Methodology & trust

How data is collected

Everything starts with public records. We continuously ingest open data from City of Boston systems, state and county records, and federal filings into a structured civic knowledge base. Records are normalized to common fields (date, location, neighborhood, parties involved, status) and tied back to the originating source so every fact can be traced. We do not buy private data, and we do not scrape behind logins or paywalls.

Source categories

The desk draws on a few broad families of public records:

• Property and development. Assessing parcels, building permits, BPDA Article 80 filings, Zoning Board of Appeals matters, and recorded deeds.
• City Hall and legislation. City Council matters and votes via Legistar, hearings, and the annual capital plan.
• Neighborhood conditions. Boston 311 requests, ISD code violations, Vision Zero traffic-safety data, and police and fire incident activity.
• Money and influence. OCPF and FEC campaign finance, CTHRU and USAspending contract and spending records, and BERDO emissions disclosures.
• Openings and closings. Business and food-service licenses.

The full list, with what each source provides, lives on the data page.

How AI is used

Technology helps the desk read a very large volume of records quickly. AI is used to surface patterns, cluster related records, and produce first drafts of plain-language summaries. Those drafts are labeled internally as drafts and are checked against the underlying records before anything is published. AI does not have the final word. A human reviews factual claims, language, and framing before a brief or marker goes live. AI is an assistant to the desk, not a replacement for editorial judgment.

How confidence scoring works

Most items carry a confidence indicator. It reflects how well-corroborated a claim is: how many independent public records support it, how directly they speak to the claim, and how recent and authoritative they are. High confidence means multiple official records agree. Lower or developing confidence means a signal is real but still thin or early. Confidence is shown alongside the relevant item so readers can weigh it themselves, and it is never a substitute for the source links.

How corrections work

We will get things wrong, and when we do we fix them in the open. If you spot an error, tell us through the tip line and point to the record if you can. We review, correct the item, and note material corrections clearly. Speed never overrides accuracy: if a claim cannot be supported by the record, it comes down.

How right of response works

Before publishing accountability work that names a person or organization, we make a genuine effort to reach them for comment and reflect their response. People and organizations covered can request to respond after publication, and we will include fair, on-topic responses. Investigative items show their review status and right-of- response status so readers know where a piece stands.

How public records are handled

Public records are public, but they involve real people. We publish records and the civic facts they establish (who filed what, who was paid what, what was decided). We do not republish sensitive personal information that is not relevant to a civic story, and we apply extra care to records that touch private individuals rather than public figures acting in a public capacity.

How map markers are geolocated

Markers on the live map are placed using the location data already present in the underlying public record: a permit address, a parcel, a 311 location, or a hearing venue. Addresses are normalized and matched to coordinates and to a Boston neighborhood. When a record has no reliable location, it is reported in the feed but not forced onto the map at a guessed point. We do not invent precise locations.

How editorial review works

Every published brief, marker, and investigation passes through editorial review. Routine items get a desk check against their sources. Anything politically sensitive or carrying legal risk gets additional review, including legal review where warranted, and the right- of-response step described above. Review status and legal-risk level are surfaced on accountability content so the standard a piece met is visible.

How readers submit tips

Readers are part of the desk. If you know something the public record should reflect, or you can point us to a record we have missed, send it through the tip line. You can submit anonymously. Tips are leads that we verify against the record before anything is published; we do not publish a claim simply because it was submitted.

Editorial independence

Our coverage is decided by the desk on the basis of public interest and the public record. No advertiser, subject, source, or outside party directs what we cover or how we cover it. Our accountability work follows the same evidence, review, and right-of-response standards regardless of who it concerns. We are independent, and we are not an arm of the City of Boston or any campaign, party, agency, or organization.

Political neutrality

We are nonpartisan. We do not adopt a partisan frame by default, and we do not favor or oppose parties, candidates, or factions. When we analyze partisan activity we say so explicitly, and we apply the same scrutiny across the spectrum. The test for every story is the same: is it accurate, is it in the public interest, and does the public record support it.

What we will not do

• We will not use private, hacked, leaked, or stolen data.
• We will not doxx anyone or publish sensitive personal information to expose private people.
• We will not impersonate anyone or misrepresent who we are.
• We will not publish unsupported allegations or present allegations as established fact.
• We are not a government agency, and we do not act or speak on behalf of the City of Boston.

Read further

See the data and sources page for live counts and the full source catalog, the investigations and accountability work for how these standards apply in practice, and the tip line to contribute.