I think Postel's Law is best understood as a growth strategy. Its tolerance facilitates easy composition but at the expense of some protocol decay. What the author does not acknowledge is that less tolerant designs schemes are often so exacting that they discourage participation.
I have to reluctantly agree with this (the noisy failure position) much as I like and have followed jon's maxim for decades.
Right now, for me, the worst "offender" (I.e. "faithful observer") is USB type-C. Hook up two high speed thunderbolt devices using a cable that supports only USB 2.0 speeds and it will "work", but slowly. There's currently no way to tell which cable is which by visual inspection.
I can see why this is a triumph of engineering and back compatibility, but I can also see why it's a problem.
