That's because 250GB is stupidly low. It equates to only 4% utilization on a 20mbps connection. I can buy 250GB in a datacenter for a dollar. And on top of that the cap didn't grow over time to reflect increased bandwidth needs and decreasing bandwidth costs.
Give me a cap that represents fair bandwidth prices and I won't object.
250GB may be low, but if you think that you can equate data center prices to prices for transit delivered to an end-user's address, you aren't thinking hard enough.
Educate me. As far as I know, the only expensive part is in establishing a connection to a user. The cost of extra data is only in the backbone parts of the network, and those do not eat a large portion of revenue.
In short, data centers are generally located at the core of networks, end users are, by definition, at the fringe of the network. All the equipment and lines between the core and the end user add to the cost to supply bandwidth through up-front purchasing cost, maintenance cost and configuration cost. All of those are costs that would not have to be paid if they did not supply internet to end users. To assume that the cost of just supplying bandwidth to a date center is the same as through all that infrastructure is to assume the cost of all that is zero, or close enough they write it off, while in truth it probably accounts for the lion's share of their overhead.
...except that the last mile has the same 10mbps or 50mbps or whatever capacity even if you're on the crippled-speed budget plan. The only cost to increase the quota is for the ISP to upgrade its major inter-site links, not the edge connections.
If even 10% of the end users fully utilized their circuits they would be far over their capacity to get those packets to their core, even if the core could handle moving them to other networks. As such, increasing the amount of transit the ISP purchases may do nothing for allowing it to be effectively utilized at the end user location.
Give me a cap that represents fair bandwidth prices and I won't object.