Futurists, especially those who claim to have a methodology beyond psychic prediction, tend to rely primarily on Moore's Law for figuring out what technology will be like 5, 10, or 20 years from now. But Moore's Law, which predicts that computing power will drop in cost by 50 percent every 18 months, isn't some absolute speed limit and some measures of technical achievement are actually moving faster than Moore's Law predicts. It's not that they violate the law (if it even IS a law) but that they take advantage of loopholes like the one where Moore's Law can sometimes be applied TO ITSELF, leading to even faster change. That's what's about to happen to television and why your new digital television probably won't matter much as a technical standard after 2015.
In order to understand this better let's first look at the broadband Internet market. I'm an American employed as a cranky technologist by an American television network, so this is pretty U.S.-centric, but there are lessons here for many countries.
Ten years ago, the United States had the fastest and cheapest residential Internet service in the world. Today U.S. residential Internet service, especially broadband, is among the slowest and most expensive. Fortunately, this is likely to change as U.S. broadband Internet services become decidedly more competitive, both in terms of cost and available bandwidth. Unfortunately, U.S. broadband adoption rates are slowing at a rate that suggests ultimate market penetration under 90 percent.
Japan went from being among the most expensive countries for residential Internet bandwidth a decade ago to absolutely the cheapest today. While some of this change can be attributed to technology improvements, most of the change can be attributed to competition, specifically the entry of Softbank BB into the Japanese broadband market. Softbank BB entered the Japanese market early this decade with loss-leader pricing that forced all the incumbent broadband suppliers to respond in kind, leading to a dramatic expansion of the Japanese broadband market where today residential 100-megabit-per-second service costs less than $20 per month.
This Japanese model does not apply well to the U.S., where there is no broadband provider willing to take the bet-the-farm approach of Softbank BB. The U.S. market also has no true national broadband ISPs that operate on a scale comparable to those in Japan. And the topology of the U.S. Internet is such that the high-bandwidth technologies applied in Japan would not work as well here simply because of a larger rural customer base.
Korea, as it is often wont to do, followed Japan in terms of bandwidth pricing. More importantly the government of Korea made it a national priority to build out the residential Internet infrastructure at government expense. This was, ironically, in part inspired by the U.S. National Information Infrastructure plan, which was intended to accomplish the same end but failed miserably. Though they took full advantage of $150 billion in tax credits, the U.S. telcos simply did not build the network they had agreed to build, yet their model inspired more successful efforts in Korea, Singapore and other Asian markets.
Of the 30+ nations that can be judged to have residential Internet service superior to the U.S., in case after case that superiority can be attributed to government funding of infrastructure, to largely urban (short-distance) topologies, or to aggressive competition.
In the United States, unlike most of the rest of the world, broadband Internet service has been dominated by cable television companies offering cable modem service on their hybrid fiber-coax systems, with telco xDSL service a less popular broadband alternative until very recently. Cable Internet service was originally coordinated on a national basis through Excite@Home, but with the failure and liquidation of that company in 2001 most cable systems were left to fend for themselves as ISPs, with varying levels of success. Since few cable TV systems have competition and regulation has tended to concentrate on television -- not data -- service, changes in price and available bandwidth have been generally dictated by whatever competitive broadband offering came from the local telephone company.
It is important to realize that bandwidth has not been lacking for U.S. cable ISPs, which typically devote to Internet service the bandwidth of one analog channel (usually Channel 80) on their systems. By adding a second data channel or (more often) segmenting their network into subnets, cable ISPs have plenty of aggregate bandwidth at their disposal but have simply not been challenged to provide it given that competitive telco products have been, until recently, limited to 1.5-megabit-per-second downloads.
Technology improvements and business model changes among broadband ISPs appear to be finally leading to significant changes in the U.S. residential broadband market. Technology is always advancing, of course, and the technologies coming into play are DOCSIS 3.0 on the cable Internet side and various forms of Fiber to the Home and Fiber to the Curb among the telcos. The business model changes involve so-called 'triple play' services where ISPs hope to make money from providing not just Internet service, but also telephone and television. The cable TV companies want to steal from the telcos basic phone service while the telcos want to steal television service from the cable companies. Since either possibility requires advanced data services and more bandwidth, users benefit.
DOCSIS 3.0 services will begin appearing shortly, offering up to 150 megabits per second, though it is doubtful that many cable ISPs will jump straight to that level given the emerging xDSL telco limitation of 24-26 megabits per second.
The telcos, notably AT&T and Verizon, are aggressively building out their fiber plants. Verizon is taking fiber directly into the home, but AT&T is taking its fiber only as far as the curb . This ostensibly limits AT&T to xDSL speed limits, though the company can use channel bonding (more than one pair of copper wires per service) to increase speeds if forced to do so by competition. Verizon is rolling out residential fiber service from 30-50 megabits per second but its equipment can jump to 100 megabits per second if needed without requiring another truck roll.
