We do not want, and should not aim for an Internet which is 100% full. Inevitable network congestion issues aside, the Internet only has real value when there is significant room for growth. It is disturbing to see the concept of "efficiency" increasingly applied in this manner, as it is indicates a fundamental misunderstanding of the Internet, and to pursue it, is to deliberately damage the Internet.
The only "efficiency" here is in how efficiently existing infrastructure is monetized, and how thoroughly startups and other newcomers can be squeezed out by the large incumbents. Metering removes any incentives to actually grow the Internet. The problem is a fundamental disconnect between what is sold and what is provided; by metering data, the relationship is entirely arbitrary, and lacking the most crucial parameter: time.
Data, which is measured in bits, is not an actual consumable, and there is no fair way to attach a price to it. What is consumed is bandwidth on the Internet links the data traverses at the time of usage; both of which are time-dependent. Needless to say, it would be insane, both technically and otherwise, to try to bill in this way. Furthermore, there still remains no incentive to actually grow the Internet, in the absence of competition.
Rather than data based metering, ISPs could be required to sell connections based on minimum guaranteed bandwidth to customers, at regulated prices. This is not the only fair scheme, but regardless, there needs to be a correlation between what is sold, and what is provided. If people want to purchase more, there must be an incentive to build out the networks, rather than to adopt a model of artificial scarcity, and bill accordingly.
Of course, selling connections fairly in this way also has technical difficulties, and it would be far simpler and cheaper if they just dispensed with all of this nonsense, and reverted to the way it used to be. Sell connections based on bandwidth, and build out the networks until there are no significant congestion issues. Other countries have proven that this model is still economically viable, even while providing people with gigabit connections. The only thing preventing it here is the lack of competition and associated stagnation of infrastructure and gouging. Let's not adopt new models of pricing which encourage more of that.