The Internet is sometimes capable of working. It can go down and has; the Morris worm wasn't even malicious (according to Morris). Designs for 'Warhol worms' which do the same thing in just a few minutes have been floating around since the mid-nineties - it's just that botnets are more profitable. And even inadvertent mistakes can cripple lots of functionality.

"It has been proved that the scale-free network is robust to random failures but vulnerable to malicious attacks."

In the real world, malicious attacks are just as valid a source of failure as randomness. (It doesn't matter why the patient dies if he dies.)