Originally Posted by pingflood 
I can see #1 and #2, but not so sure about #3. Yeah, the camera doesn't have a massive CPU, but it does have hardware optimized to do just jpeg compression. I don't think a lack of processing power leads to inherently inferior jpegs; in fact the output from most cameras can be pretty damn decent.
One example: if you compare the size (in pixels) of camera jpg versus raw export you will see that most raw converters output a slightly larger image (slightly wider view, not just stretching it). That is because processing the pixels near edges needs a more complicated algorithm(edge pixels don't have enough neighbouring pixels so they need a different interpolation codepath), so in camera processing just crops them out to simplify matters.