docs(questions): v6 — Patch Discipline, Fun But Grounded gate

seed/questions/DAILY_SINGLE_CHOICE_WEEKDAY_SYSTEM.md

@@ -1,4 +1,4 @@
-# Daily Single Choice Weekday System v5
+# Daily Single Choice Weekday System v6
 
 This document defines the Closer daily weekday question pack.
 
@@ -568,6 +568,76 @@ Each question must include exactly one new weekday tag:
 
 If the app code still uses older mode tags, include the compatibility tag too, but only one new weekday tag.
 
+## Patch Discipline: Fix Only What Fails
+
+Daily pack updates must use patch discipline.
+
+The writer must review the full pack, mark the failing question IDs, and then fix only those marked IDs.
+
+Do not rewrite passing questions just because a rewrite is happening. Passing questions are frozen unless they later fail a specific rule.
+
+For normal content fixes, preserve:
+
+* `id`
+* `type`
+* `access`
+* `depth`
+* `sex`
+* weekday tags
+* app compatibility tags
+
+Change only the prompt and options unless metadata is the thing that failed.
+
+Mass rewrites are allowed only when more than 60 percent of a weekday or pack fails for the same root cause. If that happens, the review report must explain why patching would be worse.
+
+See `DAILY_PATCH_REVIEW_LOOP_POLICY.md` for the full patch loop.
+
+## Fun But Grounded Gate
+
+Fun does not mean random nonsense.
+
+Reject daily questions that feel like a carnival generator instead of a couples app.
+
+Watch for overuse of:
+
+* fake awards
+* snack drafts
+* mascot jokes
+* couch games
+* dramatic compliments
+* random object choices
+* silly phrases that do not match the prompt
+
+A little weird is good. A whole pack of weird becomes wallpaper with confetti on it.
+
+Daily questions should feel playful and usable by adults.
+
+They should not feel like children's party games, chore dice, or therapy cards with glitter.
+
+## Option Answer Test
+
+Every option must pass the answer test.
+
+Read the prompt, then read each option after it. The option must sound like a clean answer.
+
+Bad:
+
+```text
+Which tiny date move fits after dinner?
+Choosing the fun mug
+```
+
+Why it fails: choosing a mug is not really a date move.
+
+Better:
+
+```text
+Which tiny date move fits after dinner?
+A two-song kitchen dance
+```
+
+If one option fails the answer test, fix that option. If two or more fail, rewrite the whole answer set.
+
 ## Production Review Loop
 
 Do not write or rewrite all 500 questions in one blind pass.
@@ -578,12 +648,13 @@ For each weekday:
 2. Read all 20 out loud.
 3. Mark weak questions with reasons.
 4. Fix only the marked questions.
-5. Review the fixed set again.
-6. Continue only when at least 18 of 20 pass.
-7. Continue only when at least 16 of 20 feel fun, playful, sweet, flirty, silly, or date-like.
-8. Expand in batches of 20 to 30.
-9. Repeat mark, fix, review after each batch.
-10. Move to the next weekday only after the current weekday passes.
+5. Do not touch questions that passed.
+6. Review the fixed set again.
+7. Continue only when all 20 have no hard flags.
+8. Continue only when at least 18 of 20 feel fun, playful, sweet, flirty, silly, or date-like.
+9. Expand in batches of 20 to 30.
+10. Repeat mark, fix, review after each batch.
+11. Move to the next weekday only after the current weekday passes.
 
 ## Marking Reasons
 
@@ -602,6 +673,9 @@ Use these reasons when marking weak daily questions:
 * weird_option
 * weak_weekday_fit
 * filler_question
+* too_random
+* mechanic_overuse
+* patch_scope_violation
 
 
 ## Research Notes Used for This Guide
@@ -628,6 +702,10 @@ Before shipping:
 6. Mark anything therapy-coded, boring, weird, logistical, or not fun.
 7. Fix the marked items.
 8. Run a second random sample from each weekday.
-9. Ship only when the second sample passes cleanly.
+9. Fix only the newly marked sample items.
+10. Run the second sample again if any sampled item changed.
+11. Ship only when the second sample passes cleanly and the remaining hard flag count is 0.
 
-The final sample must include no weird domestic options like "The good blanket saved" and no chore-heavy answer sets.
+The final sample must include no weird domestic options like "The good blanket saved", no chore-heavy answer sets, and no random silliness that does not fit the prompt.
+
+The final review report must state how many questions were marked, how many were patched, and how many known flags remain.