There's a big nasty dilemma hidden at the heart of mobile computing: No one knows how we'll pay for all that mobile data we're supposed to use in the next few years. The question doesn't get much publicity, but it drives some of the most intense debates in mobile, including net neutrality and the wireless bandwidth "crisis."
This is the second of a three-part series on the issue. In Part 1 (link), I talked about the tech industry's unlimited vision for the growth of mobile data, and why I think it won't come true because we'll run out of people willing to pay at the current rates for data service
In this part, I will talk about the alternate scenario, in which most people are willing to pay for mobile data and adoption of it continues to accelerate. In this case, the mobile operators will need to invest urgently in increased capacity, and even with that investment I think we'll eventually run out of cellular bandwidth.
This means the operators face two conflicting possible futures. In one, growth is about to slow down and they don't need to invest in a bigger network. In the other, they need to invest urgently in additional network capacity. For telecom execs, it's a bet-your-career choice with no clear winner. So, naturally, they are trying to get someone else to pay for the investment. That's the real cause of the rhetoric about a wireless "crisis," and it's driving much of the net neutrality debate.
To understand why this is happening, let's start with a look at the physics and economics of a cellular data network...
Mobile data doesn't scale like fixed-line broadband
When mobile operators in the US and Europe first built out their 3G networks, they miscalculated what people would do with them. They expected that new, relatively low-bandwidth mobile services like a simplified version of the Internet (called WAP) and picture messaging (MMS) would be the dominant source of data traffic on the network, and they structured it accordingly. But those new data services failed to take off, and the operators were left with a ton of excess capacity. Desperate to generate any revenue from their new networks, they offered fire-sale data plans for the newly-emerging smartphones. It didn't matter if the operator made a good profit off a smartphone data plan -- with the network already built and sitting idle, any data revenue was better than none at all.
So the operators in many regions gave us low-cost or unlimited data plans. Those plans set customer expectations for how they should use mobile data and what it would cost in the future.
Overcapacity continued on most mobile networks until the launch of the iPhone in 2007. We tend to forget about it today, but the iPhone was the first smartphone to make PC-style browsing practical and attractive for most smartphone users. The result was an explosion of mobile browsing, and almost overnight mobile data networks supporting the iPhone started to go from overcapacity to overload.
The reason for the overload was simple -- people in the developed world learned to browse first on their PCs, most of which have high-speed wired connections to the Internet. Bandwidth on these connections isn't infinite, but it's large enough that activities like file sharing and watching videos are mainstream.
When people started doing PC-style browsing with their smartphones, they brought their PC browsing habits with them. Unfortunately, the cellular wireless networks don't have nearly the same data capacity as the wired networks. So mainstream browsing behavior on a PC turns out to be excessive browsing on a smartphone, especially if you use a lot of YouTube.
Even if you're not a big video user, the normal sorts of messaging and web traffic created by a PC can overload a wireless network. A typical PC has a more or less continuous connection to the web, so instant messages and web app updates can ping back and forth constantly. But to conserve battery life and stretch network resources, a smartphone doesn't talk to a cellular network continuously; it basically says hello to the network, sends a message or a bit of data, and then says goodbye. Each little message and each app ping creates its own set of hellos and goodbyes. Send too many and they can overwhelm an operator's servers. Web apps send too many.
The operators, of course, can add additional wireless capacity to cope with the increased traffic, and they have been doing so. Cisco estimates that the total capacity of the world's wireless networks will increase by about 10x from 2010 to 2015. But these additions eventually run into physics problems. There's only so much information you can squeeze into a certain amount of wireless spectrum. At some point the cellular infrastructure overloads, and as an operator you have some ugly choices:
--You can add a lot more cell towers, reducing the size of each radio cell and therefore increasing the number of devices you can support. Unfortunately, this is extremely expensive -- not just to build the towers themselves, but for the fiber optic cables connecting them, the servers to manage them, and for the lobbying you must do to overcome political opposition to additional towers.
--You can offload data traffic to local wired connections. The operators are already pushing this hard, via WiFi. Some are also encouraging the installation of femtocells in individual homes and businesses. (A femtocell is basically a micro cell tower in a box the size of a wifi router. It gives you cellular service inside a building or business, using wired broadband to communicate back to the cellular network.) But that too is expensive: at around $200 a pop (link), it would cost about $13 billion to attach a femtocell to every one of the 67 million consumer broadband lines in the US. Plus there's the support cost for installing them, and the expense to put millions more cells in businesses, and the cost to buy the back-end servers necessary to support them. Nevertheless, some very smart people watching the mobile data market believe that femtocells are essential to the future of mobile data (link). Cisco estimates that offloading of some sort will handle about 20% of mobile traffic by 2015, and up to 40% in some countries.
