All Lessons

Filter By Learning Category Below

Filter By Keyword Below

Looking for this article recently at David Isenberg’s site, we were surprised to find that the pointer to it was broken. To help keep these memes on the Web, here’s a copy.

The Rise of the Stupid Network was originally written during a single long weekend in May 1997 by David S. Isenberg while he was an employee of AT&T Labs Research. It was officially released onto the Internet by AT&T in June 1997. It was printed (without permission, with some bad editing, and with the last few hundred words truncated) by Computer Telephony in its August 1997 issue.

If you want to cite an official, public, printed version, cite Computer Telephony, August 1997, pg 16-26.

Isenberg left AT&T in January 1998. Shortly after that, AT&T asked Isenberg to remove The Rise of the Stupid Network from the website. Isenberg wanted to make AT&T happy so he took it off. (It didn’t work; AT&T is still not happy.)

However, you can still find The Rise of the Stupid Network
at these two fine websites:

In addition, there is a Japanese translation (in several formats) by Masakazu Sato at: (Japanese PDF), (Japanese MS WORD), (GIF Image, works without Japanese type fonts.)

Find the Korean translation by Kwon Hyung Roh at:

Please come back to when you’re done reading . . .

Why the Intelligent Network was once a good idea,
but isn’t anymore. One telephone company nerd’s
odd perspective on the changing value proposition

David Isenberg – –


Design-by-assumption works as long as assumptions hold. Assumptions are shortcuts to useful efficiencies, provided they are not violated. The classic telephone company value proposition, embodied in today’s telephone network, holds:

  • that expensive, scarce infrastructure can be shared to offer premium priced services,
  • that talk – the human voice – generates most of the traffic,
  • that circuit-switched calls are the “communications technologies” that matter, and
  • that the telephone company is in control of its network.

Telephone companies still behave as if these assumptions hold despite:

  • up to several thousand-fold declines in key infrastructure costs over the last two decades,
  • a 20 year double-digit annual growth rate in the volume of data traffic, so that the volume of data traffic is now overtaking the (also growing, but more slowly) volume of voice traffic,
  • the many different data types that now travel over the telephone network (despite the fact that the network is not optimized for all these data types),
  • the many different types of “communications technologies,” from television to Ethernet, that are not part of telephone network architecture, and
  • the Internet, which, because it makes the details of network operation irrelevant, is shifting control to the end user.

The Intelligent Network is a straight-line extension of the four assumptions above – scarcity, voice, circuit switching, and control. Its primary design impetus was not customer service. Rather, the Intelligent Network was a telephone company attempt to engineer vendor independence, more automatic operation, and some “intelligent” new services into existing network architecture. However, even as it rolls out and matures, the Intelligent Network is being superseded by a Stupid Network,

  • with nothing but dumb transport in the middle, and intelligent user-controlled endpoints,
  • whose design is guided by plenty, not scarcity,
  • where transport is guided by the needs of the data, not the design assumptions of the network.

The Stupid Network is not all here yet. It is in its infancy. It needs to get stronger and, well, a bit more coordinated.

Some telephone company people realize that things are changing, and must change. But they are hemmed in by conscious, deliberate, long established telephone company practices. Many are also hobbled by less conscious telephone company mental models of “communications,” “technology,” and “customer needs.” While these people may realize that the old ways are becoming obsolete, they live in a world conditioned by an encompassing, arcane legacy that only remembers “rational,” incremental change.

(Note: here “telephone company” refers to large companies whose main business is to provide circuit switched voice calling service. In the United States, most of these are the heirs of the Bell System legacy – but Sprint, MCI, GTE, SNET, and others might also try on this shoe, and if it fits…)


It used to be more expensive to complete telephone calls than it is today. The operator-completed call gave way to call completion by electro-mechanical switch. Then, in the late 70s, the era of computer controlled electronic switching made placing calls even cheaper and more reliable.

