Yes, there are plenty of companies out there that very much depend on a network effect and aren't Y Combinator-funded: Digg, Facebook, Quora or Stack Overflow, to name a few.
Swiping is not obsolete. I don't know why you're trying to compare a virtual retailer like Amazon to shops that have an actual physical presence. The way they process payments is entirely different.
Not everyone has a smartphone. Why would Starbucks want to deal with accepting payments via mobile-geolocation and plastic for everyone else? It would just confuse the cashier. Swiping is uniform and only takes a few seconds.
Square fills the mobile payments processing niche where you need to be able to accept plastic when you're not in front of a register. Plumbers, repairmen of all sorts, traveling salesmen, people selling items at fairs/festivals, etc. These people aren't currently able to accept credit cards with real-time processing. That's the problem Square solves.
Is it not obvious? Why don't retail stores type in your credit card information instead of swiping it?
Chip and PIN doesn't seem to exist in the US so I'd imagine they're not targeting it immediately. I don't see why they couldn't expand the hardware to support it though.
I've had phone interviews where I've been asked to solve a logic puzzle with pen/paper while explaining my approach. Solving a problem you've never thought about while being put on the spot turns out to be incredibly difficult. These sorts of problems require deep concentration and quiet yet it tends to be awkward to have silence between strangers. The situation really inhibits serious problem solving.
What sort of programming problems are you giving them? Did these candidates have experience with other projects and were they able to fluidly explain their involvement with them?
It must be pretty rare to find a person who can describe a complex project in detail but is unable to write any code during an interview.