--You can buy more spectrum, but there are huge licensing costs associated with that, not to mention the cost of retrofitting your cell towers for the new frequencies and replacing all of the phones in the installed base.
Unfortunately, even if you make all of the changes above, there are some very convincing arguments that it won't be enough, quickly enough, to head off a capacity crunch if current trends continue. The growth rate of smartphones, tablets, and wireless notebooks will swamp the cellular infrastructure no matter what. Folks in the mobile industry have taken to calling this the "Moore's Law vs. Shannon's Law" problem, with Moore's Law representing the exponential growth of computing power, and Shannon's Law the fundamental limits on how much data you can push over a particular chunk of spectrum. Reinforcing the Shannon bandwidth limits is the fact that some other critical elements in the mobile data infrastructure can't keep up with Moore's Law. The number of cell sites can't increase exponentially, and handset battery capacity is barely growing at all. A crunch is inevitable at some point.
So the people who tell you that cellular wireless will replace wired broadband just don't understand the physics involved. An outstanding summary of the situation was written by Martyn Roetter, a telecom consultant (link).
Here's the key paragraph:
"Until and unless the current laws of physics are invalidated in ways that remove current limits on spectrum capacity such as are embodied in Shannon’s Law, the future will see: (a) The vast majority of broadband traffic (as distinct from numbers of broadband subscriptions) continuing to be carried (delivered and transmitted) over fixed access networks; and (b) Demands for broadband traffic from wireless or mobile subscribers outstrip the capacity of all the bandwidth available for radio access networks to handle it, even with the use of the new spectrum that can be allocated and the deployment of more spectrally efficient technologies... Bandwidth within one optical fiber is vastly greater than all the bandwidth that might theoretically be made available for mobile communications, even if every megahertz were to be refarmed for mobile services. A single mode fiber has a bandwidth of as much as 100,000 GHz, or 100 terahertz, whereas total valuable spectrum for mobile communications provides bandwidth of no more than at most 3 GHz."
Got it? What he's saying is that wired broadband traffic can continue to grow exponentially, which will create demand for mobilizing that traffic through cellular wireless -- which the cellular networks can't handle.
If data traffic continues to grow at its current pace, we're headed for a situation in which the cellular networks will be overloaded no matter what we do.
Rock, meet hard place
So we have two possible scenarios for the future of mobile data. In the segmented scenario I discussed in part 1, we run out of customers willing to pay for mobile data plans, and the growth of mobile data slows down. In the consensus scenario, customer demand continues to increase, and we run out of cellular network capacity.
These conflicting scenarios are terrifying to the mobile operators because there's no way to tell for sure which one will happen. If you knew for sure that demand was going to continue to grow, you'd invest heavily in capacity, and also start raising data prices to restrain the growth in demand to something you can actually deliver. But if demand is about to stop growing, investing in capacity and raising prices is exactly the wrong thing to do. You'll end up with excess capacity, and the price hikes will make demand stop growing even faster.
This table summarizes the dilemma (click on it to see a larger version):
An economist would tell you that this will all sort itself out in the long term, and I'm sure it will in 20 years or so. But in the meantime, in the real world, the operators have to invest in infrastructure years before the demand arrives. If you're an executive at a major operator, it is almost impossible to get the forecast right. That means you will probably either overbuild the network, wasting billions of dollars and putting your career at risk; or you will underbuild, losing share to competitors and putting your career at risk.
You can't win. It's like one of those Star Trek episodes where Captain Kirk destroys the rogue computer by putting it in a logical loop (link). If you watch closely at tech conferences, you can see the smoke seeping out of the ears of telecom execs.
Faced with this dilemma, those telecom execs naturally are trying to find a third option: Get someone else to pay for mobile data. There are a couple of options:
Option 1: Have the government pay for mobile data
I doubt that most governments would pay to make wireless data completely free for everyone, but I was surprised when I found out how much governments are already paying for mobile. For example, the US government subsidizes mobile phone service for millions of unemployed people (because it helps with their job searches; they need phone numbers so employers can all to offer them jobs). I could easily imagine that benefit being extended to include mobile data, on the assumption that poor people need access to job boards (how we'll avoid paying for their YouTube and Kongregate usage I don't know).
Governments are also being lobbied to give special regulatory treatment to wireless data. The rhetoric around a "wireless spectrum crisis" is being used to influence governments. The focus of this lobbying in the US is on taking spectrum away from the TV networks and supplying it to the mobile operators. Effectively that is a financial subsidy for the operators -- if the government forces the transfer the operators will have to pay less, and will get the spectrum faster, than if they were to purchase it on the open market.
