I have to admit that I see his point. Residential bandwidth pricing is based on the concept of oversubscription. I run a small local ISP and have seen sustainable oversubscription numbers move from around 75:1 ten years ago to somewhere around 12:1 today and am using 8:1 to plan deployments for the next 2-3 years. Equipment and transit costs have come way down, which have allowed us to keep pricing relatively stable while increasing package speeds, but as we approach 1:1 usage, there is no way to make the model work without passing on full bandwidth costs, core costs, and last mile costs to the end user. Around 3% of our subs end up costing us more in bandwidth usage then they pay us for service which is supportable for now, but as that percentage grows with increased full time streaming, we will either need to raise prices across the board, or start charging based on actual usage. What would be ideal, IMHO, is 95th percentile billing, i.e 10mbps on a 100mbps circuit with 95th percentile billing above 10mbps, but the vast majority of users just wouldn't get it.