The best that can happen with TCPA is pretty good -
it could stop a lot of viruses and malware, for one
thing.
No, it can't. That's the point; it's not like the code running inside
the sandbox becomes magically exploitproof...it just becomes totally
opaque to any external auditor. A black hat takes an exploit, encrypts
it to the public key exported by the TCPA-compliant environment (think
about a worm that encrypts itself to each cached public key) and sends
the newly unauditable structure out. Sure, the worm can only manipulate
data inside the sandbox, but when the whole *idea* is to put everything
valuable inside these safe sandboxes, that's not exactly comforting.
--Dan