I don’t even think we disagree?
The instrumentation is accurate when sited and functioning properly and maintained. Unfortunately it’s a struggle to accomplish those things sometimes. Those Vaisala sensors have accuracy greater than 0.1C, but it doesn’t mean a lot if your siting doesn’t accurately represent the 2m temp.
But we’ve called out bad ASOS data here a lot…CON, BOS, ORH, BDL, DAW…they’ve all had issues time to time. It’s usually easy to pick out the error on MADIS charts because most of them have a step change.