Blog Post

The Voice over IP Insurrection

Daniel Berninger, an old friend, a seriously smart guy and VoIP guru of sorts, and more recently senior analyst, for Tier1 Research, has been a great man to bounce ideas off. He and I have chatted about many things, and each time I come away learning something new. So last week he argued, “in the battle between Bellheads and Netheads, we’re all Netheads now.” Could not agree more. Here is his long missive on the VoIP insurrection, the best and most definitive essay you will ever read on this technology, where it is headed and why it is important. This is the second of my guest columns series where I bring the experts who know a thing or two about their respective areas of expertise.

What just happened?

The $3 billion dollar budget at Bell Laboratories did not include a single project addressing the use of data networks to transport voice when VocalTec Communications released InternetPhone in February 1995. As of 2004, every project at the post-divestiture AT&T Labs and Lucent Technologies Bell Labs reflects the reality of voice over Internet Protocol. Every major incumbent carrier, and the largest cable television providers, in the United States has announced a VoIP program. And even as some upstart carriers have used VoIP to lower telephony prices dramatically, even more radical innovators threaten to lower the cost of a phone call to zero—to make it free.

The VoIP insurrection over the last decade marks a milestone in communication history no less dramatic than the arrival of the telephone in 1876. We know data networks and packetized voice will displace the long standing pre-1995 world rooted in Alexander Graham Bell’s invention. It remains uncertain whether telecom’s incumbent carriers and equipment makers will continue to dominate or even survive as the information technology industry absorbs voice as a simple application of the Internet.

The roots of the VoIP insurrection trace back to four synchronistic events in 1968. The Federal Communications Commission (FCC) ruled MCI could compete with AT&T using microwave transport on the Chicago to St. Louis route. The same year, the FCC’s Carterfone decision forced AT&T to allow customers to attach non-Western Electric equipment, such as new telephones, and modems, to the telephone network. The Department of Defense’s Advanced Research Project Agency issued a contract to Bolt Beranek and Newman for a precursor to the Internet. And in July 1968, Andrew Grove and Gordon Moore founded Intel. Innovation in the communication sector remained the proprietary right of AT&T for most the 20th century, but events in 1968 breached the barriers that kept the telecom and information technology industries apart. For the first two-thirds of the 20th century, AT&T had manned Berlin Wall separating telecommunications and computing, but eventually, these two enormous technology tracks would be unified.

Two entrepreneurs barely out of their teens, Lior Haramaty and Alon Cohen, founded VocalTec Communications in 1993 based on the promise of packet voice technology they observed as members of the Israel Defense Force. Most military command and control used the highly survivable TCP/IP distributed data networks since the 1980’s. The challenge of transporting voice over the networks arose as an imperative to support certain very sensitive voice commands like “drop the bomb”, but the idea of commercializing packet voice did not occur to anyone until the arrival of Lior and Alon. How could slicing voice into 50 millisecond packets improve the telephone business? The tradition bound telephone industry types or “bellheads” spent their time before 1995 improving the Public Switch Telephone Network (PSTN) not replacing it.

Advances in communication from writing and paper to the printing press, telegraph, and telephone shape human progress. Some might have viewed VoIP as an interesting toy in 1995, but no one presently doubts it will dominate the communication future. The economies of scale associated with growing customer awareness and competition will produce a Moores Law like virtuous cycle of communication innovation.

The coming communication renaissance will make the 20th century telecom industry seem as quaint as the Pony Express by comparison, but what happens to the existing trillion dollar global telecom industry? The near term stakes effect millions of telecom jobs and billions of investment dollars not to mention the fact communication serves as a basic input for the economy much like oil. Can companies like Verizon and Lucent reinvent themselves sufficiently to survive the VoIP insurrection? Keep in mind the telecom incumbents remain gatekeepers for access to the Internet and regulatory uncertainly continues to weigh against investment in VoIP. Verizon presently serves as many telephone lines, collects as much in revenues, and enjoys twice the profits as the pre-breakup AT&T colossus. Let the games begin!

