reader-trial.md

Reader trial protocol and results form

This short protocol evaluates whether a newcomer can run and understand the document-job journey. It is a template for future readers; no external participants or timings are reported here.

Ask each participant to use a fresh scratch directory and the exact package under evaluation. Provide the link to the document first-run tutorial and no additional walkthrough. Ask them to:

1. create the synthetic source and configuration;
2. run process, inspect-job, and verify-job;
3. explain what completed means and whether it implies review;
4. run the interruption and resume section;
5. apply the review record tied to a page and output, then identify the still-pending pages.

Record where they hesitate, which file they use to answer each question, and which command or sentence needs clarification. Do not ask them to call a network service or describe the synthetic reviewer as an independent person.

Results form

Field	Entry
Package version and source SHA	pending
Package mode (wheel/sdist/source)	pending
Participant identifier	pending
Date and timezone	pending
Completed process without help	pending
Correctly separated completed from reviewed	pending
Recovered without duplicating page 1	pending
Correctly rejected stale receipt	pending
Correctly identified pending review pages	pending
Commands or wording that caused hesitation	pending
Observed time, if voluntarily recorded	pending

Do not combine these fields into an accuracy claim. A reader trial measures whether the workflow and explanations are usable; it does not establish that the transcription is correct.