Five-viewer validation — EIGHT SECONDS OF DUSK

Purpose

Test the riskiest claim: a first-time viewer can complete the 48-second work, understand that their hold affected what remained, feel something specific without explanatory copy, and want to share or support another work.

Participants

• Five opted-in adults who were not involved in production.
• At least three already consume digital art, animation, narrative games or interactive fiction.
• No incentives in the first test; disclose that the work is AI-assisted and unfinished.
• Operator-mediated local browser or private one-to-one screen share only. No public URL or social posting.

Procedure

1. Say only: “This is a 48-second interactive film. Headphones recommended. Please use the pointer if the work asks you to.”
2. Record whether the viewer starts, reaches 00:48, moves the boundary during seconds 20–28, replays, captures a frame, or asks for a link.
3. Ask without suggesting answers:
   • “What happened?”
   • “What did your action change?”
   • “What feeling or memory remained?”
   • “Would you send this to one person? Why or why not?”
   • “If the next work were free, would you subscribe or follow?”
   • “Would you voluntarily support the next work at ¥100, ¥500, neither, or another amount?”
4. Store only anonymous participant labels P1–P5 and aggregated responses. Do not collect names, email addresses or sensitive data.

Success and failure

• Success: 4/5 finish; 3/5 independently describe encounter, separation and something remaining; 3/5 recognize their hold affected the result; 2/5 would share; 2/5 choose any support amount or 3/5 request another work.
• Failure: fewer than 3/5 finish, fewer than 2/5 understand the action, or no viewer would share/follow/support.
• Stop rule: do not polish during the five sessions. Complete all sessions before interpreting results.

Decision after test

• Success: fix only repeated usability/audio issues, then reassess public-release quality.
• Mixed: one focused iteration on the single most repeated failure, then retest with three new viewers.
• Failure: stop production and compare a different work concept or business model rather than extending this prototype.

Anonymous result sheet