I just thought of this as I was looking at how well I was doing with my contribution to distributed.net The whole idea behind this is if someone could get access to a large number of computers, even if they're not top class workstations, they can crack these tough cyphers. But then the point to be taken is how do they get access to that many machines? Here's where the paranoia comes in.
So, lets say someone who was working on part of some system, oh like linux.(the kernel) And they felt a need to crack one of these cyphers. They could put some code in the system to do the work during idle cycles. They would need to set it up so it would only contact the main server when ever the user did something that transmitted packets. This would keep transmit lights from blinking except when it was expected. Think you could find these with a packet sniffer? Well, keep in mind that a packet sniffer works by talking with the kernel, and the kernel could easily mask those packets. Even if they were from a different machine, so long as it was the same kernel. Of course, one could avoid most of the hiding of the packets if they made some reason for them being there, like some different kind of protocol or some such.(you know, the manager approach instead of the programmer approach.) .
It would pretty hard to detect. Although in a system like linux, sooner or later(ok, sooner) someone would see this code. But think about a closed system. Think about a popular closed system. One that runs on thousands of computers, is slow and bulky, and continually demands faster, bigger hardware to work. I'm telling you, it could be done.
On a side note, has anyone ever sniffed packets on a network with wintels? wtf are all of those packets of data??