This is a flawed analysis because it assumesI'll repost my earlier scenario which demonstrates as such.
First we start with the assumption that data caps are being put into place to better user experience. But we know that Internet usage throughout the day is not constant, but rather has peak times and lull times. We also know that there are heavy users who use are constantly downloading so they are using their Internet during the peak times and during the lull times. During peak times, you have the average users competing for bandwidth with the heavy users, so every one has an equal share of the available bandwidth. Now if we assume 2% of users are heavy users, then the math is the following:
b = available bandwidth
u = total number of users using the Internet at one time
With heavy downloaders and average users using the Internet at peak times, each user gets:
b / u
If we assume that the heavy downloaders don't use the Internet during the peak times (remember data caps supposedly improve overall user experience), then each user gets:
b / 0.98u
The percentage improvement for the average user if all heavy downloaders stopped using the Internet during peak times:
(b / .98u - b / u) / (b / u)
=(1/.98 - 1)(b/u) / (b/u)
=1/.98 - 1
=.0204
=2.04%
So the average user will see ~2% more available bandwidth if all the heavy downloaders stopped using the Internet during peak times.
1) the same probability that you will be sharing bandwidth with a light user and a heavy user. Heavy users are much more likely to be using the net than light users (more so off peak hours, but during them as well).
2) everybody is using his/her connection to capacity all the time. I watch video streaming from certain TV channels, and it uses ~1mbps, probably 99% of the time I'm not using much more than that (I have a 10mbps link). Heavy users downloading shit ton of torrents and stuff simultaneously generally try to d/l to capacity.
Say there are 10 peak hours, a light user might be using it actively for say 4 hours, and probably <5% of the available bandwith on average (email and web pages are downloaded a lot faster than you can read them e.g.). A heavy user will be on for all of those 10 hours using 80%+ of his bandwith.
Now do an analysis for this, much more realistic, case...