I often joke that my ability to formulate good Google queries is the single most important part of my software engineering skillset. Except, you know, I'm not really joking.
(I taught intro to CS for the first time last year, and watched people get very hung up by subtly broken things they found on the innertubes. To the point where we simply told them not to - we'd much rather have them asking the course staff, until they've developed the experience to sort out the wheat from the chaff themselves.)
I eventually got things working, though I feel far from confident about openGL ES. Actually, embedded shader code, packed arrays, limited debugging and my rusty matrix algebra doesn't make it easy.
Google is also very good at what I call "meta-guidance." If I search for something and don't find any good results, there's two possibilities: a) I'm not asking the right question, or b) I'm trying to solve the wrong problem.
If the candidate knows the problem he/she is facing and can formulate good query to look for a solution; that's a ++
Then I have to exhaustively Google to make sure that it's not an obvious Google search away. Then I start listing the steps that I've tried before, and I realize a possibility that I hadn't considered before.
The possibility works, and I close the window. Oh.
These days, I usually just walk the dog. By the time that I get back, I've usually figured out the answer or figured out how I'm going to look for the answer. The dog doesn't mind that his owner is babbling about code for the entire walk because he's a dog.
I find it to be very irritating if it happens for people that should know better...
In general I find that asking them to explain why they think this is causing the behaviour you're seeing they can usually work out they're wrong/convince you they're right. Of course this only works if they don't scream "I've found the bug" every time you step forward.