Well, as evidenced by this thread, candidates are frustrated by the interviewing process at Google, in particular by poor/generic/boring questions that test a narrow/irrelevant topic. So, yes, I took pride in selecting questions that aren't like this in order to find what the candidate is good at, not what they don't know.
And rest assured that I always made it very clear to the candidate that abandoning a question they are stuck on is best to maximize their chance of doing well in the interview. If I have a 45-min time slot to spend on a candidate, I don't want to waste 30 min on a single coding challenge that they do poorly on and have barely 15 min to cover other topics. If I get a sense they won't do well after 10 min, I stop it, move on, and that leaves us 35 min to do other coding challenges. I have more chances of finding what the candidate is good at in 35 min than in 15 min.
And rest assured that I always made it very clear to the candidate that abandoning a question they are stuck on is best to maximize their chance of doing well in the interview. If I have a 45-min time slot to spend on a candidate, I don't want to waste 30 min on a single coding challenge that they do poorly on and have barely 15 min to cover other topics. If I get a sense they won't do well after 10 min, I stop it, move on, and that leaves us 35 min to do other coding challenges. I have more chances of finding what the candidate is good at in 35 min than in 15 min.