Public records, rebuilt into datasets that can be questioned.
Public records, rebuilt into datasets that can be questioned.
A Data Infrastructure Project
Much of what we want to know about political life is already recorded somewhere public — election returns, arms transfers, conflict events, parliamentary speech. Most of it sits in formats nobody can analyse.
Corpus Labb takes that material out of its original form, verifies it against itself, and publishes it as datasets that can be questioned and reused.
official reports retrieved across seven election types, covering every national and local election since 2002.
rows released, from national totals down to district level — with ballot-box detail for selected provinces.
arithmetic checks run against the data itself: party votes must sum to valid votes, turnout must reconcile.
Each project is a standalone dataset with its own access rules and documentation.
District-level results for every national election since 2002 — parliamentary, presidential, referendum and local — retrieved from official records, cross-checked and reshaped for analysis. Includes ballot-box level data for selected provinces.
Live · RestrictedA short climate-negotiation simulation played on phones around a table. Twelve delegations, one contested text, two rounds to the gavel. Built for the SETA stand at the COP31 Green Zone; runs during live sessions.
Live during sessionsLinking climate indicators to conflict and displacement records, to test where environmental stress actually registers in political outcomes.
In developmentArms transfers, military expenditure and national defence profiles, assembled into a consistent panel for cross-country comparison.
In developmentCorpus Labb is an independent research workshop working at the intersection of international relations, security studies and computational methods.
The work is deliberately unglamorous: locating official sources, extracting what they contain without distortion, and making the result legible to other researchers.
Method & Access
Method. Sources are retrieved directly from the publishing institution, never from secondary compilations. Each dataset carries its own verification layer — internal sums, cross-period consistency, comparison against independently published totals.
Access. Datasets under active development are access-controlled while they are being validated.