IP on everything
The forces putting the Internet in a position to displace the PSTN have nothing to do with voice as an application. The cycle of innovation that serves as an engine for the information technology industry makes all of the elements needed to enable VoIP faster and cheaper. The forces reducing the price of processing power exist independent of VoIP, but as a computation intensive real-time application VoIP benefits from the persistence of Moore’s Law. The number of people with access to the Internet and, in particular, broadband access continues grow with or without VoIP, but all communication applications benefit from network effects. The Internet backbone latency declines absent specific requirements from VoIP. For example, a connection between New York and Hong Kong that might have suffered 500ms round trip delay in 1996 experiences less than 150ms delay today. The emergence of Wi-Fi and other unlicensed wireless developments associated with IEEE 802 working groups moving VoIP into the mobile realm represent yet another example of this dynamic.

No one originally expected the Internet to make the PSTN obsolete, but they did seek to avoid the vertical integration making the PSTN inflexible. The people shaping the Internet viewed themselves as users not service providers, where as the PSTN existed to serve the interests of the telephone company not users. The PSTN works as one integrated system synchronized everywhere to the same clock. All of elements that make up the Internet work independent of each other. The coordination happens through an exchange of messages. This difference sets the Internet on a pace of innovation that allowed the rudimentary capacity and performance available in 1970 to improve at a rapid rate unleashing the present Darwinian survival of the fittest process. On going performance improvements continue to give data networks the capacity to efficiently address an ever broader range of applications.

The arrival of VoIP in 1995 corresponded with the arrival of a PC (i.e. Intel 486 processor) capable of managing the encode and decode processing in real-time. VocalTec offered packet voice solutions starting in 1994, but the need for special purpose hardware to support real-time two-way communication slowed deployment. Dialup Internet penetration reached nearly 20% in the United States in 1995 and interest in VoIP accelerated in 2003 as broadband penetration rates approached 20%. Interconnection with the PSTN remains the largest cost for VoIP service providers, but the need for PSTN interconnection falls as broadband penetration increases. It only took ten years after the arrival of the commercial Internet for most people to get an email address. The same outcome seems likely with regard to VoIP telephone numbers. In any case, the challenge of producing a low cost VoIP version of traditional telephone call seems largely solved. The most interesting developments remain to come as VoIP enables capabilities that go beyond the plain old telephone call. With the price of the traditional telephone call going to zero, there will be enormous incentives to offer more value.

VoIP turns telecom into a simple extension of consumer electronics business, because Internet applications exist without metering for time and location. Users of VoIP need not worry about the destination or duration of their calls any more than someone sending an email or browsing the web. People do not pay each time they play a CD, and communications seems headed in the same direction. Microsoft X-Box Player already offers VoIP for participants in multi-player games. Metering and billing calls can easily cost more than delivering the service itself, and the flat rate access billing model eliminates the need for solving inter-carrier compensation.

The decoupling that produces rapid improvements in connectivity and processing platforms also facilitates software development. People working on VoIP applications don’t need to change the nature of the Internet with each new application, and everyone with a computer becomes a potential member of the Internet development team. Applications of the Internet from email to the web to instant messaging and VoIP without exception have come from the tinkering of entrepreneurs rather than an industrial research center backed by market research.

The competition between VoIP and the PSTN shapes up much like highways versus railroads. The operator of the PSTN and railroad own their transport network. VoIP companies and car companies do not. Railroads and the PSTN support a single type of usage. Highways and the Internet allow all user types to commingle. The emergence of highways empowered people to control many more aspects of their transportation needs rather than depend on the schedules and railroad routes available. The Internet accomplishes the same thing for communications. Automobiles and highways gave rise to an entirely separate industry and provided the basis for new types of commerce. The Internet offers the same promise, and corporate chieftains with traditional telecom assets find themselves in the same position as the railroad barons when Henry Ford got rolling.

III. Disconnect
Alexander Graham Bell lasted only three years in the telephone business. His financial backers took over American Bell not long after 1876 in a manner and for reasons not unknown to entrepreneurs today. Lawyers and financiers shaped telecommunications far more than engineers for most of the next century as various forms of monopoly dominated. The accomplishments of Bell Laboratories does not change the reality of industry stagnation. No area of technology progressed slower than the telephone business in the 20th century as comparisons with automobiles, aviation, or healthcare reveals. Innovations all revolved around obtaining efficiencies from the perspective of the telephone company. Telephone service changed very little from the perspective of end users.

