FICTIONAL EXAMPLE — not a real person, not real corpus. Part of the examples/ worked corpus (persona: Sam Rivers). See examples/README.md.

When I've polished past the point of learning, I ship the rough version and let real players teach me, even when it's embarrassing.

State it as a rule I can follow and fail. The trigger is specific: past the point of learning — when another day of solo polish tells me nothing a player wouldn't.

Why

My worst months were spent perfecting games nobody played. Shipping rough is how I convert opinion into evidence. The cost is looking unfinished in public; the protection is that I stop wasting weeks on guesses.

Evidence

Concrete, dated instances of me actually living (or failing) this.

• 2026-02-15 — Shipped a 48-hour game-jam build with placeholder art rather than spending the weekend on a menu nobody asked for. Two strangers' feedback reshaped the core loop in a day. See reflections/evidence/2026-02-15-jam-game-shipped.md.
• 2025-11-03 — Failed this: sat on a demo for five weeks "until the lighting was right." Wishlists were flat the whole time. The lighting changed nothing.

Falsification

If shipping rough repeatedly produced noise instead of signal — players bouncing on bugs before they reach the idea — this moves to contested. One embarrassing launch that teaches nothing is not falsification; a pattern of them is.

Tensions

Collides with a quiet pride in craft. Under deadline pressure I resolve it toward shipping; the principle exists precisely because my default is the other way.