"Yeah," I said.
"So how come you weren't on Xnet last night?"
I was grateful for the distraction. I explained it all to him, the Bayesian stuff and my fear that we couldn't go on using Xnet the way we had been without getting nabbed. He listened thoughtfully.
"I see what you're saying. The problem is that if there's too much crypto in someone's Internet connection, they'll stand out as unusual. But if you don't encrypt, you'll make it easy for the bad guys to wiretap you."
"Yeah," I said. "I've been trying to figure it out all day. Maybe we could slow the connection down, spread it out over more peoples' accounts --"
"Won't work," he said. "To get it slow enough to vanish into the noise, you'd have to basically shut down the network, which isn't an option."
"You're right," I said. "But what else can we do?"
"What if we changed the definition of normal?"
And that was why Jolu got hired to work at Pigspleen when he was 12. Give him a problem with two bad solutions and he'd figure out a third totally different solution based on throwing away all your assumptions. I nodded vigorously. "Go on, tell me."
"What if the average San Francisco Internet user had a lot more crypto in his average day on the Internet? If we could change the split so it's more like fifty-fifty cleartext to ciphertext, then the users that supply the Xnet would just look like normal."