The whole net neutrality debate is fraught with profound misunderstandings of how the internet works. This T-Mobile/EFF fiasco included. We think of bandwidth

The whole net neutrality debate is fraught with profound misunderstandings of how the internet works. This T-Mobile/EFF fiasco included. We think of bandwidth in terms of maximum throughput, which makes sense for a dedicated circuit, since often the customer of a dedicated circuit plans to keep it nearly saturated (consider an office phone system, for example). We also care about latency and packet loss, but these are generally application specific concerns. Internet browsing works pretty well with high latency, but VOIP does not. Generally, it's always possible to overbuild the circuit and exceed the minimum performance requirements. A small office could install a second T1 to handle the extremely rare case of all phones and all fax machines being in use at once, but it may prefer to save the money and use a single T1. The second T1 could be viewed along the lines of an insurance policy against a usage spike, and the decision not to buy it could be considered "living dangerously" or voluntarily exposing the company to that risk in exchange for money saved. T-Mobile is not selling a dedicated bandwidth circuit to its customers, in spite of what the EFF wants us to feel indignant about. Yes, we all know the slippery slope the EFF is concerned about, but what T-Mobile is doing is quite reasonable. Imagine that you are the only person using a cable modem in the whole neighborhood. There is ample bandwidth available to you. But if all the neighbors are also using it, there is throttling and prioritization going on so that the finite bandwidth can be shared optimally. What does optimally mean? That is very much in the eye of the beholder. Maybe you play network games and would prefer that the network be optimized for low latency and UDP. Maybe your neighbor runs a BitTorrent node. Maybe someone down the street downloads raw SETI data for some kind of analysis. Each person has a different optimal QoS that he/she wishes the service provider would impose on everyone else. T-Mobile has strong incentives to keep its customers happy. By analyzing the bandwidth usage, it has found a way to offer bandwidth at a lower cost by imposing a specific kind of QoS filter on the data. This is precisely the purpose of QoS, being able to "stretch" bandwidth to accommodate more traffic by placing limits and prioritization on the traffic, with the goal of meeting all of the application specific requirements that the circuit would ideally meet. So, quite reasonably, T-Mobile has found a way to constrain some of the most egregious bandwidth hogging behavior its customers do, and has engineered a way to offer the constrained service at an improved price and packaged in a way so customers perceive greater value from the service. Yes, there might be someone trying to download SETI data over the T-Mobile LTE connection, or trying to compete in networked gaming, or all sorts of edge cases, but like Amazon's choice of which products it offers as Prime, T-Mobile has chosen a broadly appealing QoS filter which will result in most customers perceiving better value. Additionally, it allows T-Mobile to leverage content providers in infrastructure planning risk, which helps reduce the amount of bandwidth speculation going on, and overall makes the circuit perform more like a dedicated circuit. This point is subtle but true. So while the T-Mobile CEO seems a bit over the top, the EFF has (I think) gone over the line in an ill-conceived attack on one of the more progressive service providers. I personally donate to the EFF and proudly display the laptop sticker, but I also understand QoS and the complexity of shared infrastructure planning.