![Image](http://www.penstarsys.com/reviews/sound/bluegears/opamps/wave.jpg)
When we digitally record, the computer takes in the analog signal and picks specific points along the wave form to graph in internal memory. The points selected are dependent on the bit rate. The higher the bit rate, the more often points are plotted. In lossless audio (.flac) files, the bit rate is such that the computer is graphing as much of an analog signal as possible - making it virtually indistinguishable, numerically, from the original signal, even to a computer (there is also the issue of fixed-point vs floating-point but that is an entirely different tangent that I couldn't care less about). As you can see by the image, digital recording looks like a series of staircases linked together, going up and down. These "stairs" are the individual plot points the bit rate has specified should be sampled and stored until the wave is completely analyzed. After the wave form is recorded and all plot points have been registered, the computer goes through and does extrapolations (there's a real math term for this but I can't remember) to "smooth out the stairs" according to what points have been graphed. With a lower bit rate, there is more space between nodes, and thus more imagination left to the computer to fill in the gaps. And since computers are not sentient beings, it relies on faulty code that idiot computer programmers write to make up the difference of those gaps. That's why when you hear something with a 64kbps rate, it sounds crazy shitty compared to 192kbps or 320kbps.
Of course, the closer to the true analog wave, the bigger the file because there is more data being represented. And THAT is why Fade was reduced to almost 1/3 of its normal bit rate to fit as a file attachment.