Answers about what this archive holds, how its records are recovered and dated, and how to use, cite or correct them. Each answer restates what the records and the methodology statement already establish; where something is not established, the answer says so instead.
It is an independent preservation project that holds and presents the surviving record of Free Press Houston, an independent Houston newspaper published from 2003 to 2018. The archive is a separate organisation from the paper: it did not publish the newspaper, and the paper ceased publishing in 2018.
Material is presented as primary source material — covers, issues, recovered articles, photography, illustration, artwork, ephemera and contributor records — each with the evidence for what it states.
Free Press Houston was founded by Omar Afra, who published and edited the paper across its run from 2003 to 2018. His contributor record holds the issues, mastheads and bylines the archive has recovered for him.
The span is 2003 to 2018, the paper's full run. Holdings include digitised issue objects, recovered article texts, cover and poster artwork, photographs, ephemera such as maps and passes, transcribed mastheads, contributor records, festival and event records, and versioned datasets generated from those records.
Counts of what is actually held — digitised issues, content-complete issues, recovered articles — are printed on the About page and regenerated from the data rather than stated in prose.
Much of fifteen years of writing and artwork was dispersed when the paper collapsed in 2018, and the material that survives is being collected from many separate sources. The archive is incomplete and actively growing; issues, articles, images and records are added and reconciled on an ongoing basis.
Where a record is thin, the record says so. An article with no transcription is marked as untranscribed rather than presented with partial or machine-guessed text, and an uncorroborated date is left unstated rather than filled in.
Material is recovered from printed originals, press proofs held as PDFs, email attachments retained by the paper, and surviving pages of the paper's former website. Each recovered object is logged as a source record with its type, retrieval date and confidence before anything is published.
Original files are kept unmodified as the source of truth; web-facing images and page views are derived from them. Where the same article survives in more than one form, the versions are reconciled into a single record with several sources rather than duplicated. Every record carries a provenance state and a confidence level: confirmed, probable or unverified.
Software, including machine-learning tools, assists with discovery, text extraction, de-duplication and name matching. A match a tool proposes is not a fact the archive states: attributions, identity joins and publication decisions are reviewed by a person against corroborating evidence.
Original publication dates are preserved; a date of digitisation or migration is never presented as a publication date. Each article record is in one of three date states: issue-corroborated, where the piece is bound to a held issue and the issue's month governs; corroborated, where a day-precision date is established from a dated address, byline date, correspondence or printed page; or undated, where nothing could be corroborated and no publication date is emitted. Issue records carry month precision only.
Dates shown on the paper's former website are not reliable. The archive records reconciled dates and prints the basis for each one on the record under Date evidence.
Each contributor has one canonical record at /contributors/{slug}, and every byline, credit and staff listing recovered for that person resolves to it. Names are recorded as printed on the held material; where an uncertain spelling survives, it is kept as found rather than silently normalised.
Roles and relationships are only stated to the extent the evidence supports them. Where a role is known from the publisher's own account rather than from a held document, the record says so on its face.
Full-text search covers recovered articles, contributors, issues and records. Browsing is by issue index, covers, articles, collections, places, events and contributors; a collection page acts as a finding aid, listing the articles, editions and objects it holds.
Every article, issue, contributor, collection and object record is also served as JSON at the same address with .json appended, carrying the same facts as the page.
Both festivals were produced out of Free Press Houston, and the archive holds the records it has recovered for them: edition records, printed coverage, artwork and ephemera, and staff listings taken from held rosters or stated by the publisher. They are presented as collections and event records within the paper's archive, not as separate institutions.
Record pages carry a Cite this record block that generates a human-readable citation and machine formats from the record's own metadata. The original publication date and your access date are kept separate: a record without a corroborated publication date is cited as undated.
Record addresses are stable. When a record is corrected, it is corrected in place and its URL does not change, so citations do not break.
Write to omarsafra@gmail.com with the record and what is wrong. For a photograph or artwork credit, evidence that establishes the attribution — an original file, a contact sheet or a printed page — is what allows the credit to be changed.
Corrections are made in place and entered in the public corrections log with the date, the record, the change and the reason.
Yes. Suggestions, corrections, missing work, old issues, artwork and photographs can be sent to omarsafra@gmail.com. Submitted material is treated like any other source: it is logged with its provenance, and what the archive states about it is limited to what the material itself supports.
