TVSs are fine for what they're fine for... a common place I see them is a voltage clamp in a switching supply. Instead of an RCD snubber or clamp, they use a 200V (or so) TVS, which sees pulses up to one amp or so, and not too much average power. That's a perfectly fine application for them -- it's basically heat that fuses the silicon chip, so as long as the average power and pulse energy aren't too much, it'll last forever.
There isn't much silicon inside the things, and silicon is expensive, so they blow quickly under any kind of huge stress, like expecting them to handle power line transients.
The other crappy part about any kind of TVS is the curve. A typical 1.5KE15 for instance is designed to protect 12V supplies, so (at 25C) it's guaranteed to not conduct (much... 1uA) at 12.8V, start drawing significant current by 14.3V (min) to 15.8V (max), and clamp 21.2V (max) during a pulse of 71.7A (peak). So your circuit needs to withstand say 24V to survive that kind of fault.
Same goes for MOVs; a 240VAC device might be rated 370V (at the "significant current" point) and clamp a big pulse in the 500-600V range. The difference is, MOVs are massive hunks of ceramic, so they can withstand gangbusters energy spikes without failure. (They're notorious for failing anyway, because too-small devices tend to be used. Damage is cumulative, with an exponential weighting, so you can use an MOV like a zener diode practically forever, but you'll only get a few zaps in the 100A+ range before even a large one gives up.)
But even if your switcher shuts down at that voltage (overvoltage limit..?), it has to withstand all that voltage, too. So you can easily get fried transistors despite best intentions. But hey, we're talking, like, lightning strike or something, so if it escapes without too much vaporization (besides the MOV), that's not doing too bad.
Don't worry about consumer stuff; it isn't designed to be repaired, so they don't bother with it at all (only a fuse, or fusible resistor, for required safety protection).
Tim