I believe the "stupid nature reference" is just to state where they got their inspiration from, and it also serves as a non-technical analogy that laymen can understand. It's a pretty standard practice that you'll find in many CS textbooks. Also, you're making a lot of assumptions and outright illogical statements.
1. You clearly aren't very knowledgeable about network/system security. Sure, you can have complete security if you leave your computer off or don't connect it to an external network, but that's not always practical. A computer sitting disassembled inside of a 20-ton safe isn't very useful. Software is written by human-beings, and the more robust the software is, the more likely it is to have bugs. If a bug can be exploited, then you have a security problem on your hands. You may know everything there is to know about security (at the moment), but there's no telling what new exploits will be found in the future, and what new viruses will be written to use them. And, usually, it's only after a virus has infected lots of machines that security researchers pick up on it and and learn how to detect/remove/prevent it.
2. How do you know that the servers sitting between you and your e-mail server/online banking server/ecommerce site server are secure? Just as malware writers use encryption to communicate securely with their botnets, so can security researchers communicate with their "ants." And the idea is to run the application on your entire network so that the network uses its collective computing power to detect signs of a potential new infection. This is only meant to be used on large networks and is more concerned with preventing an infection from spreading through the entire network rather than protecting a single PC.
3. That reads a lot like, "[If] you want X, you should do Y. However, Y usually doesn't work." If monitoring network activity doesn't work on the dangerous malware, then wouldn't it be a good idea to try another method? What exactly is your point? That computer security is futile? This has nothing to do with the idea being proposed, which is merely a way of applying distributed computing to the deployment of network security. The article only mentions that the ants will look for evidence of a security threat, not what type of evidence it is looking for.
There is no perfect solution to computer security, which is why it's an area of continuous research. Of course malware writers will try to hide, disguise, or otherwise improve their worms and viruses. Likewise, security researchers will continue to improve their detection/prevention techniques. It's an ongoing arms race. That doesn't mean nothing can be done to improve the state of computer security.