We don't mind advances in kernel security, however wasteful excess in
the kernel has yet to be carefully evaluated.

We don't need ssp to have a secure system. Good administrative practices
is the single most important thing we can do for our systems. 
Therefore using ssp to protect unproven attack models (such as the linux
kernel) could be merely waste of the systems valuable resource pool. It
doesn't mean it shouldn't be implemented on systems to in an effort to
secure/harden them, I am just not a fan of hype without proof that 
a) It functions as advertised and is documented as doing so. 
b) that the trade off of performance is not to great and that real time
i/o can still occur.

I would use ssp, or whatever new implementation of security there is, if
it's right for the job and have been proved that the system won't start
degrade performance to an unacceptable levels.