The collection of data networks that make up the Internet exist entirely separately from the PSTN. The public switched telephone network refers to the network addressed dialing by telephone numbers. Home users may access the Internet by dialing telephone numbers, but none of the Internet hosts with permanent IP addresses depend on telephone numbers for connectivity. The links may share conduits. The equipment may get housed in the same buildings or share maintenance staff, but the PSTN infrastructure remains entirely separate from Internet infrastructure. One could turn off all of the equipment supporting the PSTN without effecting the Internet. The PSTN and Internet come together only through special gateways, because they transport messages by entirely incompatible means.

The amazing durability of plain old telephone service (POTS) does not arise from the fact it represents some kind of wonder technology. POTS persisted for business reasons associated with monopolization of telecom and not technology or sound quality. Humans can perceive sound from 50 Hz to 20,000 Hz. POTS captures sound between 300 and 3300 Hz. The copper analog loop associated with the Bell System in the US and monopolies in other countries add various other impairments to this already limited representation of sound. POTS remained dominant until the arrival of VoIP, because of the high barriers to entry facing insurgents.

The local telephone loop still uses analog technology introduced in the 1940’s. Touch Tone phones arrived in 1963. Engineers at Bell Labs understood growing demand required offering new services, but the nature of the PSTN made it largely impossible. Imagine the challenge of software makers in a computer industry stuck with something like the Intel 286 processor for 40 years. The PSTN as a communication platform does not improve in performance, whereas efforts to create demand for computers gets facilitated by Moores Law. The PSTN supports only three kilohertz sound, so everything including data needs to get converted to voice (e.g. the modem squeal). The complexity of the software controlling the large network based circuit switches utilized by Verizon, AT&T, and others make them unsuitable for implementing new services. Each new service requires re-inventing the telephone network, but the revenue prospects for the new services don’t justify the expense. The PSTN’s resistance to innovation meant demand grew no faster than the larger economy or 3-4% per year during the last 50 years.

One attempt by AT&T to improve voice quality in the early 90’s illustrates the PSTN’s handicap. Marketing studies indicated customers might prefer a low end (i.e. bass) audio boost. Sony implemented this with a “MegaBass” switch on their Walkman product line. AT&T wanted to do the same thing in hopes of competing with MCI and Sprint on voice quality rather than price. The consumer Vice-President at AT&T, Joseph Nacchio, pushed through an $800 million project to get the job done. AT&T could not simply install a switch on telephones analogous to the Sony’s solution. AT&T had to alter the signal processing incorporated in echo cancellers throughout the network. These network wide modifications produced irate customers not more customers. The higher sound levels caused operators distress and amplified existing network quality problems. A long period with lots of effort followed to unwind the TrueVoice implementation. Joseph Nacchio famously left to launch Qwest.

The arrival of VoIP in 1995 generated not a rush to hire engineers but rather a rush to hire lawyers. The Americas Competitive Telecom Association (ACTA) submitted a petition encouraging the FCC to regulate VoIP software makers like telephone companies. The move prompted Jeff Pulver to launch the Voice on the Net (VON) Coalition with Daniel Berninger’s help in 1996. More recently, the FCC issued a notice of proposed rulemaking addressing VoIP in March 2004 with the goal of bringing the Internet and PSTN together under a single rational regulatory regime. No one expects the regulatory issues to get resolved soon as the FCC faces a challenge similar to a railroad regulator addressing the emergence of highways.

The telephone incumbents also control a key input needed by the VoIP insurgents – connectivity. It does not represent an accident that there remains a significant disconnect between the capacity of connectivity within the premise associated with ethernet local area networks and the capacity in the middle of the Internet associated with fiber deployments and improvements in the capacity of fiber. The most capacity and greatest performance improvements exist where there exists the most competition. The fact that local access remains a bottleneck or completely absent follows from the weakness or absence of competition for so called first mile connectivity. The phenomena gives rise to the broadband deployment debate, but the disconnect depends less on technology than the absence of market forces compelling providers to improve performance and reduce their prices. The growing list of access technologies offers some hope (e.g. copper, coax, wireless, fiber, power lines, satellite), but it remains unknown whether there will exist sufficient competition between these media to reduce the access bottleneck.

