Aluminum Polymer

Electrical failure of ceramic caps is essentially nonexistent. When speaking of ceramic failure, that's precisely what anyone else means: mechanical damage.

Some products don't sit in quiet, stable racks all their lives...

Tim

MLC ceramics have tiny sandwich-layer features, and manufactured-in flaws DO short them, not infrequently. By the time you see the failed unit, there's so much secondary damage that you might think that damage to be the cause (even when it isn't).

Oh, no, it's certainly not extremely rare for MLCC caps to short. It's the normal failure mode when the board flexes under them. It's particularly bad for larger caps (1206, 1210, and above). We're required to use series caps, mounted at right angles to each other, when the source isn't current limited (enough to prevent smoke).

Well, we don't see them. We'd usually miss an open or a wrong value, being that most of our caps are bypasses.

Bypass caps are the largest problem, particularly those on the power inputs. Coupling caps into power amplifiers are the other real problem (DC offsets are bad).

Your operating environment is pretty benign, obviously. It is a serious problem with ceramic caps, enough so that specialty caps with "soft (flexible) termination" are made to mitigate the problem. They're quite expensive, though.