I would argue that a 4 GB plan should provide 4 GB for a given fee, whether it gets used up over one month or one year. And when that is used up, they should bill you for another 4 GB.
The main reason for having clear quotas is to allow the ISP to estimate for bandwidth usage. As another poster said, it's not the total volume that really matters, but the peak rate of data flow. By saying "you can download up to X GB this month", over a large number of customers, you get fairly predictable usage patterns. If you say "you can download X GB whenever you like" to a large number of customers, you get very difficult to predict usage patterns, which means that provisioning sufficient bandwidth for peak usage becomes difficult.
The 'ideal' solution would be for the ISP to provide a real-time view of their available capacity on each of their transit links, and for users to self-regulate their traffic. If the usage is below say 50%, then you can download as much as you like for no fee. As the utilisation increases, those who have already used a lot of bandwidth get a lower quality of service, unless they pay an increasing fee for "premium" bandwidth. That way, if you're a low usage type of customer, you can just use the internet whenever you want and get full speeds. If you're a high usage type of customer, you can monitor the available capacity and reduce your usage during peak periods, go nuts during low usage periods, and pay a low fee. Or, you can pay heaps and go nuts all the time.
The reason that solution won't work is because most people wouldn't be able to understand what the heck was going on, and many of those who did understand it really can't be bothered with that kind of micromanagement. So, a periodic quota provides a kind of in-between point which is easy enough for users to understand (especially if you provide nice graphs and such) and which also gives the bandwidth provider a reasonable stable and predictable usage pattern.
And if a download drops midway and has to be restarted from the beginning (or if a page fails to load and requires reloading everything), the phone company should have to eat that cost.
The problem with this is that it's difficult to identify who's at fault if a download drops. Why should your phone company have to pay more because you're trying to download from a site that's unreliable and keeps dropping offline? What if the download failed due to user error, e.g. moving outside of the service area during it? What's more, even if you do manage to establish clear rules, trying to prove whose fault it actually was after the fact will be very difficult.
My attitude is that I'm paying a monthly fee that provides up to 5 GB per month and I'm only using a fraction of that, I'm wasting money. Thus, I might as well find a way to max it out every month.
Yep, and this is in fact what I do with my home internet connection: non-critical downloads are kept aside until toward the end of my billing cycle, at which point I let them loose since I don't particularly care if I end up shaped for a few hours before the quota resets. But again, this is predictable behaviour which makes network capacity planning easy.
On the other hand, I have a 1 GB quota for my phone which I never get anywhere near (lucky to break 100 megs) and I've never tried to maximise my usage of that... but my phone isn't my primary internet access mechanism, so I guess the psychology is a bit different.