They're not open-source hardware in that sense. What they publish is a 19-page pdf of schematics of how the components are put together [1] -- and even that has redactions due to NDAs (like page 14 -- the say they have a "super NDA" with the GSM chip provider that bans them from even publishing the datasheet!) The components themselves (like the Samsung SoC and the GSM chip) are completely closed.

"Open" in that they show you how they put together the black boxes. "Closed" because it's all made up of black boxes.

[1] http://wiki.openmoko.org/wiki/Neo_FreeRunner_Hardware#Hardwa...

[2] http://lists.openmoko.org/pipermail/community/2007-January/0...

"Reflections on Trusting Trust", Ken Thompson's 1983 Turing Award lecture: http://cm.bell-labs.com/who/ken/trust.html

Also, there is an interesting article on trusting hardware ( http://theinvisiblethings.blogspot.fr/2009/03/trusting-hardw... ) which was reposted two months ago on HN ( http://news.ycombinator.com/item?id=3656522 ).

Even then, trusting trust is not really all that undetectable in a world of different compilers and different hardware. Or, perhaps more accurately, the power of the adversary has to increase by a lot to maintain the illusion of non-infection.

http://www.acsa-admin.org/2005/abstracts/47.html