But as said earlier hard to explain to grandma Jones, hell it is obvious from this topic that even EEs sometimes have no clue and start brabbling all kinds of nonsense from their statistical experiment with n=1. 
Oh, FFS no. If you have to explain how a particular lightbulb works to Grandma Jones, then you have failed. Your lightbulb has failed. Your product is a failure and it deserved to fail. It's as simple as that.
Lightbulbs are not esoteric engineering products intended for use by experts. It's lighting, as in the first or second household application of electricity. People have expectations for how a lightbub will perform and nobody who doesn't have a particular interest in lightbulbs should have to build up a working knowledge of the constraints and limitations of a particular trechnology or implementation in order to make decisions about it, especially if the product is being marketing as retrofit.
I don't see the relevance of the rest of your post. Do incandescents fail if they are switched too much? Sure. But people know that. And the incandescent manufacturers aren't making any new claims. Their basic claim is the same as ever: this lightbulb works like that last one and costs $0.80. Will any old bulb work in an attic that where it is used for 30 minutes a year? Sure. In fact, there is absolutely no reason to put an LED there since the total lifetime energy use is negligible.
You make an interesting but ill-conceived point about n=1, that people should not form judgments from their very limited experience. Though this is true, I've got some news for you: that's indeed how they do it. They're not going to consult government reports on lightbulb performance. They are not going to sift through Consumer Reports lightbulb reviews. They are not going to build up a spreadsheet collecting data from their friends and neighbors. They are going to buy one bulb off an endcap at the local hardware store for $8 under the promise that it will save them some money in the long run, and they will screw it into the next available socket. And if that bulb fails in a year or two, they're going to remember that outcome, and that will be the end of $8 bulbs for that person for a decade or more. (or until they have no choice)
That is how people make these kinds of decisions. Economists call this "bounded rationality" and contrary to being a sign of stupidity or laziness, it is a reasonable way to make decisions in a complex world where you don't have the time or resources to know about everything. Given this behavior, LED manufacturers would be smart to design product that are extremely unlikely to fail in the promised time. This means, for example, that if you want to make a claim of 20 years life, you might want a product with an MTBF several times that. It also means not shoving the restrictions and limitations of the bulb into a blob of fine print on the side of the box where it serves primarily to protect the company rather than inform the consumer.
Finally, in my case, I have had 7 LED failures in 6 sockets in just over 7 years. It's more than large enough N to know that there is something amiss between the label on the package and reality.
PS -- I don't like that I have to play the "anti LED" guy in a thread like this. I actually like LEDs and want them to work. There is a reason I keep buying them. They have qualities that I rather enjoy and in fact, I prefer the light output of them over incandescent. There is not a single incandescent bulb in my house today and only a few fluorescent remain in fixtures that have build in ballasts and are in locations that are difficult to access for replacement. But I have also been burned by LED retrofit bulbs repeatedly, and man, I can not I not stand fanboy bullshit from people who will tell me to my face that my direct experience is irrelevant or that worse, that I would not have this experience if I wasn't so damned stupid.