I don't want to sound overly hostile here, but have you actually read the post? Many (if not most) kernel bugs can be mitigated with existing technology, and there's ongoing research that will bring this down even further. There are certainly scenarios where assuming that any level of compromise may be significantly deeper than you imagined is the correct response, but that's not a supportable response in the majority of cases.
Looking at it another way - if application security is important, the game is over, the attacker has already got in via the network. We're bad at writing applications, so we shouldn't expose them to the internet.
All of the above. Do the best you can at network security, and try to get better. Do the best you can at application security, and try to get better. Do the best you can at kernel security, and try to get better. And do the best you can at intrusion detection, and try to get better.
Looking at it another way - if application security is important, the game is over, the attacker has already got in via the network. We're bad at writing applications, so we shouldn't expose them to the internet.