You want a service level guarantee (99.9% available, with some extra wiggle room for peak hours and degradation. With electricity, you're paying for energy generation, plus grid maintenance. With internet, you're only paying for grid maintenance.
The econ 101 solution to this is variable pricing based on hours of the day, to shift consumption to off peak hours. However, that's not attractive to ISPs for a few reasons:
1. Customers don't have a good understanding of how much data various activities use.
2. Just like airline tickets, you may be paying substantially more or less than your neighbor for the same service. ISPs do not want to add additional pricing transparency or control - they would much rather adjust the rate as high as they can regardless of usage, then lower it for individuals newly signing up for service or when existing customers threaten to cancel.
3. Many large ISPs also own media and cable TV businesses that are in direct competition with high bandwidth services like Netflix. Putting caps on that apply regardless of time of day directs users to their own offerings, which do not count towards the cap. It's also a hedge to gain revenue as cable dies, and extract rents from it's replacement.
4. If you run a gym, your best customers are the ones who show up least often - the same is true for the current ISP model. A cap is a good way to shed heavy users and provides a chilling effect on existing users who don't have a good grasp of computer networking.
I think it's dangerous to let those with deep pockets prioritize their traffic over mine.
If only a finite amount of bandwidth is available to the world, then yes, the rich will be able to spend more to take bandwidth from us.
However more cable can always be added, so what would actually happen is that the pie would grow.
This assumption is the same one behind energy grids, and assumes an ideal world with no bad actors.
Instead, what happens is that incumbent players with the deepest pockets buy up all the (relevant) competition and then corner the market on capacity.