Implicit in much of this tutorial is the concept of computational feasibility. Some attacks on cryptographic protocols can be done on computers, while others exceed the capabilities that improving computers will obtain. Of course, just because one line of attack is computationally infeasible does not mean that a whole protocol, or even an algorithm involved, is secure. Attackers can try approaches other than those you protect yourself against.

We refer to a protocol that is computationally invulnerable to any form of attack as "computationally secure." Keep in mind that "human factor" approaches are really properly described as "compromises" rather than as attacks per se (especially in this context). However, it turns out that we can do even better than computational security. Let's take a look in the next panel.