While it is true that some good may come from the TCG's efforts, I really think that Tom Perrine was really on the ball back in 2002 when he effectively asked the question "Why are we here, again? (And why does no one remember the last time we were here, again?)" here:

http://www.usenix.org/publications/l...fs/perrine.pdf

Here are some highlights:

in 1973, the seminal report on computer security – the “Anderson Report” – had been published for the US Air Force. This report called for better software design practices, better programming languages, and something new called a “security kernel.” It also suggested using formal mathematical models to prove that the kernel would operate correctly. This paper also, almost as an afterthought, described what we now call “automated intrusion detection” and noted that a primary way to compromise an operating system was to exploit “insufficient argument validation.”
Yes, the Anderson Report described buffer overflows as a proven penetration method and ways to avoid them 30 years ago.We’ve obviously come a long way since then. So far that we need Palladium and TCPA.
(Palladium and TCPA are of course related to the TCG's current hardware-based security ideals)

People started to design and develop “security kernels.” These were small, well-defined cores upon which an OS could be written that would be small and “verifiable” using formal methods. This gets around the problem that verification methods and human minds weren’t ready to deal with analyzing very complex systems. The idea was to concentrate all the security features, and only the security features, into a small kernel that would provide the base upon which a secure OS could be layered. This was imposing “least privilege” on the operating system itself, allowing the operating system to have bugs and yet not be able to compromise security.
(Hmm so software-based security might work to prevent every little flaw from being an security issue, who knew?!)

Although KSOS (and SCOMP and Multics) made significant advances in computer security and software design methodologies and helped us to understand the problem of software quality and assurance, they have been mostly forgotten. These OSes, and their contemporaries, provided many features and services that are continually rediscovered or even “invented” every few years for new operating systems. Palladium and TCPA are just the most recent efforts to cover the same ground. In Orange Book terms, they are trying to go “beyond A1” into “trusted hardware,” without first getting to B-level software architecture.
Skipping steps... now there is the way to a secure system.

cheers,

catch