Difficulty / reward tuning (telemetry-driven)

[ryanabooth]

Tune gold rewards, decay rate, and capacity costs based on real usage data.

Why

Current tuning constants (FarmTuning, QuestTuning, StreakTuning) were picked by feel during the pivot. Real player behavior will reveal which are too generous or too punishing — e.g. if 90% of players never re-roll because cost feels too high, or if plots die too fast for casual users.

Blocker

Needs an analytics backend. Options:

• TelemetryDeck (privacy-friendly, ~$10/mo)
• Plausible Analytics (web-style, less app-shaped)
• Self-hosted PostHog
• Skip the backend, ship an in-app debug log that the user reads off TestFlight feedback

Decision is "what backend" first, then implementation is a few-day project (event taxonomy + emit at the right call sites + dashboard).

Acceptance criteria (post-unblock)

• Analytics backend chosen and wired
• Events: quest_claimed, quest_rerolled, plot_died, plot_replanted, capacity_upgraded, gold_earned, gold_spent (with dimensions: kind, gold amount)
• Dashboard answers: median re-rolls per user per day, plot death rate per goal type, cosmetic adoption rate, capacity-upgrade conversion
• At least one tuning constant adjusted based on data