An usher comes in and asks all of you to move to the right side of the auditorium, so that all of the seats on the right side are empty. You and four friends sitting next to you are wearing blue shirts. The usher asks your four friends to leave and gives you a sign that says "5" to indicate that there were five people in blue shirts sitting there originally. He does this every time he sees more than one kid in a row wearing the same color shirt. As people leave the auditorium, the ones who remain move to the left, making sure there are no empty seats between them. Eventually, the auditorium, which was almost 50 percent full at the beginning, is only five percent full. This is how file compression works.

Later, the expansion program does the equivalent of reading the numbered signs and understanding, for example, that your sign with a "5" means that four more kids in blue shirts need to be there. The kids who were asked to leave are called back into the auditorium and sit according to the signs.

The software that expands the files to their regular size uses a decoder -- a kind of dictionary that tells it what the signs mean. If you know the code, you can easily determine where the kids should sit when they come back in. A compression program simply looks for patterns that will reduce the file size as much as possible. The more repeated patterns the dictionary finds in a file, the more the file can be compressed.