How our newswire works

What is published

The server checks official feeds hourly and selects up to six recent releases per UTC day, with a maximum of two per source. New entries must be no more than 48 hours old and must have a usable date, summary and accessible source page. A quiet news day does not create a requirement to fill a quota.

Each brief names its source, links to the original and distinguishes the source’s publication time from its first appearance here. Excerpts are reproduced with attribution, may be shortened, and have HTML formatting removed. No new facts, quotes or photographs are generated automatically. They are institutional statements, not independently verified reporting or recommendations.

Scheduled data analysis

Our Data Desk publishes original calculations, comparisons and charts using two documented formats. On Mondays after 10:00 UTC, it compares five currencies over the completed week using ECB reference rates. On Thursdays after 10:00 UTC, it checks Eurostat’s monthly unemployment dataset and publishes when a new common observation month is available for all 27 EU countries. The server checks due work at 24 minutes past each hour. The first issue of each format establishes the series when enabled.

Each analysis states its period, source, calculations and limitations. These are automated statistical reports, not interviews or investigations. Source snapshots are retained privately for reproducibility. Unchanged observations do not produce another article; revisions update the original with a note. Missing or invalid data delay publication. These original analyses are open to search indexing and included in the article and news sitemaps.

Scheduled explainers

Nine analyses and explainers on games, technology, business, trading and politics are scheduled for Tuesdays and Fridays from 23 October to 20 November 2026, at 10:24 UTC. They are prepared with AI assistance from dated primary sources, with original interpretation and labelled hypothetical calculations. Game features are source-based coverage, not claimed hands-on reviews. Each article identifies its sources and research date. Before first publication, the server checks availability and selected key facts on the primary source page; a failed check holds the article. This limited check does not independently verify the source or rewrite the article. Time-sensitive previews have an expiry and are held if their publication window is missed. Full texts remain private until publication. See the publication calendar.

Published explainers enter the ordinary sitemap, site search and RSS. They are labelled as explainers, not breaking news, and are excluded from the news-only sitemap. From 24 November the server continues with a fresh-data Steam comparison on Tuesdays and, from 27 November, a Treasury yield-curve report on Fridays. This recurring schedule has no end date; existing currency and employment reports also continue. Steam comparisons use a disclosed sample, US prices, native-platform flags and aggregate review counts; they are not hands-on tests. Treasury reports show original basis-point calculations, not forecasts or investment advice. Missing or stale data hold an issue; unchanged or previously covered records do not create filler articles. Hosting outages delay publication, with at most one overdue explainer released per day when service resumes.

Editorial maintenance and topic selection

The server checks article structure each hour, rotates source-link checks across the week and refreshes official-source topic suggestions daily. New prepared explainers are held when source checks fail, titles repeat, paragraphs duplicate or an unsupported claim of hands-on testing appears. These are limited automated checks, not independent verification of every statement. Changed sources are flagged for review rather than silently rewritten.

A private weekly operations report records publications, source problems, research candidates and a restore-copy check. Topic scores reflect freshness and practical relevance; they are not measured search volumes. New topics are research outlines and are not automatically promoted to finished journalism. The weekly reading list collects published original reports.

Continuing comparisons: selection and calculations

Steam editions inspect at most 32 entries from the featured new-release list, prioritising games also shown among top sellers. Eligible games have a listed release date within 45 days, a US price and native-platform flags, no mature-content descriptors, and at least 50 reviews in the selected query. We compare up to three previously uncovered games with the highest review counts in that sample; at least two are required. The result is a sample, not a ranking of all new games. Reviews include all languages and Steam purchases with Steam’s off-topic filter enabled. Positive share is positive reviews divided by positive plus negative reviews; it is not Valve’s rating algorithm or a representative player survey. We retain aggregate counts, not review text or identities. Prices are US-dollar snapshots, not local offers or lifetime cost.

Treasury editions use the latest complete official par-yield observation before Friday and one on or before seven days earlier. All five maturities must share both dates; a source gap above four days holds the report. Yield changes and the 10-year/2-year gap are calculated in basis points: one percentage point equals 100 basis points. Par yields are fixed-maturity estimates, not realised investment returns. The analysis identifies movements without inferring their cause or issuing a trading recommendation. Each edition retains its dated snapshot rather than silently rewriting history.

The recurring job retries a held issue no more than once per six hours. After an outage it prepares the latest eligible edition, avoiding a backlog of stale reports. Existing configuration, stored drafts and earlier articles are preserved. Publication status.

Approved sources

• NASA official feeds — space and science; NASA usage guidance.
• Eurostat — public statistical releases; © European Union, CC BY 4.0 and reuse conditions. Eurostat is not responsible for this presentation.
• European Central Bank — statistical releases; attributed reuse conditions. The originals are freely available.

Selection and limitations

Duplicate links, future dates, stale records, unsupported sources and incomplete entries are skipped. High-risk personal allegations and medical claims are excluded from automatic selection. Automated checks do not replace editorial judgement or guarantee a source’s accuracy. A source outage leaves the last successful edition available.

Syndicated brief pages are excluded from search indexing and the Google news sitemap. The indexable newswire collection helps readers follow the source material. Our existing reported articles retain their own URLs and indexing.

Corrections

When a source changes a recent excerpt, the brief can be updated with a visible note while keeping its original publication time. Older corrections and disputed claims require editorial review. Report an error or rights concern with the article address and supporting information.

Read the newswire → · Browse all reporting