People who read my blog must love long entries. They hardly have a choice ;) But even I balk at trying to write a single entry on the economics of a neutral internet vs. a differentiated internet. The brief version for those of you who cannot be bothered with reading it all is that the short-term benefit of differentiated traffic is trivial compared to the value gained by fostering the long-term growth of that traffic. Put another way, telcos wanting to de-neutralize the net are in danger of killing the goose that laid the golden egg. “That is too brief”, I hear you cry. Okay, so instead here is part one of the low-down on why the neutral net is in everyone’s best economic interests, and why that may not be enough reason to persuade everyone to keep the net neutral. Look forward to parts two, three and four over the next few days.
Politicians and lawyers run most of the world. Technology makes the world a better place. Politicians and lawyers usually have no understanding of technology. Draw your own conclusions on whether politicians and lawyers make the world a better place…
Is it a good or a bad thing that the US Department of Justice has argued against net neutrality regulation in its submission to the US Federal Communications Commission? Well, the answer depends on whether you want a vibrant or decaying US economy. If you want a vibrant US economy, it is a bad mistake. If you want to see the US economy continue to slip and slide, then the DoJ has just given it another little push in that direction.
The DoJ press release trumpets the risks of reduced consumer choice and reduced investment in networks. Let us stand back for a minute and just think about what that means, beginning with consumer choice. Consumer choice is measured in the variety of suppliers they can pick from. The power of the internet is that it has given people choice on a scale like never before. Today, anyone with internet access has a range of choice that would have been incredible just twenty years ago. Information, entertainment, retail products, news… the list of areas where the internet has opened up choice goes on and on. Think of them all as suppliers. Some supply content, some services, some goods, and you can find and access them all because of the internet. You can even compare the prices and hence make the free market even more efficient – in other words keep prices down. So if the concern is with choice, then anything which maximizes the number of suppliers on the internet – whatever it is they seek to supply – is in the best interests of the consumer. Net neutrality, by seeking to prohibit preferential treatment of one kind of supplier over another, helps to maximize the number of suppliers. It denies the opportunity for any supplier to get preferential treatment over another, which helps to ensure the maximum real choice for the customer. So the DoJ logic that net neutrality might inhibit consumer choice sees the world upside-down.
The other concern of the DoJ was with investment in networks. To understand the concern here, we should just recap on what the issue is. The issue is whether networks get to differentiate between different kinds of traffic. What does “differentiate” mean in this context? It means (a) set grades of priority for, and (b) set prices accordingly. When exactly would a prioritized network make any noticeable difference to anyone? Well, most of the time it would not. Think of packets on a network like cars on a road. If there is spare capacity (i.e. the roads are clear) it makes no difference whatsoever if one packet has priority over another. They both travel down the road at their own good pace and arrive at their destination at the same time either way. The only time it makes a difference is if the network has a bottleneck – the equivalent of a junction using our road metaphor. On a net neutral network, the first packet to arrive at the junction is the first packet to leave it. On a differentiated network, higher priority packets get to jump ahead of lower priority packets at the junction. This would be like having to let high-priority drivers go first every time you stopped at a red light. Does that make a difference to performance? Yes, of course it does. The higher-priority packets will get to their destination first. But the extent of the difference depends on the load and number of bottlenecks. If there is plenty of spare capacity, then resorting the queue at every junction makes very little difference. If, on the other hand, you have gridlock, then jumping ahead makes a big difference. So there is very little economic incentive to pay for higher-priority unless you think the network is going to be clogged up. Which begs the question of why the DoJ is convinced that allowing networks to set differential rates will encourage investment. On a simple economic analysis, it should do the opposite: there is no reward for building spare capacity into the network and potentially significant rewards for trying to maximize network usage, as this would justify higher differentials in price for different grades of traffic.
This last argument is not scare tactics from a net obsessive. Just take a look at some of the arguments being pushed by networks to justify differential pricing. AT&T helped produce scientific research that argued undifferentiated networks need extra capacity. Or at least, that is the rather simplistic headline that was generated from the research. Why anyone bothered to research this is beyond me – anybody with an intuitive grasp of mathematics would have jumped to the same conclusions as the researchers did without even bothering to do the sums. If you read the academic paper the research team wrote on why differentiated networks would be more cost-effective, the maths fundamentals are so simple that I can explain them in a few sentences. Imagine my driving scenario as above, where we compare what happens depending on whether high-priority cars get to go to the front at red lights, or whether it is first-in, first-out. Then set a target duration for the total journey. In the prioritized scenario, only the high-priority cars have to arrive within the target duration. In the unprioritized FIFO scenario, all cars have to arrive within the target duration (because that is the only way to be sure that the most important drivers do). Imagine then you have steadily more traffic on your road/network and that delays at red lights get longer as a result. Guess what? It turns out you need to spend less on building extra road/network capacity to compensate for increased traffic if you only care about whether the priority cars arrive on time. What a surprise. I cannot argue it is science fiction but it definitely is science phooey. If this argument is right, which of course it indisputably is, then it follows that networks can save money on network investment if they can prioritize traffic. But this completely contradicts the DoJ argument, which was that not prioritizing traffic might lead to less network investment. The only way to interpret this is that networks will simply not bother to improve their network unless they can get specific users to pay for specific improvements. Presumably the DoJ have completely discounted the need to improve the general level of service in order to stay competitive with rival networks.
Read tomorrow for part two and the logic behind the DoJ’s arguments.