> "Clogging the network" would be solved by carriers actually bothering to invest in their networks
This is exactly my point. With unlimited data plans, carriers have little incentive to upgrade their networks. It increases costs without directly increasing revenue. With pay-per-GB, the incentive is clear: upgrading the network increases usage which directly increases revenue.
The real incentive is improving speeds makes shared-bandwidth networks more cost effective. You can serve the same number of customers for less cost. Each connection receives its data quicker then disconnects so someone else can use the air space.
Time is the more important measure of cellular network usage than bytes. I think it would be more convenient for customers to be billed per minute. At $10/GB if my average speed is 5 Mbps that's $0.30 a minute. Or $30 to watch a full-length movie in HD. Makes me wish Netflix still had mail delivery.
> You can serve the same number of customers for less cost.
No, upgrading the network allows you to serve more customers at higher cost (the cost of the upgrades). If you have unlimited data plans, you can only recoup those upgrade costs by recruiting more customers because existing customers pay the same regardless of the speed. Recruiting more customers is also costly and affected by many more factors than just network quality. The connection between network upgrades and increased revenue is tenuous and indirect.
In contrast, with pay-per-GB, network upgrades increase usage and thus increase revenue from existing customers as well as new ones. The connection between network upgrades and increased revenue is direct and immediate.
I agree that $10/GB is far too expensive; I'm sure Google would love for it to be cheaper but the pricing isn't their choice to make. When carriers get some competition from Project Loon and/or Elon Musk's satellite thingy, hopefully prices will go down.
> No, upgrading the network allows you to serve more customers at higher cost (the cost of the upgrades)
Upgrades are a one-time cost. You charge a higher rate for access to the faster network until the investment is paid off then you phase down to all customers. Then your operating cost is lower per subscriber in the long run.
These days, upgrades are a continuous cost. By the time you've paid off an upgrade it's time to upgrade again. Charging a higher rate costs you customers, and with fewer customers you don't even need the extra capacity you just bought. Also many customers are on contracts so you can't increase rates right away.
In contrast, with pay-per-GB, upgrades increase revenue automatically without losing customers. You don't have to increase your rates and you can probably even decrease them and still increase revenue.
The direct increase of revenue comes from more customers arriving now that you're combining a robust network with unlimited data. Both of those things even in isolation are pretty significant attractions; having both of them is even better.
Barring that, however (as I mentioned elsewhere in this topic), congestion issues are already resolved in the wired broadband realm by selling speed rather than data quantity. There's little reason why the wireless world can't do the same by selling access to 2G/3G/4G separately, then making every plan unlimited (or, at the very least, have way bigger caps for everyone, as is the case currently with some of the not-so-great broadband ISPs).
Basically, you'd have T-Mobile's approach, but with a lot more flexibility. In combination with a no-contract carrier (like T-Mobile) or reseller (like Google Fi), a customer could, say, subscribe to a 2G or 3G plan for normal usage, but then subscribe to the 4G plan for a day or so if he/she wants to binge-watch Game of Thrones or somesuch. This way, the congestion issues are addressed (there's a direct monetary incentive for carriers to build more robust networks, since the faster data subscriptions would cost more and generate more revenue) while not imposing arbitrary caps on the quantity of data "used" by a customer. Win-win.
This could be even further extended by providing the 2G and 3G plans with an optional scheme of "we'll give you $X days of 4G per month" (similar to T-Mobile's "we'll give you $X gigabytes of 4G per month"). That way, our Game of Thrones bingewatcher doesn't have to deal with changing the subscription; he'd instead have a week's worth of 4G per month that he could activate on a per-day basis (for example). An even sweeter deal would be for these unused days of 4G to roll over to the next month up to a year (or some other suitably-reasonable length of time), like what T-Mobile already does for its non-unlimited plans.
This is exactly my point. With unlimited data plans, carriers have little incentive to upgrade their networks. It increases costs without directly increasing revenue. With pay-per-GB, the incentive is clear: upgrading the network increases usage which directly increases revenue.