One run, or a pass rate?
One run, or a pass rate?
One green run feels safe to ship. Watch what it hides.
▶ Run the case
↻ start over
the case: fix the broken date
✓
Ship it ✓
You shipped on a coin you flipped once.
One run
is a fact about one run.
Ten runs
are a fact about the agent.
An agent is a distribution, not a function.
Grade the distribution.
↻ watch again