Here's how the lobbying works. This is an excerpt from an e-mail sent to me recently by a PR firm working for a group called the Internet Innovation Alliance:
"IIA's Blog: The Spectrum Clock is Ticking
Writing for Forbes, Lawrence J. Spiwak, President of the Phoenix Center for Advanced Legal and Economic Public Policy Studies, warns Congress that more spectrum needs to be freed up for mobile broadband and it needs to be freed up soon: Like it or not, the clock is ticking on spectrum exhaustion, both for consumers and our public safety professionals. Unless we want a market characterized by higher prices, failed data sessions, dropped calls and stifled innovation, policymakers need to implement a cohesive spectrum policy with a large degree of urgency."
At first glance, that reads like a standard plea from a bunch of web companies worried about the mobile network getting overloaded. But the backstory is that both IIA and the Phoenix Center are reportedly funded by the mobile operators (link, link). So this isn't an independent assessment of the situation, it's the operators sending us a message. And the message is: "Give us more bandwidth or we'll trash your phone service." I think that's a bit disingenuous -- unless the operators seriously mismanage their networks, we won't end up with both bad service and higher prices. But they're right that without more spectrum we'll definitely get one or the other.
Option 2: Make web companies pay for mobile data
Several of the leading operators in Europe recently argued that big tech companies like Apple and Google should be forced to pay to use the wireless networks. Although they don't put it this way, they're asking the big Internet companies to subsidize mobile data plans for users (link).
The CEO of Telefonica said the web companies "use Telefonica’s networks for free, which is good news for them and a tragedy for us. That can’t continue."
Here's the CEO of France Telecom (link):
"The real risk of everything is collapse. Nobody utters this loudly enough, but the real issue for the world is a collapse of the network or some local collapses. We are the people with pipes. We are supposed to invest heavily in pipes in order to bring the capacity which is necessary to sustain the explosion of consumption and usage and data traffic in our networks. At the same time, the people that create this traffic…are not really incentivized to manage properly, globally, the traffic. There is an unbalance in the overall system, which in our view is a major problem. It is totally impossible to absorb such an explosion in traffic without first, clearly investing massively in spectrum and equipment, and second, without introducing some new pricing approaches."
This is the heart of the whole debate about net neutrality. I believe it's not really about mobile operators trying to give an advantage to their own services, it's mostly about the operators trying to open up a second revenue stream because they're afraid they can't get enough revenue from users to support future growth.
For the operators, charging web companies a fee seems intensely attractive because the fee could be scaled to the amount of traffic they generate (unlike the flat-rate data plans that users prefer), forcing the web companies to use bandwidth more efficiently. It also would let operators increase their revenue without directly reducing user demand. Basically, the web companies would subsidize a shift from wired to wireless computing.
The third option
I can see why the operators are pushing on both of these options. They're in a difficult situation, and it would be very helpful to them if somebody bailed them out (link). I might be trying the same things if I worked for an operator. But there is a third option for managing cellular data overload, and it deserves to get a lot more attention:
In almost every other industry in the world, you're responsible for charging enough money to support your business. Yes, sometimes you have to make investments before you know how much demand there will be, and yes, sometimes that creates a lot of risk for your company. But that's why they pay you the big bucks, Mr. or Ms. CEO.
I don't understand how we as a society came to the conclusion that wireless data should be different. Is there some religious commandment that people must be allowed to stream Netflix on the subway? Or maybe those big Cisco growth forecasts have led us to think that endless growth of mobile data is a ravenous beast that will cause immense suffering if it's not fed more bandwidth.
Baloney. If the network is overloaded, raise your prices until you either get enough money to expand the network, or you force people to use less data. If you want network bandwidth used more efficiently, show users the cost of the data they use and they'll demand more efficient apps and devices on their own.
I bet a price increase from $50 a month to $80 a month for mobile data would end the bandwidth crisis overnight.
Not only is higher pricing the simplest way to manage network overload, it's going to happen no matter what we do. Even if we give the operators all the bandwidth from the TV networks, and get the web companies to subsidize wireless service, all that will do is delay the crunch for a few years. More traffic will switch from fixed-line to wireless until once again the network saturates and prices go up. It is inevitable.
What it means
When we plan for the future of mobile, we need to be realistic, and a little bit humble, about what we can change and what we can't. We need to learn to live with the things we can't change, and focus on doing a good job of managing the things we can.
Here's my list of the things I think we can't change about mobile data because they are driven by economics and physics:
--Most data traffic will be wireless only for the last 100 feet (30 meters) that it travels from your device to the nearest hotspot (whether it's WiFi, femtocell, or something else). So we need to be careful about our terminology. Most data could well be technically "wireless" in the sense that it passes through WiFi at the end, but that is a meaningless distinction for the purposes of this article; most of it won't pass through the cellular data network.
