10:36 playing time, 7629 frames, 41h41m, ~20s between frames, Nov 14-16, 2007
This is the first time lapse that I've time-stamped. I felt it was useful as without the time-stamp, it's hard to tell exactly what time it is, especially at night. It also allows you to see how traffic changes with each rush hour.
I think this time lapse would have worked pretty well with a very short interval, perhaps a second or less. Obviously, I couldn't have run it for nearly two days, but it would have converted the traffic from random noise to the main subject. The most interesting stuff happens in the second evening when various mists and clouds come in, so feel free to fast-forward to that.
The biggest problem with this time lapse is the huge number of frames don't actually contribute to the study. I could shorten the 10 minute running time by increasing the frame rate (or by not including all the frames), but at higher rates the traffic just looks even more random. Things would work better if the view changed much more smoothly.