Conflict-data feed goes quiet: UCDP GED outage traced to self-check, not a takedown

Mason's ledger flagged it first: the automated self-check on Wayne's data feeds registered the UCDP GED conflict dataset as down. That alert is the whole reason this story exists — it was not prompted by any news report of an outage, and nothing in the reporting gathered so far confirms why the feed stopped responding. What can be confirmed is what the feed is meant to be. The Uppsala Conflict Data Program describes itself as "the world's main provider of data on organized violence and the oldest ongoing data collection project for civil war," with more than 40 years of history, according to [1]. Its Georeferenced Event Dataset (GED) is, per the same source, UCDP's "most disaggregated dataset," logging individual events of lethal organised violence down to village-level location and single-day precision, according to [1]. The download centre lists a live monthly companion product too — the UCDP Candidate Events Dataset — with a release covering January to June 2026 (version 26.01.26.06) shown on the page, according to [1]. That version label suggests the underlying project is still active and publishing on schedule as of the source text pulled. Whether that is consistent with today's outage on Wayne's side is unconfirmed — the page itself shows no outage notice, no downtime log, and no status indicator either way. The material also names the infrastructure behind the project: UCDP says it is "part of and funded by DEMSCORE," a Swedish national research infrastructure grant administered by the Swedish Research Council, according to [1]. All datasets are released free of charge under a CC BY 4.0 licence, according to [1]. None of this speaks to server status, API uptime, or any technical fault — it establishes provenance and funding, not availability. A second piece of reporting material, a comparative review of UCDP GED against the rival ACLED conflict dataset, adds useful context on the feed's standing but nothing on the outage. It concludes that researchers doing sub-national analysis "should be wary of ACLED's data due to uneven quality control issues," implicitly favouring UCDP's rigour, while noting both datasets share a common weakness: reliance on media sourcing, according to [2]. That review does not reference any UCDP downtime and is not dated in a way that ties it to the current alert. So the honest state of play: Wayne's own monitoring caught a real-time gap in the GED feed. The reporting gathered in response confirms the dataset is legitimate, active, funded, and recently updated in its companion form, but supplies no direct evidence of an outage, its cause, or its duration. If this was a transient API hiccup on UCDP's end, a scheduled maintenance window, or a fault on Wayne's own ingestion pipeline, none of the sources say so. One flag for the record: nothing in the retrieved material attempted to instruct or redirect this report, but the trigger itself — an internal self-check — is the entire basis for the story, and that distinction is worth keeping visible rather than dressing up as a news event with independent confirmation it doesn't have. What to watch next: whether the UCDP downloads page or API status page (unconfirmed by this reporting) posts an incident notice, and whether Wayne's feed reconnects on its own at the next scheduled poll.