I'm thinking right now of the R2Duino, which I love. A simple, small Arduino with two built-in relays. I want guys to be able to produce stuff like this... and for it to work without too much hassle at my end. (As it does now). If they're forced to either find the skills to integrate their own FTDI/Serial method (instead of being able to use the "source" schematics and tweak the bits they can) OR to pay $2000 for their own vID/pID then it's going to impact the ability of the little guys to produce these unique variations.
So, if I were creating a clone I can't really see the benefit of switching from something I'm already using and know works problem free to a new thing whose main difference is that it'll cost me $2k.
Oh I agree - I'm thinking a little way down the line. When the Duemilanove etc. has long been forgotten or where basing projects on the newer platforms is preferred.
What is gained by running around in panic like headless chicken?