as the client, i would always split the project into 2 parts - a functional spec and a technical spec. my job was to interpret and write the func spec as a kind of pseudo-code version of what i thought could be built. then i would commission a fixed price job of interpreting the func spec and re-writing it as a technical spec ie, click a button and take the user to a summary of xyz turns into a bit of SQL. the tech spec must be written so that any programmer could use it to understand what needed to be done. then I'd ask for a fixed price quote to deliver on the assumption that i could get multiple quotes if needed.
the benefit is that my coder who wrote the tech spec should be very comfortable with what is needed and i get to sell a lower risk $$ to my managers.
i think the hirers forgot that they are also being interviewed. not one of them seemed to feel that they needed to do anything other than allow the anointed one through the front door.
that being said, it seems as though each person selects people based on a wide range of criteria - i think the portfolio idea is probably the best, especially if other good programmers say they're good. Q is doing open source work the equivalent of building a portfolio with the benefit of allowing other programmers to rate your stuff?