{
  "authorship": "Five self-authored fictional cases, not vendor results or actual observations.",
  "cases": [
    {
      "id": 1,
      "text": " 1. A background track starts on page entry   Fictional setup:  The imaginary Lantern page contains a lyric video with a play button and a separate ambient loop configured to start on page entry. Its proposed handoff says the video starts only on request. That describes one component while missing another. Assume for this exercise that the loop would continue beyond three seconds; no actual recording was timed.   Authored decision:  Split the handoff into two rows. For the video, record the intended play action. For the loop, record page entry as the trigger and identify its proposed independent control. Prefer removing the automatic trigger and presenting a clearly labeled start action before the visitor hears anything.   Review record:  Write expected state as loop silent until requested, with observed state pending. Add a follow-up to enter the public page and identify every audible source. A later reviewer should name the revision and actual result. The fictional proposal cannot justify a passed review. If the implementation has no separate loop, remove that assumption and document the real source inventory instead of treating the example as evidence. "
    },
    {
      "id": 2,
      "text": " 2. A muted video sits beside an audible embed   Fictional setup:  On the invented Harbor page, the hero video is muted but a lower preview panel has its own audio track. The designer adds a mute icon beside the hero and describes the whole page as muted. The players are separate sources in this exercise; their implementation and actual playback have not been tested.   Authored decision:  Replace the page-wide assumption with a source map. Name the hero and preview independently, and associate controls with their sources. If a shared control is proposed, specify that both sources must respond. Visual grouping does not demonstrate shared behavior.   Review record:  Prepare separate observations for page entry and control activation. The useful question is which source remains audible, not whether an icon changes appearance. A quiet hero does not settle the preview row. Leave observations pending until someone inspects the identified page. Retain an unresolved source as a separate task rather than merging it into an optimistic overall result. If the second player is offscreen, keep its location in the record so the reviewer can find it. "
    },
    {
      "id": 3,
      "text": " 3. The only instruction is to lower system volume   Fictional setup:  The imaginary Fern page offers no source control for its automatic soundtrack. A help note tells listeners to use computer volume keys. A reviewer proposes marking this sufficient because the music can become silent. That conclusion ignores the difference between controlling page audio and changing all computer sound.   Authored decision:  Reject that proposal as evidence of independent control. System mute affects other audio the visitor may need. Specify a source pause or stop action, or an independent volume control that can silence the source while leaving overall system volume available. This is a requirement for a fictional revision, not a claim that the control exists.   Review record:  Add the chosen control and intended scope. A later observation should describe the source response and whether system volume was unchanged. Do not invent a successful screen-reader test or infer compatibility from a slider. Keep such checks outside the evidence until performed. Record any implementation limitation plainly, including a control that only changes an icon or a slider whose actual target remains unclear. "
    },
    {
      "id": 4,
      "text": " 4. A hidden player restarts on return   Fictional setup:  The invented Orchard page lets a visitor stop its background player. Its navigation opens a lyrics section and then returns to the main panel. For this exercise, returning reinitializes the player and loses the stopped state even though the visitor did not intentionally request music again.   Authored decision:  Treat the return path as another trigger. Identify which component owns playback state and specify the intended return state. A control that works once does not describe the entire journey. Prefer a return state that waits for a deliberate, clearly labeled start action.   Review record:  Write a sequence: enter, stop the source, visit lyrics, return, observe. Record each step rather than only the first stop. If implementation differs from the fictional assumption, preserve the real observation. Reloading and navigation can deserve separate rows because they may create different player instances. Do not assert that every browser restores or resets playback in the same way. Save the environment and path actually used, and leave other paths pending until reviewed. "
    },
    {
      "id": 5,
      "text": " 5. The visitor intentionally starts a labeled preview   Fictional setup:  The imaginary Meadow page has a button labeled Play song preview with sound. No audio is intended before activation. A reviewer groups it with automatic-start examples because playback begins immediately after a click. That grouping loses the intention communicated by the label.   Authored decision:  Record the intentional activation separately from page-entry audio. The label makes the consequence clear in this fictional design. Do not use that distinction to excuse an unrelated background source that starts simultaneously or to claim the whole player is accessible. Keyboard operation and understandable controls remain separate checks.   Review record:  Save the proposed label, starting state and affected source. Ask the reviewer to observe whether any sound starts before activation and whether the action starts only the expected preview. Mark both observations pending. This case demonstrates classification, not behavior across browsers. Keep a vague button label as a revision task rather than assuming any click expresses a wish to hear music. Preserve the actual label in later evidence so another reviewer can assess the same interaction. "
    }
  ]
}