In those days, computers, including those that controlled switching, were still considered expensive, scarce resources. When I worked in the nascent electronic toy industry in 1979, a single insight that eliminated six transistors paid my way. And the same factor – the need to save two expensive bytes of memory – laid the basis in this era for the Year 2000 Problem (stay tuned to the eleventh hour news for more on THIS story!).

Now computer circuits are thousands of times cheaper. Moore’s Law is what we call the ongoing improvement in computing cost and power. But in the 70s it was not generally known to be a `law’ – to most telecommunications engineers (and to humanity in general), it has become the most game-changing wild card played in recent times.

Telephone networks have been designed for optimal use of scarce resources. The local exchange in your city, which handles the last four digits of your telephone number, theoretically could handle up to 10,000 telephones, e.g., with numbers 510-547-0000, 0001, 0002, et cetera through 510-547-9999. But the switching office is not designed to handle 10,000 simultaneous calls. It is designed to handle far fewer, maybe one tenth of that, based on the assumption that even in the busiest time of the day, only a fraction of its telephones will be active at any one time.

The network works as long as engineering assumptions (e.g., the length of a call, the number of call attempts, etc.) do not change. But let the assumptions change episodically (e.g., Rolling Stones tickets go on sale), or structurally (calls to Internet service providers last several times longer than voice calls), and the network hits its design limits – completing a call becomes a matter of try, try again.

What if network design were based on another assumption – that computation and bandwidth were cheap and plentiful?


Once the telephone companies began doing digital switching, the idea that you could do “intelligent” things with calls was not far behind. The concept of network control was extended to let various centralized resources – digital switches, databases (Service Control Points) and signal processing systems (Intelligent Peripherals) – communicate among each other by extending the telephone network’s control protocol (SS7).

As noted above, the main force motivating the Intelligent Network was a telephone company attempt at “vendor independence” so telephone companies could get better deals from their suppliers. Thus, Intelligent Network specs were meant to encourage vendors to design their equipment to work in a multi-vendor environment – to interoperate. As a side benefit, almost an afterthought, some of the newly specified equipment could also interoperate with the business systems of certain customers – but only via limited, cautiously designed interfaces. Virtually all of these services center around call completion, automation, and billing. This, in a nutshell, is the concept marketed as the Intelligent Network. Some Intelligent Network service examples include:

  • Routing calls to different numbers than the one that the caller originally dialed (this is the basis of e.g., 800 service).
  • Giving caller choices before the call is completed (“push one for domestic reservations,” etc.).
  • Saying, “Calling Card, Collect, Third-Party, or Operator” to control payment options.
  • Verifying that the calling card number is valid in “real time.”
  • Supplying calling party numbers directly to customers for database lookup (which is why I must verify from my home phone that I got my Citibank card in the mail).

Expensive computers, intertwined in central network operations, do this. Belief becomes reality. But wait! The telephone companies are now losing design hegemony – the news that “The Internet is here!” is beginning to penetrate the telephone company inner sanctum.


The astute reader might by now suspect that the main beneficiaries of the Intelligent Network are the telephone companies themselves. Nevertheless, telephone companies propound a “philosophy” that the Intelligent Network makes it easy to introduce new services and new technologies, and to meet new customer needs.

New customer needs, when they are detected, filter into the telephone company slowly. Some needs, the ones with big, obvious, immediate payoffs, get attention from decision makers, who then request a business case, which must then get approved. The next step is the development plan, followed by the Operations, Administration, Maintenance, and Provisioning Plans. Then if all goes well, the telephone company might begin the process of implementation. This can take years, or even decades (witness ISDN).

If you hate hanging on hold, you are part of a huge latent market – do you know anybody who doesn’t? Yet, telephone companies have yet to use Intelligent Network capabilities to effectively ameliorate this problem. Now, suppose Internet Telephony gets as good as telephone company telephony (see below), and some enterprising independent programmer wants to make a product that solves the problem of being on hold. They would simply write an end-user application and sell it from their web site. If it works, and people like it, they will sell lots of it. If not, they might try again. But they don’t have to go through any long, bureaucratic economic justification, business planning, and technical development processes – they just do it. Internet Telephony, because the Internet Protocol works at the level that user software manages the session, takes the telephone company out of the value equation.


