And as was suggested earlier in the thread, the circuit, case, etc could be under the GPL just like the GumStix, with the BIOS being GPL'ed, etc. allowing multiple manufacturers to produce these units.
I think the market is ready for a portable unit you can play at home, and which *CAN* but doesn't nessesarily have to be your portable music player too, which can run any number of different games, not just those which are licensed through a central authority.
Oh. I agree. Look athe the success of projects like MythTV in fact I would suggest that MythTV is a very good base to start since it has really good media support all ready.
I don't want to discourage development at all. A GPL game console would be excellent and a great feather in the cap of LinuxBIOS.
Just don't be thinking you are going to get the same kind of performance you get from a commercial unit without spending 2 to 3x what they cost. Haveing really a reallly good software engine may help to minimize this but to get a _really_ good software engine you have to have good knowledge of the hardware. With the current graphics landscape thats hard and boderline impossible to get.
One glimmer of hope. If you can work with these guys
http://lists.duskglow.com/mailman/listinfo/open-graphics
you might actually be able to drive a small market for it.
I'm interested in it since Bitworks is capable of producing board runs on the level you are talking about. Say like 25 - 100 units. You can do like the MegaSquirt people do. Start an escrow account and take orders until you get enough that you can have it built. If the schematics, and gerbers, and BOM are alreday done then thats a lot of the cost of having a custom run built.
Of course whoever does the prototype(s) is going to pay a chunk. Perhaps you could seek some funding somewhere? Most of the groundwork issues could be worked out with COTS stuff thats available already.