There's a story about some people who were writing the software for an early avionics computer.(Originally from a speech at the NAE, and also quoted on NPR)
One day they were visited by the weight control officer, who was responsible for the total weight of the plane.
"You're building software?"
"How much does it weigh?"
"It doesn't weigh anything."
"Come on, you can't fool me. They all say that."
"No, it really doesn't weigh anything."
After half an hour of back and forth, he gave up. But two days later he came back and said, "I've got you guys pinned to the wall. I came in last night, and the janitor showed me where you keep your software."
He opened a closet door, and there were boxes and boxes of punch cards. "You can't tell me those don't weigh anything!"
After a short pause, they explained to him, very gently, that the software was in the holes.
On a related note, there's an article in Internet Week :
Washington Decrees Microsoft Must Sell Software By The PoundThat led me to think about how much software *does* weigh. Assuming you start with memory randomised, equal numbers of zeroes and ones, how many electrons are needed compared with memory filled with all-zeroes or all-ones? Which should be assumed to be the initial state? Does a program have more ones than zeros, and what is the mass difference? Not that it matters, it's less than the difference in weight between a hot CPU and a cold one. And enormously outweighed by the difference between a CPU with one person's fingerprint on it, and someone else's. What I'm most interested in is whether the weight is positive or negative, not the actual amount.
...In order to establish a "fair" pricing regime for alternative middleware, supposedly to protect competitors like Netscape from unregulated "tying," the Justice Department wants to set software prices based on the weight of the code. I am not making this up. Take a look at judgment provision 3(g)ii, reproduced below...
Oh yes, Selling England By The Pound is THE classic album by Genesis. Catchy title too.