1. Establish the two works
The source work and screen version are identified separately. Source identity and factual metadata are reviewed before they are treated as stable reference information.
2. Research the relationship once
The Adaptations Studio uses one comprehensive research package to gather the supportable source-to-screen relationship: overall premise and coverage, story, characters, relationships, events, world and setting, themes, structure, tone, endings, evidence, and optional chapter/volume/arc-to-episode mappings.
The system does not create separate Reader Path summaries, Source Experience essays, Screen Experience essays, verdicts, rankings, recommendation copy, or fidelity scores.
3. Preview before import
Research output is previewed before it becomes project data. Source facts, orientation, comparison records, evidence, and mappings can be inspected, and proposed records can be deselected when they are unsupported or unnecessary. AI output is never treated as evidence merely because an AI produced it.
4. Store individual comparison records
Accepted research is organized into structured comparison records. A record keeps the source detail, screen detail, change type, likeness, significance, spoiler scope and level, confidence, supporting evidence, and—when useful—what one version uniquely contains.
That structure lets the public app regroup the same reviewed facts without rewriting them.
5. Review every record
Comparison records are worked individually as Accepted, Needs Work, or Rejected. Proposed or Needs Work records remain unfinished. Evidence and source facts can also be verified independently. Integrity checks surface missing sides, unsupported claims, unresolved source identity, spoiler conflicts, duplicate records, and open reader corrections.
6. Publish only the approved projection
The backend keeps the research history, evidence, review state, unresolved issues, and internal notes. DDPublic receives only the published project and accepted public comparison data. A project can therefore remain published while its current canonical reprocessing or correction work is visibly incomplete in Admin.
7. Protect spoilers without hiding the size of the comparison
Public spoiler controls use each record's spoiler scope and level. When protection hides reviewed records, the page can show visible / total counts—such as 3 / 4—so readers know additional comparison material exists without seeing the protected content. Show Spoilers reveals the full reviewed set.
8. Display the same data many ways
The default public view is Content Comparison. Optional reader lenses change ordering and spoiler visibility for drama watchers, source readers, or people new to both. Topic filters, change filters, “what each version gives you,” source reference information, and reader lenses all come from the same canonical project.
9. Reprocess and correct without duplicating the project
Older Adaptations can be reprocessed through the current research workflow. Completion reflects whether that current research and review work is actually finished. Public Report Correction submissions can be routed back into Integrity as review work, researched, resolved, and then republished from the same canonical dataset.
Governing principle
Research once. Store once. Review what matters. Display many ways. Adaptations should be comprehensive and orderly underneath while remaining easy to explore on the public site.