It is my opinion that what should be accomplished in an interview is not an A to Z question and answer session trying to find every tidbit of PHP knowledge one has. There should be some well thought-out technical questions, not designed to show their flashy skills, but designed to expose the interviewee's thinking power.
For example:
Ask the interviewee to solve a technical problem, such as, "Given an array of integers from 1 to 1 million, write a function to determine which integer is missing in the range." At this point, give them a pen and paper (or white-board) and let them actually try to solve the problem. Don't rush them for time (but don't let them take all day either). You can prod them with ways to go about it, but above all make sure they are solving the problem. You can learn a lot about a programmer by how they deal with seemingly trivial problems, and how they deal with writing code under pressure.
Once you have finished that portion, ask them to optimize the code for speed and efficiency. If they had to cut the processing time by 25% or even 50%, how would they do it. At this point, it could just be conceptual. You may not even have a way to solve it yourself immediately, but this is irrelevant. Make the programmer THINK! I've found that when you ask questions that make a person step back and think, you learn much more about them than their resume or a "tell me why you're a good programmer" question will ever reveal.
As a side note, one question I have yet to ask in an interview (but can't wait to get the chance) is this:
"Having seen Google Maps, how would you duplicate or enhance the software design (objects, data structures, web services)?"
It's a super abstract question, and the actual answer doesn't really matter, I'm more interested in how they approach it.