Chris, Some quick points.
1. As to cost. Once a source is selected, then we'll know for sure. These little boards were born in small quantities as side projects. Admittedly a 100% free hardware computer will likely be a specialty product appealing to a narrow group which in the grand scheme of things means low sales volume.
In such a scenario a low volume producer (a specialty or small board maker) is all that is needed.
2. The Cubieboard is very interesting, lots of support there too in their community, but it is not clear that their board is easily available to turn into a product. The Olimex board, which to seems to be very close to the Cubieboard in its main parts, is industrial quality, and everything is 100% open and available, at least that's what they say. The BeagleBone cards are neat but are not free and with them, there are hoops to go through should one wish to make a product based on their board.
I am new to the finer points of free hardware and free software. How far does one go? How deep? I mean the Cubieboard seems "free" as far as I can tell, however it apparently is not "open hardware". This would seem to say one cannot look at its intimate details. This would not be a good thing if it were software so is it good when it is hardware? I would say no.
Stallman isn't 100% clear on free hardware as far as I can tell. Sometimes he is ok with non-free things baked in that cannot be changed.
Also, hardware has become a more important issue to him. A surprising chunk of his LibrePlanet speech was about the need for free hardware. https://trisquel.info/en/forum/rms-talk-libreplanet
I've noticed that people who write things that are very close to hardware, I mean programmers, need as many, and as clear, details as they can get. It would seem the more open the hardware, the faster and better the programming. So I lean toward that.
3. The difference between a "modern desktop" and "good enough" won't be a question within a year. About every 6 months there is a new board. Always a power improvement, but sadly not always an improvement toward freedom.
Faster chips/boards are available now. But then they use GPU's like PowerVR, which is far from being known or open or free. The Mali GPU is "the one". That guy, Luc Verhaegen, needs help. http://libv.livejournal.com/ But even so I'd expect he'll have useable base code before this year ends.
Android is taking over Linux when it comes to graphics drivers. Wrappers have been created to use Android binaries and most Linux distributions seem ok with that which is not in favor of freedom. It seems most distributions believe they need to do something "now" to keep GNU/Linux in the game and remain an option other than Android which has taken over probably way too much of ARM and device markets. So they'll use the wrappers and run, saying they'll fix things right later. But this of course is a bad idea long term, "real" code needs to exist, no matter the timing.
This is also a long way of saying that once the GPU can really be used instead of the CPU for graphics on these boards, there will be very noticeable improvements in speed/power even on the older ARM boards.
4. I want to do more research on the Loongson. Beyond that if anyone can point to other processors that are worth a look, I would appreciate it.
x86 is quite known, I'm not sure there is a "free" way in? Doesn't seem like it, although I haven't looked into that area much. Seems like if freedom were possible there, there would be more examples.
ARM is big and growing. A nice thing about ARM is that the company began almost like these board makers in that they've always had (or maybe used) a strong community base to develop their products.
Since they do not manufacture anything, only license, this too is different from the x86 world. As mentioned in earlier posts, many independent factories produce ARM licensed products for many uses. When all is said and done, there will need to be a Trisquel or gNewsense ARM version at some point. I would expect that the "Replicant" phone project would have to be ARM based right now.
At the March LibrePlanet, Ruben was all for ARM when the time and conditions were right. http://quidam.cc/01-04-2013/talk-libreplanet-2013 (around 33 minutes in.) The conditions are getting very close to that now, but that's my view. If 100% free hardware were here, and if it is was ARM based, maybe that would be enough.
ARM is fascinating and I urge everyone to learn more about it. Since it is used in phones, tablets, TV's, and other "things" it is here whether we like it or not. Why not have it be as free as possible? Seems to work very well too, except in many implementations it is not free….but in others it appears to be!!
5. If you mean by "fizzle" that with the speed of change in the tech world, by the time a board is identified as "good", everyone has already rushed to a new one which isn't, and the old good one is discontinued, that has me concerned as well. However I see a couple of possibilities.
For example, Olimex seems pretty small, stable, open, whatever. Their philosophies on parts and price seem good. They could make more boards, old or newer, I don't see them as dropping and running to the next big thing. But that is just guesses based on reading, I really don't know. It will be something to find out if an ARM board is "the answer".
Also the Australian company Miniand has a thing for "unrestricted hardware" and for asking "will it run Linux?" when they create products. Maybe there is a way to work with them somehow too. They have an office/history in China. Plus Mimiand is already making some very close to free computers right now. A few tweaks and maybe they could have a free one.
And I'm with you. It seems everyone drops the ball on WiFi. Everything is good, check, check, check, then the WiFi! Nope, not free.
There are a lot of things to check out yet. I am happy to see your interest in these possibilities.
