I got to thinking again recently about security on the physical layer, and came to a few points. Programmers can only work within the guidelines of the hardware given to them, today most hardware devices are still relitavely simple or blind when it comes security. The 32-bit architechture compared w/ 64-bits provides little doubt that buffer attacks against memory processes are only a little more secure on a 64-bit architechture. We have commercially available routers being to make themseleves available w/ capabilities to prevent or minimized the affects of DoS attacks. What I am getting too, is that i think we need to stop sticking to mainstream ideas when developing hardware componets. Hardware makers for computer companies need to invest in new technologies for the consumers and retail. Ones that give the programmer more power to develop complex applications and processes that render or perform flawlessly under little to no stress, while also having more control over security. </rant>

more complex = more problems...

when it comes to saying a buffer can only hold, say, 20 characters, and then doesn't make sure it actually is =< 20, what is hardware gonna do to pervent programmer's error? or maybe there's an HTTP exploit that directly sends commands to shell, hardware's just doing what it's told...
mostly all i can think of that hardware can stop is generic overflows... and they are taking preventive measures for those (ex.. amd made a processor to "stop buffer overflows").. there are very many types of attacks hardware can't do anything with though

good thinking though