC_POLICY1 | 1,378 |
| EPU_ECONOMY_HISTORIC | 1,335 |
| GENERAL_GOVERNMENT | 1,226 |
| WB_2432_FRAGILITY_CONFLICT_AND_VIOLENCE | 1,179 |
| GENERAL_HEALTH | 1,178 |
| WB_133_INFORMATION_AND_COMMUNICATION_TECHNOLOGIES | 1,144 |
| WB_621_HEALTH_NUTRITION_AND_POPULATION | 1,112 |
| MEDICAL | 1,097 |
| WB_840_JUSTICE | 1,086 |
| WB_678_DIGITAL_GOVERNMENT | 1,086 |
| TAX_ECON_PRICE | 1,070 |
| CRISISLEX_T11_UPDATESSYMPATHY | 1,063 |
| KILL | 1,043 |
| EDUCATION | 1,037 |
| ARMEDCONFLICT | 945 |
| MEDIA_MSM | 928 |
| EPU_POLICY_GOVERNMENT | 917 |
| WB_694_BROADCAST_AND_MEDIA | 898 |
Named-entity extraction over article text, surfaced names ranked by how often the news ecosystem mentioned them in the sample. Honorifics, common-noun false positives, and disambiguation collisions are not filtered — this is the raw instrument reading.
| Person | Mentions |
|---|---|
| donald trump | 352 |
| los angeles | 118 |
| marco rubio | 112 |
| nicola sturgeon | 68 |
| narendra modi | 63 |
| pope leo | 57 |
| gavin newsom | 48 |
| benjamin netanyahu | 46 |
| craig covey | 46 |
| magnifica humanitas | 46 |
| peter murrell | 41 |
| whatsapp linkedin | 39 |
| santon downham | 38 |
| jesus christ | 36 |
| james manning | 33 |
| joe biden | 32 |
| jane barlow | 32 |
| esmaeil baghaei | 29 |
| rerum novarum | 25 |
| kevin warsh | 24 |
| ted cruz | 24 |
| tom morgan | 23 |
| masoud pezeshkian | 23 |
| lindsey graham | 23 |
| barack obama | 23 |
| andrew matthews | 22 |
| caroline abrahams | 21 |
| dezi freeman | 21 |
| swanholme lakes | 20 |
| mojtaba khamenei | 20 |
Date sampled: 25 May 2026 (UTC).
Sample: 8 of 8 attempted GKG files, one every three hours across the UTC day. Each file is a 15-minute window. Roughly 120 minutes of attention sampled out of 1,440.
Counted: 6,094 GKG records ingested, yielding 4,255 distinct themes and 11,771 distinct persons.
Excluded: source article URLs are not fetched. Tone scores, locations, organisations, V2-format columns are present in the raw data but not shown here.
Script: pipeline/first_reading.py in the project repository. Pure standard library, reproducible from the raw GDELT URLs cited below.
Files cited:
What this is not: an issue of The New Relation. Issues join attention to instrument measurement (the publication's signature move) and pass through synthesis and self-evaluation. This page is the attention half alone, hand-run, before either pipeline exists.
What comes next: the same scaffolding for one physical instrument (USGS earthquakes is the candidate), then the smallest possible join between the two.