Alex Dudley, VP of PR at Time Warner, mentioned something interesting in our interview with him:

“Look, I don't know why this is such - why this is foreign to folks. You know, you're either paying for consumption... I mean, the concept of paying for what you consume is not a foreign one. I understand that it's different from the way we've charged for the Internet in the past, and we admit that. But the concept that you pay for what you use is how you buy just about everything”.

It is, you know. Paying for data usage is a very simple concept. Anyone can understand it. The problem is that it’s an oversimplification of the concept. That is – consumption based billing may be how you buy gasoline and bread, but it really doesn’t accurately reflect the way that networks and communications services actually work.

Trying to explain these complex concepts without oversimplifying is one of the reasons we’ve spent – what, six posts now? – on TWC’s data caps.

For example, I’m not sure I accept at face value Time Warner’s claim that there are (or will be) congestion problems in the future. I tend to subscribe to the idea that these data caps ultimately have nothing to do with preserving Internet performance for Time Warner’s customers, but everything to do with trying to preserve the market for cable television while exploiting the existing monopoly Time Warner has in certain markets for a quick burst of short-term revenue.

Time Warner earlier used the metaphor of “going dutch on a Salad/Steak lunch” to show the relationship between light and heavy users. Ars Technica responded to it this way:

“TWC's steak/salad analogy breaks down when it's crafted more accurately. The real question is whether you would even have lunch with a friend at a restaurant that charged $45 for a salad and $200 for a steak. Certainly, in a free market, most people would go elsewhere.”

So, I’m skeptical. Time Warner charges $1/GB for something that costs them $0.10/GB, they’re only doing this in markets where they have clear monopolies and competitors do not have strong footholds, and they’re doing so despite the fact that the profitability and capacity of broadband has actually outpaced demand. In short, there’s no actual “problem” that the “solution” is going to solve.

But even so, one needs to be a devil’s advocate sometimes. We’ve constantly argued that data is not the limited resource, bandwidth is. It’s an argument which ultimately boils down to: “It’s not too much data that causes congestion, it’s too much data at once.” Therefore, prices should continue to be based on bandwidth, not data consumption.

The only flaw in this argument is that, if you cram enough data down the pipe, it will lead to congestion eventually. Multiple people would have to be hitting close to the maximum of their data capacity nearly all the time during peak hours, but it is plausible.

So, if you were to assume, despite the evidence, that congestion on Time Warner’s cable service is a problem and that data overuse will lead to congestion, does that necessarily mean that Time Warner’s plan is an acceptable one?

No. There are other and better ways to handle congestion.

In fact, we can think of five separate plans that would be more effective at managing congestion, both in the near and short terms. Some of them are non-trivial, some of them non-neutral, but they would all manage congestion problems better than trying to change user behavior using exorbitant fees for consumption as a stick.

The fact that Time Warner isn’t using any of these plans – and chose to use the data cap plans – implies to me that they were either unable or unwilling to come up with better alternatives.

Plan 1: QoS-based packet prioritization for light users over heavy users.

Time Warner could, if it want to, deploy new hardware and software close to regional network routers to flip a user from the highest QoS levels to a lower one if they’ve been saturating the connection for a period of time. Comcast uses a similar system, flipping users who use more than 70% of their allotted bandwidth over a 15 minute period to a lower QoS priority over the next 15 minutes when the line is congested.

There’s no change to packet priority when the line is not congested, the system is identifying users who are using the most bandwidth at the time of congestion, and it immediately improves the relative performance of the light user while neither blocking or charging extra for the data for any user.

Tracking QoS policies on an individual IP address basis rather than by protocol would be a large rollout, perhaps even requiring new modems at the client-side to get the best results. But Time Warner was going to have to do that anyway with the rollout of DOCSIS 3.0 technology to enable faster speeds, and it was going to have to track usage anyway in order to bill for it. If congestion really is a problem, this is a practical solution considering the unique opportunity provided by the DOCSIS 3.0 rollout.

[Full disclosure: Network Performance Daily is the company blog of NetQoS, which is a enterprise network monitoring solutions vendor, many of which monitor the impact of QoS policies. On the other hand, we’ve got four more ideas below that don’t need quite as much network monitoring, and Time Warner’s welcome to use these as well.]

