Kind of depends on your criteria a bit I suppose.
Many System 1 machines mimicked the EM equivalent for functions. Any bugs on these?
I've always felt the software from Larry Demar seemed pretty solid. Often taking into account situations that one might not normally consider, and then having software accomodate for various and potential mechanical malfunctions. A bit of a segue, but if you are looking at situations where you could intentionally try to mess with the mechanical switches, in ways which would not normally occur or be anticipated, this could be the way to create some odd software situations which some might consider to be bugs.. but the Demar code seems to somehow think of everything.