--Most of us will continue to have some sort of broadband cable connecting to our homes and offices, or to a point very close by (like the lamp post in the street outside your window). Forget those visions of cellular replacing the wired broadband network; in the developed world it can't happen.
--The cost per-byte of cellular data will be significantly higher than the cost per-byte of wired data. The difference will be large enough that we'll be aware of it and it will alter the way we use our devices.
--Flat rate unlimited cellular data contracts will go up in cost, or will be replaced by much more variable pricing for most users. This is already underway at some operators. For example, Verizon is rumored to be about to move from $30 per month for unlimited data to a tiered plan that ranges from $30 per month for two gigabytes to $80/month for 10 gigs (link). I don't usually like consumer price hikes, but in this case the change is long overdue.
--As the relative cost of mobile data rises, most of us will use cellular data primarily as a supplement to the wired network when we're on the go. We'll become religious about turning on WiFi in our smartphones and tablets, and making sure it can connect at home and at work. Because cellular data is more expensive, many of us will try to avoid using very data-heavy apps on the cellular network.
This means cellular data use won't be carefree. That may not sound like a big difference, but in consumer terms I think it is. We've been making the assumption that cellular data can directly replace fixed-line data, just as cellular phones replaced fixed line phones for many people. "Go ahead! Use it anywhere! Be free!" But for an aggressive user of mobile data, that can't happen. Our use of cellular data is going to be much more nuanced, managed, and carefully thought out than our use of cellular voice. I think many of us will look at our cellular data budgets the same way we look at our automobile budgets. Some people will spend more, some less, but I think most of us will be aware of the cost and manage it actively.
The wired Internet will continue to set the tune. The ongoing role of fixed-line broadband means that many leading-edge web apps will continue to be designed around the capacity and responsiveness of fixed-line networks. This is another subtle but very important difference, because it means the mobile operators will continue to play catch-up to customer expectations set on the wired networks. There will be exceptions; some features of cellular data (such as location) will drive unique mobile apps. But in most application categories, rather than shaping the future of the Internet, mobile operators in the developed world will be pushed to deliver an Internet experience that evolved on fixed-line networks.
Here are the things I think we can change about cellular data:
--We can alter the share of total data traffic that moves through the cellular networks. By transferring spectrum and giving the operators other favorable treatment, we can make the overall capacity of the cellular data networks higher than it would have been otherwise. Basically, we can make the mobile operators bigger. That may delay the onset of mobile network congestion, and enable some classes of web applications to be more successful in cellular (for example, low-res video streaming). That can have a big impact on individual users and app companies. It will also have a big impact on the ultimate revenue and profitability of the mobile operators, which is why they are lobbying so hard.
--We can probably change the size of the average mobile data bill, but only temporarily. The more revenue streams we give to the operators, the more mobile data we'll probably get for a given user price. However, as I mentioned above, keep in mind that if mobile data is made cheaper, people will use more of it, which will eventually saturate the network and cause prices to rise. So any money we save on our mobile data bills will probably be temporary.
--The decisions we make in the next few years will profoundly change the economic structure of the wireless data industry. Changes in regulations and pricing rules will have a huge impact on the ability of small companies to compete with large ones in mobile, and will determine who pays for the whole thing. This could decide whether the mobile internet looks more like the wired Internet (low barriers to entry, lots of companies) or cable television (high barriers to entry, dominated by a few big players).
I think the most important thing about the three points above is that they're all driven by government regulation. The rules we set for the mobile Internet are going to determine the ultimate size of the mobile operators, how they are funded, how competition works in mobile data, and how much power is held by the various players.
That scares me. I prefer to have winners and losers in a market chosen by customer decisions, not government ones. You can't blame the mobile operators, or the big web companies like Google, for lobbying the government on these issues. But I don't think their interests are necessarily the same as the rest of the industry, let alone consumers. Also, most of the big players are driven by quarterly revenue, and in some cases they are pushing for changes that I think will help them in the short term but would actually hurt them in the long run.
I wish there were some scenario in which we could tell governments just to butt out and let the market decide, but governments are already deeply involved in allocating spectrum, and there's no practical way to undo that. So I think it's important that we all have a very thorough, open discussion of the government decisions to be made and the sort of wireless industry they'll produce.
That's what I'll cover tomorrow.
In part 3, which I'll post tomorrow (link), I'll give my ideas on how we should structure the mobile data market. I'll also talk about the opportunities this new world of mobile data will create for companies in mobile. In the meantime, I welcome your comments and questions. This is a big, complex issue, and I don't pretend to have it all figured out.