When they figure a way to bill users differently between "long-distance" and "local" packet transmissions, they can start to see applications/services put in ef

When they figure a way to bill users differently between "long-distance" and "local" packet transmissions, they can start to see applications/services put in effort to adapt. Also, mechanisms like L4S for high-fidelity ECN feedback allow for higher link utilizations without end-to-end QoS impact. IPv6 reducing routing table fragmentation should enable reasonable prefix based peering, along with billing senders according to the used peering. We don't need fast flat rates, we need cheap bulk traffic and flat rates for effectively-not-oversubscribed bandwidth. Maybe with a way to roll over unused priority bandwidth at some multiple into the next billing period to count against that period's bulk traffic usage.