An important secondary motivation for this fiber rollout is that telcos are not required to share such facilities with competitors as they have been required to share copper infrastructure under the Telecommunications Act of 1996. So while there may be competition in the neighborhood from cable modems, once the fiber is in and the copper is out the telcos need never again fear competition from Competitive Local Exchange Carriers (CLECs).
Over time there will be other types of broadband ISPs that may provide competition and thereby spur service improvements. One possibility is WiMax, but WiMax is NOT a service that can compete for true high-bandwidth (above 10 megabit-per-second) service on an economical basis. The same applies for so-called 3G and emerging 4G wireless data services from cellular phone companies, which are also limited in total aggregate bandwidth.
While the number of U.S. residential broadband users is continuing to increase, the rate of that increase is slowing according to several surveys by the Pew Internet & American Life Project. Extrapolating these numbers suggests that ultimate broadband penetration will be comparable to cable TV, or around 85 percent. This slowing of growth may be inspiration for the growing telco vs. cable battle over triple play digital services, with the idea that some telephone users (where market penetration is already 97+ percent) will be induced to buy broadband service to lower their telephone costs.
There is a base cost of around $20 per month for providing broadband Internet service irrespective of the allocated bandwidth based on published cost assumptions of international ISPs. This $20 is essentially the ISP overhead and is unlikely to decrease significantly no matter how inexpensive bandwidth, itself, becomes. So while there has been some moderation in broadband subscription rates in recent years, it has been minimal, with ISPs generally using introductory specials, rather than permanently lower rates, to attract customers. Broadband customer churn is minimal, probably due to local monopolies and not wanting to give up ISP-branded e-mail addresses, so there is not significant price pressure. Rather, there is pressure to provide greater bandwidth at the same price. Nearly any U.S. residential pricing model, then, will have a base subscription cost of around $30-40 per month but with the ISP allocating increasing amounts of bandwidth for that unchanging payment. In this instance the ISP is generally hoping to make much of its profit on value-added services like Voice-over-IP phone service or movie downloads. After staying for years at an average 1.5-megabit-per-second download speeds, broadband ISPs are moving to an average of 6 megabits per second in 2007-2008, 24 megabits per second in 2010-2012, and 100 megabits per second in 2014-2016.
Okay, so that's the bandwidth picture, but what will we do with it? Here's where Moore's Law reenters the picture. As processors get more and more powerful they will migrate into many consumer electronic devices, especially televisions. It's not that televisions will become computers but that televisions will become more and more computerized. This leads to a very interesting effect. As we've seen above, Internet bandwidth growth is already defying Moore's Law, growing at 100 percent per year with roughly flat prices. Add to this increased processing power in televisions and we'll get decreasing bandwidth requirements as televisions will be able to run more powerful codecs. Where many PCs today can't do real-time 1080p video decoding in software, simply throwing in another generation or two of Moore's Law will eliminate that problem completely.
So bandwidth will get cheaper and cheaper while our entertainment devices will be doing more and more with available bandwidth. Add to this the slow evolution of video standards and we'll have cheap bandwidth and even cheaper processing power colliding with the wall that is the 1080p HDTV standard, leading to ultra-low per-stream costs for entertainment providers.
To explain this another way, traditionally industry would react to such cheap bandwidth by jumping us all to 2K or 4K displays or maybe 3D, and that might well happen eventually. But until then we've decided as a technical culture that 1080p -- so-called Full HD -- is as good as it gets. Jumping beyond 1080p will require years of haggling and during that time 1080p will become cheaper and cheaper and cheaper to do, leading to that inflection point alluded to in this week's column headline.
Around 2015 is the time when the cost of sending a separate 1080p video signal to every Internet-connected viewer -- or POTENTIAL VIEWER -- will be the same as using a broadcast model and sending that signal through the air. After 2015 there will be no scaling limits, no processing limits, no decoding limits. And since individual video streams mean individual commercials with a requisite CPM (cost per thousand) bump of up to 10X, commercial television as we know it will die, replaced by consumers choosing from a menu or recommendation engine what they want to see when they want to see it.
Just follow the money.
Commercial stations will repurpose their bandwidth for alternate wireless services, eventually shutting down their digital transmitters completely. And PBS, which can't create a marketplace all by itself, will follow.
I'm not saying here that you shouldn't buy that new DTV, because it will fit into most any emerging system. But I am telling you that the era of the television programmer, where some guy at the network or down at your local station thinks he knows in what order and on what days the audience really wants to watch TV, well that era will be gone forever, seven years from today.
"I, Cringely . The Pulpit - http://www.pbs.org/cringely/pulpit/