Lab
First experiments with data and code — data stories and analyses. Open, unfinished, in the making.
Experiments · ongoing, refreshed each night
Ongoing experiments with open data sources — each names its sources and discloses how it is made.
-
The Protocol — Daily figures from twelve open sources
Every night a pipeline writes the minutes of the planet's session — from twelve open, citable sources, deterministic, no LLM. Every agenda item ends the same way: Resolution: adjourned.
Method sheet → -
Half-Life — How fast attention to disasters fades
For every major event with fatalities, a pipeline measures how quickly attention decays — half-life by half-life, next to the constants of physics. Intake follows a published rule; nobody selects.
Method sheet → -
Parallax — How Wikipedia language editions differ on contested topics
The same contested thing across several Wikipedia language versions — and the measure of which claim each version states and which it conceals. The Japanese description of the Senkaku Islands, for instance, never mentions the territorial dispute.
Method sheet → -
The Policy — Climate cost, computed from market data as a „premium"
An insurance policy on the present, its premium recomputed each night from real market data. The market has long since priced in the climate catastrophe — and the premium is rising: +179% since 1998.
Method sheet →
Studies & data stories
-
Study: Überflug
SGP4 in the browser — which Earth-observation satellites currently have a location in view. A technical étude, not an artwork.
- study
- satellites
- orbital
- sgp4
-
Analytics Engineering: BigQuery and dbt as the foundation
How BigQuery and dbt turn raw events into reliable, tested models — the foundation for any dependable decision.
- BigQuery
- dbt
- Analytics Engineering