Gunnlaugur has a good point, if RO were open sourced you would probably find it being taken in a dozen different directions, and end up with a dozen half finished versions, because there is no agreed roadmap for its future development either amoungst users or the programmers that might work on it. Everyone is concerned addressing what they percieve is a particular fundemental weekness that has to be tackled above all else, and no one is considering how to re-enforce the few things that give RISC OS its unique selling point.
As for embedded work, this can filter back many useful things to desktop RISC OS. For example improved networking with DHCP support, STB browser development (although with the continuing delay of O3 you could be forgiven for overlooking that), plus PCI and USB support which were first developed for embedded devices. All sorts of new things could come out of future STB work, such as support for streaming video to graphics cards from DTV or DVD. Even things such as synchronising between PDAs and remote severs, or downloading music to MP3 players are features proposed for future STBs.