The use of the phrase always seems so condescending, as if all complicated topics are so easy to understand that they can be explained to a random grandmother in a few minutes. When someone says this phrase, it's frequently code for "I don't understand the words coming out of your mouth--could you use less jargon?" But saying it flatly admits ignorance, and we monkeys can't have that--it'll lower our status! Especially if we're interviewing someone who is supposed to become a good subordinate if accepted into a position--we can't have someone smarter than us!
It's a weird exchange when the inferior (in the immediate sense) member of a pair is the one being condescending. It's a proud display of ignorance, a rude display of ignorance. Personally, I think rudeness should be restricted to those who have something to be rude about--e.g. the person who is trying to explain something while avoiding all the jargon-words that he's taken for granted 8 hours a day in his school or work. Phrases like "My grandmother could understand this!" or "My grandmother's dog already understands this and has moved on." are used by those in the knowledgeable position trying to teach, as a way of being rude as well as trying to say that the topic isn't so hard--any fool can learn it given the time.
Rather than talking purely in terms of knowledge bases, though, an alternate solution to our information exchange can come from the concept of inference levels. When someone doesn't understand something, you are speaking at a level higher than them--sometimes many levels. A talk at a science conference might involve details above the inference level of most people attending, but if it's a good talk, then the speaker will first step down to the common inference level and connect it to the one he goes on to talk about. (He won't go down too far, that can also appear condescending, but he doesn't need to if he assumes his audience are scientists in the field.) So this is what I suggest asking: "Please bring it down to my inference level." Sometimes this is impossible without dedicating a considerable amount of time to the subject--not even Feynman could explain electricity and magnetism to a layman with no knowledge. Others try to cheat and discuss rubber bands for field lines or expanding balloons for expanding space or a rubber sheet for gravity--all these are horrible.
Of course, I could be wrong. But I'm willing to change my stance if someone shows me their non-technical grandmother understanding the concept of lazy evaluation in programming languages, which means being able to come up with and use an "original" example (i.e. not one shown in the discussion/example material). Understanding does not mean nodding your head politely, though I don't have anything against that as often understanding will come later after more pieces of the mysterious puzzle are put together, and sometimes work just needs to begin. (Jeff Jonas uses the puzzle metaphor to describe how in big data systems, if your system is clever enough, more data makes you faster (while for many companies more data makes you dumber and you have corporate amnesia (I recently got a "car insurance quote" from Progressive even though I'm already a customer)): if you have a bunch of puzzle pieces, and you're not sure what the picture is because you don't have the box, you can nevertheless start putting pieces together and after a point you get more and more pieces coming together faster until the picture is complete.)
Posted on 2011-05-15 by Jach