Editorial standards

Stateside Daily doesn't write or republish articles. Our editorial work lives in the analysis layer: deciding which publishers to ingest, classifying their political lean and reliability, grouping stories that cover the same event into comparison clusters, and ranking what surfaces on the home page. This page documents how each of those decisions is made.

1. Source selection

We aggregate from established news organizations that publish via public RSS, meet a minimum editorial standards bar, and add coverage breadth not duplicated by other sources we already track. New sources are added by the editor; reader nominations go to corrections@statesidedaily.com.

The bar a candidate publisher must clear:

• Editorial masthead. A publicly named editor or editorial board, accountable for what runs.
• Bylined reporting. Articles attributed to a named author, not anonymous content farms.
• Corrections discipline. Visible corrections policy and a track record of issuing corrections rather than quietly editing or deleting.
• Original reporting share. At least 30% of content is original reporting, not exclusively syndication or aggregation. (Pure aggregators are excluded — that's our niche.)
• RSS hygiene. Working RSS feed with sane item counts and correct timestamps. We re-audit feed health weekly.

Sources that demonstrably fabricate, persistently mislabel opinion as news, or have failed integrity audits at major independent monitors are not included. The full list of sources we currently ingest is published at /sources.

2. Bias classification

Every source carries a political-lean classification on a five-point scale: left, lean-left, center, lean-right, right. We also track a separate reliability rating (high / mixed / low) covering sourcing discipline, factual accuracy, and willingness to correct.

Classifications are based on three independent inputs, in this order of precedence:

1. Independent media-bias monitors. AllSides, Ad Fontes, NewsGuard, and (where available) academic longitudinal studies. We use the consensus rating where monitors agree; we average where they disagree.
2. Editorial board endorsements. Stated political endorsements from the publisher's editorial board, where declared.
3. Internal sampling. A reading of 30 randomly sampled stories from the source's last 90 days, scored on framing language, sourcing breadth, and topic selection.

The classification appears on every cluster page next to the source name as a colored chip. The aggregate distribution across a cluster is visualized as a spectrum bar so readers can see at a glance whether a story is dominated by one lean.

Classifications are reviewed quarterly. Reader disputes (with evidence) are reviewed within 7 days; see corrections.

3. Cluster algorithm

When the same event is covered by multiple publishers, we group the coverage into a cluster. The cluster page surfaces every source's headline side-by-side. Clustering is a two-pass process:

1. Cluster ID match. A hash of the lead noun phrase + key entities is generated for each headline. Headlines with matching hashes are grouped immediately.
2. Title signature dedupe. Across remaining headlines, we extract the top-5 significant tokens (sorted) and merge clusters whose signatures overlap by ≥80%. This catches stories with slightly different phrasings that the cluster-ID hash split.

Stories from the same publisher are de-duplicated within a cluster; we surface only the canonical version per outlet.

4. Ranking

The home page is not a single editor's pick. Stories rise on a hot-score that combines:

• Cluster size — how many independent publishers are covering this. More sources = larger story.
• Recency — newer headlines decay slower.
• Source weighting — high-reliability sources contribute more weight than mixed/low ones, but we include all rated sources to avoid suppressing alternative views.
• Click signal — anonymized aggregate click rates feed into hot-score with low weight; we don't want a handful of viral clicks to dominate.

We do not rerank for political balance — the spectrum bar is the transparency mechanism, not a thumb on the scale.

5. What we don't do

• We do not republish article bodies. Our summaries are short fair-use excerpts (typically < 50 words) with a link to the original.
• We do not write or rewrite news content. Our work is classification and surfacing, not authorship.
• We do not generate AI imagery for news context. Images come from publishers' canonical metadata.
• We do not accept payment for placement, classification, or ranking. Advertising is via standard Google AdSense slots, clearly labeled.
• We do not endorse candidates, parties, or causes. We have no opinion column.

6. Disputes & corrections

If you believe we've misclassified a source, broken a cluster (joined unrelated stories or split related ones), or ranked something inappropriately, write to corrections@statesidedaily.com. We respond within 7 business days. Sustained patterns of misclassification across multiple readers trigger a re-audit. See corrections policy for full SLA.

7. Independence

Stateside Daily is an independent operation. We are not part of any media conglomerate. We are not funded by political organizations, candidates, or PACs. Our revenue comes entirely from Google AdSense, displayed in clearly-labeled slots between content. We accept no money for editorial placement, classification, or ranking.