The Internet breaks the telephone company model by passing control to the end user. It does this by taking the underlying network details out of the picture.

Let’s look at how this works in the case of voice. To the telephone company, there is one main way of transmitting voice – sampled in 8 bit bytes, 8000 times a second, for an aggregate rate of 64 kbit/s. The entire telephone network is designed around this rate. But if you want to send voice on the Internet, you can encode it at any rate you want, and send it at any rate up to the one that the slowest underlying network link supports. The recipient must have the right decoder running in her intelligent terminal, too.

The very name, Internet, denotes that it is designed to network networks. You can use Internet Protocol on an Ethernet to communicate with an X.25 network, an FDDI network, or a modem – lower layer protocols are submerged, made irrelevant. So if you are on an (e.g., 10Mbit/s) Ethernet, and your endpoint application wants to send better quality 256 kbit/s voice, no problem. You can’t do that with the telephone network.

Or, with a different application (on the same endpoint and network) you can send six different interwoven 10 kbit/s voice streams to six different destinations at the same time. And you don’t have to tell your Stupid Network provider anything about it, or pay a premium to install anything special. The network provider becomes virtually irrelevant – the user controls the relevant capabilities.


I contrast the flexibility of a Stupid Network with my experience as a member of AT&T’s True Voice technical team. AT&T True Voice was a valiant attempt to improve circuit switched voice quality as much as possible in the context of current network architecture. If we had not been constrained by network architecture, the easiest way would have been to increase the sampling rate or change the coding algorithm. But to actually do this, we would have had to change every piece of the telephone network except the wires. So we had to work within the designed 64 kbit/s data rate.

An astute AT&T perceptual psychophysicist (and a friend of mine) determined that voice quality could be substantially improved by boosting the bass part of the signal, that part of the audio spectrum between 100 and 300 cycles per second. But as we set out to implement this conceptually simple improvement, we kept running into the problem that there were too many places in the network that had built in “intelligent” assumptions about the voice signal – echo cancellers, conference bridges, voice messaging systems, etc. – and too many devices that depended on these acoustic assumptions for their correct operation – modems, fax machines, and a surprising number of strange devices with proprietary analog protocols. After about two years of intense effort, we made a noticeable difference, one that most listeners preferred (if asked explicitly), but it was not as large as it could have been. There was too much “intelligence” intertwined with the basic transport.

The True Voice experience led me to see the advantages of a network – a Stupid Network – that would let you stuff bits in one end and get them out the other without getting tangled up in cobwebs of legacy assumptions. Want a different voice quality? With a Stupid Network, you’d get a different program, install it in your intelligent end user device and run it.


There is no longer first-order economic justification for a telephone company to engineer and control scarce, expensive, network resources – the basic conditions no longer obtain. The age of plentiful computing is here. I have a multi-color, three dimensional screen saver that uses the entire capacity of my 200 MHz Pentium. The designers of the Intelligent Network never imagined such “wasteful” use of processing “intelligence.” The age of plentiful bandwidth is just around the corner, as several families of technologies (fiber, satellite, cable modems, xDSL, LMDS, and low power TV, to name just six) line up to break the local bandwidth bottleneck, and as the capacity of backbone fiber has risen from 2 to 6 to 10, 20 and 40 Gbit/s over just the last few years.

The age of centralized control is ending too, with the rise of the next generation of Internet – and especially the appearance of circuit-like Internet mechanisms, such as those in the latest version of Internet Protocol (IPv6), designed to tame delay and improve real-time two-way Internet voice.


A new network “philosophy and architecture,” is replacing the vision of an Intelligent Network. The vision is one in which the public communications network would be engineered for “always-on” use, not intermittence and scarcity. It would be engineered for intelligence at the end-user’s device, not in the network. And the network would be engineered simply to “Deliver the Bits, Stupid,” not for fancy network routing or “smart” number translation.

