Provenance

How a reader is made

1. 1

   The source

   A work believed to be in the public domain in the market it is sold in — Canada first. The text comes from an archive (Project Gutenberg, Wikisource, Faded Page) and the exact source is recorded with the reader. Public-domain status, translations, illustrations and audio are checked separately, because rights differ by country and by asset.

2. 2

   The screen

   Each work is screened against HummingRead’s published Content Philosophy — apolitical, areligious, with exclusion filters — and given a ruling: use, use with an era note, or set aside. Sensitive history is ruled on title by title.

3. 3

   The adaptation

   The story is engineered to a band under a written canon: word window, sentence length, vocabulary load, section count, picture count, the pause questions and the Let’s Talk questions. The About page states the source, the author and the year of first publication, exact where known.

4. 4

   The pictures

   Pictures are made to a brief for each section in a house style for the band. Every picture is audited against its brief before release; a picture that fails is redrawn or the reader is held.

5. 5

   The checks

   An automated audit checks the reading level, the sentence and paragraph laws, the picture density, the layout and the print files. A sample is read by a person. A reader that fails is fixed in the record and rebuilt, never edited by hand.

6. 6

   The release

   A reader is released when its source, region, checks and files are all in order. Only released readers are counted on this site.

Readers are engineered from the original public-domain text with AI under HummingRead’s published canon and Content Philosophy, checked by automated audit and sampled by a person. Pictures are made with AI to HummingRead’s brief for each story.

The catalogue is built from thousands of screened source works and grows as readers are prepared and released; the number shown on the library page is the number ready today.