Those are two very different questions - the second is so abstract it's essentially meaningless.
The first however - is not a bad question at all.
Note that it is not a 'brain teaser' with a specific answer a-la Microsoft's 'river crossing' type.
A similar question used in consulting is 'if Quebec separated from Canada, how would they divide the debt' although that is less mechanical.
The 'Portland mechanic' question is actually quite good, and I enjoy asking them.
The objective is simply to see how people might think creatively at arriving at an answer.
I once asked the infamous 'how man tennis balls on a 747' question - and the interviewee started calculating how many balls would fit in the area of various body parts, it was ridiculous.
You can tell a lot by how they answer such questions, and get a good grip on their common sense, however, that type of question might be more suited for business analysts.
They are good questions to just get a rough idea of how someone might think about ambiguous problems that need solving - of course, they could never be used in isolation.
FYI - these arbitrary data questions are quite valid for software. We don't expect people to write a perfect 'bubble sort' or know specifically how the various sorts work - rather - the questions just provide an opportunity for someone to work through a simple problem - and then code the solution.
A crazy number of people have trouble with basic things, it's important to weed them out. Smart, coherent people will shine in these scenarios and it will generally be clear.