Well, if porting to an open hardware platform like neuros is out of reach (because of time/efforts/etc...), then I think there are only 2 ways to build a mass-production model for a floss media center STB:
* bet on eeebox and similar, like the ASUS NovaP22 (see specs tab:
http://promos.asus.com/US/evensite/Nova/index.htm)
* study specs of cheap STB's already in market like those in
http://www.globalsources.com/gsol/I/HD-s.../23932.htm and try to get linux+xbmc running
But IMHO, the ideal scenario for mass free/open source media centers in people's living room is something like:
- A set-top box hardware which allows the user to modify it as he/she likes (just like the PC, you can change memory, processor, stack hard-drives...)
- A base O.S. which enables the user to install/uninstall their preferred media center app (I'd personally start with linux+xbmc/boxee, but the platform should be open to others willing to port their app to run on this STB)
- Downloadable/upgradable media center applications available on the web (working pretty much like the apple tv patchstick to install/upgrade xbmc/boxee).
With such base architecture, there would be little or no reason for the masses to buy proprietary boxes or software for media center.
There would be an open hardware standard to run such multimedia apps and enrich the user experience in infinite ways (each media center app will offer something different: tv tunning, social networking, p2p, voip... etc...) and the user is free even to have 2 or 3 installed on their box!
This is what the Freedom Box project is all about. Do you think this is something interesting? If it is... I'd like to invite you to share your ideas and experiences (here or in a separate place if you like). We are open to discuss and build the steps to make this a concrete reality.