Spend enough time doing threat modeling, and it becomes plain that the phrase “secure system” has different meanings depending on context. Some examples:
Business computers need to be secure against hackers, criminals, and industrial competitors. Military computers need to be secure against all those threats plus enemy militaries. Some business computers, those that run the telephone service are a good example, need to be secure against military threats as well.
Many urban transportation systems use prepaid farecards instead of cash. Similar prepaid phone card systems are used throughout Europe and Asia. These systems need to be secure against forgery in all of its forms. Of course, forgeries that cost the forgers more than legitimate use are not a problem.
E-mail security programs need to ensure that e-mail is secure against eavesdropping and alteration by any type of attacker. Of course, the program cannot protect against manipulation at the end points: a Trojan horse in the computer, a TEMPEST attack against the computer, a video camera that can read the screen, and so forth. Encrypted telephones are the same; they can secure the voice conversation in transit, but can do nothing about room bugs.
The trick is to design systems that are secure against the real threats, and not to haphazardly use security technologies with the belief that something good will come of it. The way to do that is to build a security policy (sometimes called ...