Fundamentally, it would be a Stupid Network.

In the Stupid Network, the data would tell the network where it needs to go. (In contrast, in a circuit network, the network tells the data where to go.) In a Stupid Network, the data on it would be the boss.

Instead of fancy “intelligent” network routing translation, in a Stupid Network, intelligent end-user devices would be connected to one or more high speed access networks – always listening for relevant information, for data addressed to their owner. Sometimes a “communication” might be a few bits, perhaps a short, pager-type message. Other times, it might be longer, like email. In the event of the need for two-way voice communication, an initial message might state the identity of the “caller,” and/or inquire of the whereabouts of the owner. The intelligent end-user device could apply its knowledge of where its “owner” was, and who the caller was. Then, if it were programmed to do so, it could launch a message to its owner, telling of the call, the caller’s identity, location, and any other information. It could also forward as much information as practical.

End user devices would be free to behave flexibly because, in the Stupid Network the data is boss, bits are essentially free, and there is no assumption that the data is of a single data rate or data type.


In the current telephone network, voice is the assumed data type, unless specially ordered, high cost services are ordered. But in the Stupid Network, because the data is the boss, it can tell the network, in real time, what kind of service it needs. And the Stupid Network would have a small repertoire of idiot-savant behaviors to treat different data types appropriately. If the data identified itself as financial data, the Stupid Network would deliver it accurately, no matter how many milliseconds of delay the error checking would take. If the data were two-way voice or video, the Stupid Network would provide low delay, even at the price of an occasional flipped bit. If the data were entertainment audio or video, the Stupid Network would provide wider bandwidth, but would not necessarily give low delay or absolute accuracy. And if there were a need for unique transmission characteristics, the data would tell the Stupid Network in more detail how to treat it, and the Stupid Network would do what it was told.

The Stupid Network would let you send mixed data types at will – limited only by the knowledge and imagination of the application programmer community. One way voice messages, multi-way voice conferences, two-way video, email, documents, audio and/or video entertainment, whatever, could be mixed and interspersed at will, within and between sessions. You would not have to ask your Stupid Network provider for any special network modifications – its only function would be to, “Deliver the Bits, Stupid.”

One thing about the Stupid Network is clear – the physical elements that comprise the network would be neither expensive nor scarce. There would be little profit margin in shipping dumb bits. There would be lots of high value Business Ideas supported by the Stupid Network, above and beyond transport.


A rudimentary form of the Stupid Network – the Internet – is here today. The telephone companies are beginning to realize this. Fearing erosion of their control and, more importantly, their revenue stream, they have been quick to call for the banning of Internet Telephony, quick to call for the federal imposition of charges on Internet access, and slow to implement widely available, reasonably priced broadband services. This creates a chicken and egg problem – while the hungry wait for dinner and breakfast.

A powerful leading indicator of the Stupid Network will arrive when entrepreneurs who have no vested interest in maintaining telephone company assumptions begin to offer profitable, affordable, widely available data services. Watch Metricom’s Ricochet modem service, an early entry in this market. Will entrepreneurial broadband service follow? There are several early efforts, for example, Sky Station International, which plans to launch self-propelled balloon-based transcievers over major cities to deliver personal 1.5 Mbit/s service. Meanwhile, we will see how advances in Internet Technology (such as IPv6 and the Internet II initiative of leading universities) evolve – here the ability of the Internet to offer low delay services, such as two-way voice, is the key indicator.

To counter these threats, the telephone companies are now speeding deployment of Intelligent Network services, much like sailing merchants responded to the threat of steam by inventing faster sailing ships in the mid 1800s. The beneficiaries of this accelerated Intelligent Network deployment are big businesses – who can offer cheaper help-desk type services with lower human labor costs. Nevertheless, despite this current Intelligent Network buy-in, if big business finds that it is better served by the Stupid Network and premises based intelligence, it will not hesitate to switch.