Plan 2: Graduated Decline of Bandwidth during Peak Hours based on usage

If the idea to manage individual QoS policies based on usage is too complex or too expensive, a less accurate but more effective solution would simply be to track data usage and slow down heavy users during peak hours.

Rather than a data cap, or billing based on data consumption, this plan would track data consumption and use it as a sort of a loose rule of thumb as to how much bandwidth the user is using, and during peak times, throttle the bandwidth down to slower, but reasonable, broadband speeds.

For example, a user on a 15Mbit down, 2Mbit up connection would always have that speed during off-peak times. But he only gets that 15Mbits during peak times if he’s downloaded less than, say, 15 gigabytes during peak hours for that billing period – or, if measuring consumption only during peak times is too difficult, say, about 50 gigabytes overall.

At that point, during times of congestion, his speed is knocked to 7Mbits. Consuming more data would knock it down to 3Mbits during congestion, and even more would knock it down to 1.5 Mbits during congestion. Again, however, off-peak times – the times without congestion problems – still retain the normal speeds.

A sudden drop in speed can be frustrating; and this plan may cut user speed far more aggressively than is actually needed. But it can be implemented rather quickly and easily compared to a QoS policy solution, while having much of the same effect.

Plan 3: Charging for extra consumption, rather than all consumption.

This plan turns the broadband consumption assumption on its head. The major problem most people have with the broadband cap is that there’s no way around it – all plans have some sort of cap. This is, perhaps, one of the reasons why the (probably true) conspiracy theory around Time Warner trying to protect their Cable TV interests keeps coming up.

But, instead of offering 15Mbit “Turbo” plans, and then complaining about “heavy users,” why not give “heavy users” a chance to pay more for extra speed instead of charging them for the data they used to get for free?

For example, you could bring everyone down to a standard 5Mbit plan, with unlimited data for a reasonable monthly rate.

However, if users require more bandwidth, they would then pay extra for a certain number of gigabytes at that higher bandwidth.

So, assuming that you charge $45 for the 5Mbit unlimited access, a Turbo plan could charge, say $65 for 15Mbit access with 60GB at that speed. When the 60GB cap is maxed, users can choose to continue along at the standard 5Mbit access until the next billing cycle, or perhaps they can buy another pack of “high-speed gigs” at a reasonable rate (say, $20 for another 60GB). Unused gigs would roll over to the next billing cycle.

This achieves exactly what Time Warner argues it is trying to achieve – making the heaviest users pay more compared to the lightest users – but doing so in a way that no one has to worry about Internet service suddenly costing more than they expected. Friday Night Lights Season 1 won’t cost $31 to download anymore, if you’re willing to wait 14&1/2 hours to get it, compared to 5 hours.

Plan 4: Lower QoS during Peak Hours based on protocol
I don’t like this plan. I thought of this plan and I don’t like it.

It’s simple and it’s brutal. During peak hours, decline the protocol associated with the highest data consumption and with the most saturation – BitTorrent. During all other times, leave it alone.

As a tool, I love BitTorrent. BitTorrent actually helps you take advantage of the capacity of the network, instead of limiting you to the slowest link along the route from server to client. It is one of the wonders of the Internet, and despite the stigma it has as a “pirate’s tool,” BitTorrent is extremely useful for any application where you have to get large amounts of data to multiple users.

This is a stupid plan, and I don’t endorse it. It opens thorny neutrality issues, creates problems for people wanting to use the torrent protocol in other applications (World of Warcraft, for example), and is unfairly scapegoating one set of Internet users.

As bad as this plan is, it’s still better than caps.

Plan 5: Reselling service and letting the free market hash it out.

It could be argued that the broadband infrastructure in this country is a natural monopoly, like many utilities. However, they are certainly not regulated like utilities.

What might be an option would be going the Canadian route – letting Time Warner handle the backbone and parcel out bits of bandwidth to the highest competing bidders, who then compete with each other. One company might differentiate itself by offering lower prices with a cap, the other, higher prices without a cap. Another company might advertise fast speeds, another low speeds, another good tech support – and if customers are unhappy with one service, it would be much easier to switch to another service.

This might require regulation, as it did in the Canadian market, but if we can’t run DSL and Cable and Fiber to every home, perhaps this is the best way to preserve the best prices for the consumer.