I don't disagree with the author, but I do think that the mechanisms he's asking for, with the exception of the stronger sandbox enforcement, fall under the hea

I don't disagree with the author, but I do think that the mechanisms he's asking for, with the exception of the stronger sandbox enforcement, fall under the header of "magic pixie dust" security. Randomizing memory layouts and enforcing W^X may help prevent buffer-overrun attacks, but it won't help in cases like the ARDAgent backdoor, or even in the case of something like the current Ruby vulnerabilities. (W^X is particularly useless when you're talking about a dynamic language runtime -- the scripting code pretty much has to live in writable memory, unless you're compiling everything to C first and using dlopen() and kin.) The kauth framework and Scheme-based sandbox language already built in to the XNU kernel from Leopard on allows a lot of work to happen today, if security vendors decided to get off their collective asses and offer a comprehensive OS X security add-on. Unfortunately, none of the major AV/anti-malware vendors have a reputation worth a damn in the OS X community. They spent so many years selling "do-nothing" AV products to Mac users that no one takes them seriously now.