The MVP Scoper

Help me define the minimum version worth building to learn the most.

The full vision: [DESCRIBE]
What I want to learn from the first version: [DESCRIBE]
My constraints: [TIME, MONEY, SKILLS]

Provide:
1. The one core thing the MVP must do to test the key assumption
2. What to ruthlessly cut from v1 (and why it can wait)
3. Whether I even need to build software, or a "concierge"/manual version
   would teach the same thing faster
4. The signal that would tell me to build more vs stop
5. The trap of building too much before learning

Bias me toward the smallest thing that produces real learning, not the
most impressive demo.