Docs/Skills and code/skill script-cohort-analysis
.claude/skills/script-cohort-analysis/references/topic-prompt.md
Blind topic comparison: You'll see short, neutral topic cards (T1, T2, …). Each card describes only what a video is about. You haven't seen the videos and don't know how well they were made or how they performed.
Blind topic comparison
You'll see short, neutral topic cards (T1, T2, …). Each card describes only what a video is about. You haven't seen the videos and don't know how well they were made or how they performed. That's deliberate: you are judging the idea, not the execution.
The audience is described in your task file.
For each listed comparison "X vs Y", decide which topic more viewers in this audience would choose to watch if both videos were equally well made and packaged. Use your own judgment of this audience's interests. Don't apply a fixed formula. (The criteria are deliberately left open, so the topic measure doesn't bake in any hypothesis the analysis is testing.)
Answer "X" or "Y" (the card IDs as written, not "first" or "second"). Use "tie" only when you truly can't choose. Give a reason of 15 words or fewer. Some pairs appear twice in opposite orders. Judge each on its own; don't try to stay consistent with yourself.
Output one JSON file:
{
"judgments": [ { "x": "T3", "y": "T1", "winner": "T3", "reason": "…" } ],
"notes": "one sentence on what made topics strong or weak"
}judgments must contain every listed comparison, in the order given. Run the validation command from your task file until it prints OK. Your final message is one line: the path and the validator result.
Show the raw text
# Blind topic comparison
You'll see short, neutral **topic cards** (T1, T2, …). Each card describes only what a
video is about. You haven't seen the videos and don't know how well they were made or how
they performed. That's deliberate: you are judging **the idea, not the execution**.
The audience is described in your task file.
For each listed comparison "X vs Y", decide which topic **more viewers in this audience
would choose to watch** if both videos were equally well made and packaged. Use your own
judgment of this audience's interests. Don't apply a fixed formula. (The criteria are
deliberately left open, so the topic measure doesn't bake in any hypothesis the analysis is
testing.)
Answer `"X"` or `"Y"` (the card IDs as written, not "first" or "second"). Use `"tie"` only
when you truly can't choose. Give a reason of 15 words or fewer. Some pairs appear twice
in opposite orders. Judge each on its own; don't try to stay consistent with yourself.
Output one JSON file:
```json
{
"judgments": [ { "x": "T3", "y": "T1", "winner": "T3", "reason": "…" } ],
"notes": "one sentence on what made topics strong or weak"
}
```
`judgments` must contain every listed comparison, in the order given. Run the validation
command from your task file until it prints OK. Your final message is one line: the path
and the validator result.
Shown as it is in the repository at build time (2026-11-03); the home directory appears as <home>. If this page and the file differ, the file is right. Back to the catalogue.