The Telecom Act of 1996 and the World Trade Organization telecom agreement of 1997 can be seen as attempts to preserve oligopolistic hegemony of the telephone companies. The thrust of both is to allow big companies to band together to create a marketplace dominated by a few large players in place of government control. Will there be unintended consequences of these agreements? Count on it! Will they hasten or impede the advent of the Stupid Network? Hmmm.


The shift from scarcity to plenty is often the harbinger of new value propositions. For example, as computer power got cheaper and cheaper in the 1980s, there was much talk of a shift in value from hardware to software, but it was not easy to see how the shift would unfold. In fact, it appears that only one person (Bill Gates) understood it fully. The changes that now portend the Stupid Network are likely to shift the telecommunications value proposition from “network services” to something else. If I knew what it was, I would not be wasting my time writing these words.

Given that disclaimer, I have three brief observations:

  1. It is rare that a market is completely killed by the next generation of technology. Neither TV nor the VCR killed the movies. Neither the minicomputer (alas, remember them?) nor the PC killed the Mainframe. We still have ships and railroads, though their markets are both diminished and changed by the car and airplane. The “paperless office” exists – but mine is cluttered with books, memos and magazines that are printed on paper. So it is likely that the Stupid Network and the Intelligent Network will exist side by side for some time, or even share merged definitions, functions, and value. It is also likely that “deliver the bits” companies will exist in a Stupid Network world, but given much lower profit margins, they will not look much like telephone companies.
  2. Telephone companies themselves could cannibalize their own product. Smarter companies often field new products that replace current profitable product.
    • Sony does this several times a year – it tries to learn from its own mistakes faster than its competition, fielding new products that improve on its old before such improvements become obvious to their foe.
    • Boeing does it – the 757 and 767 cut into the top of its 727 market and the bottom of its 747 market with fuel efficient, and crew efficient new designs – we can only hope that Boeing does not become complacent now that it is has beat out its strongest competitors.
    • Intel does it – having been the first to articulate Moore’s Law, it now drives it with a new, more powerful chip every 18 months or so, long before the old chip is obsolete – it realizes that if it stops, there are other chip makers that would be glad to take leadership of that market.
    Telephone companies could do it too, but it is unlikely as long as their senior managers prefer to talk with lawyers, regulators, consultants and financiers more than with experts in their own employ.
  3. Telephone companies could reinvent a place for themselves as purveyors of new values propositions brought by the Stupid Network. They will have to, because their old value proposition will erode as the Stupid Network grows. In a “deliver the bits” world, so much information, and so many courses of action, will be available, that there will be a great need for known, trusted authorities. Businesses with brand reputation and staying power will be guarantors of transactions, holders of critical information, organizers and filters of information, and even voices of reason, leadership, and “objectivity.” (Of course, they will need to HAVE reason, leadership, and objectivity to do this.) There will be other roles for big companies in the world of the Stupid Network, and “forgetting organizations,” who are able to abandon old models when new ways no longer support old assumptions, will find them.


Former Shell Group Planning Head, Arie deGeus, in his master work, “The Living Company” (Harvard, Boston, 1997), examined thousands of companies to try to discover what it takes to adapt to changing conditions. He found that the life expectancy of the average company was only 40 years – this means that telephone company culture is in advanced old age. De Geus also studied 27 companies that had been able to survive over 100 years. He concluded that managing for longevity – to maximize the chances that a company will adapt to changes in the business climate – is very different than managing for profit. For example, in the former, employees are part of a larger, cohesive whole, a work community. In the latter, employees are “resources” to be deployed or downsized as business dictates. As the Stupid Network arrives, as the business idea shifts from scarce physical infrastructure to something more knowledge based, company culture will need to adapt to the truth that, “Nobody knows as much as all of us.”

Whatever we discover to be the new Stupid Network value proposition, my working hypothesis is that it will be based on intelligent end user devices, intelligent customers, employees whose intelligence is valued as a corporate asset, and companies that can learn.