The Pony Express service failed prematurely in 1861 when railroad interests convinced the United Sates government to withhold a promised contract leading its entrepreneurial backers into bankruptcy. Consensus on the technology merit of VoIP does not assure its future. The telecom incumbents like the prospects for VoIP to make their operations more efficient, but they will fight mightily the potential for VoIP to alter the telecom industry balance of power. Campaigns toward national and local regulatory interventions can certainly alter the pace of change. The losses suffered on investments motivated by the Telecom Act of 1996 leave decision makers in the capital markets twice cautious. The combination of regulatory and technology openings associated with VoIP seem a powerful force for change as in the case of MCI leveraging microwave technology and regulatory openings to set up the break up of AT&T in the 70’s.

59 Responses to “The Voice over IP Insurrection”

  1. I have read the comments about Skype, and I agree that it is a program that works well without any hassle. It’s not going to make a big impact though. It uses a proprietary protocol rather than an open standard like SIP, its routing algorithm used to traverse firewalls and NAT is currently flawed (a call between two European people can get routed through Australia) and it’s stuck to PCs (and Pocket PCs) that have Skype software. BT will hopefully release a full blown SIP solution in the UK in the coming years (so far we can purchase a VoIP service over broadband but only as a 2nd line, due to legal emergency call regulations).

    What I can say about Skype is that it will at least accustom people to the idea of enjoying a conversation with someone in high quality sound, for less. Ideally the telcos will make the switch as transparent as possible for those who still want to be able to pick up a regular handset.

  2. When WiMax becomes available, I wonder if that will change the face of cellular service. It would seem that Pulver’s FWD would be a likely alternative for cellular service since WiFi based calls would be free, other than the cost of the connection, if applicable. PDA’s with facilities for VOIP may easily replace non wifi enabled cellphones as a preferred method of making phone calls.

  3. The author makes some great points, but I can’t believe he didn’t mention Skype!!!

    Mark my words, Skype is THE PSTN killer. If you haven’t used it – try it. You can talk from US to China to Europe and it sounds crystal clear, better than a landline or cell phone!

  4. A few comments… I’m using VOIP through . It works quite well, but delays are considerably more than with PSTN. Of course, cellphone delays are similar. We had similar delays when international PSTN calls were carried by satellite. Will consumers put up with the delays? Per minute pricing on long distance PSTN calls seems to be approaching the access fees for final delivery at the destination of the call.

    I remember seeing an ANALOG packet switching system in the early 1970s. It was called TASSI (I think that’s the spelling) and was used on the undersea cable from San Luis Obispo CA to Hawaii. This cable was a single piece of coax and had vacuum tube amplifiers on the bottom of the ocean. It used frequency division multiplexing with each voice converted to a single sideband RF signal on the coax (remember the “monkey chatter” you used to hear on long distance telephone calls from interference from adjacent carriers?). Anyway, TASSI would fill the dead space on one channel with another voice requiring delivery. Since most voice conversations tend to be half duplex (one person talks, the other listens, then they switch roles), this doubled the capacity of the cable.

    As to the future of telecom companies, it seems there are several components to a telecom company. These are the local infrastructure (local loops), long distance circuits (all digital now), and switches (both local and long distance). The change to VOIP would only involve the switches (change from circuit to packet switching). It seems that telecoms would still play a large role in providing circuits and some switching.

    As to regulation of VOIP, it seems to me that there is indeed some sort of unfair playing field at the moment in that telephone companies have to pay into universal service funds, etc. Even if these costs are listed separately on your bill, that does put them at a cost disadvantage to other communications methods that do not need to charge for this. However, if your local internet connectivity provider were to ALSO pay similar fees, it doesn’t seem that a VOIP provider (which seems to just be a directory service unless they actually drop you onto PSTN) would not have to pay the fee. Communications businesses (those that actually provide the circuits) should have similar taxation (either reducing taxes on PSTN or increasing them on internet connectivity providers. How the FCC can claim that an internet service provider is an “information service” and not a “communications service” is beyond me.

    Finally (will it ever end?), it seems that even though there currently appears to be a bandwidth glut, bandwidth does cost money. It seems that the actual metric should be bits time distance. If you are running an application that requires a lower bandwidth and run it for less time, more bandwidth is available on shared media (everything but the local loop on DSL) for other users. As users demand more bandwidth for more time, additional infrastructure needs to be built, and that costs something. Further, distance becomes a consideration since a shorter distance transaction frees up bandwidth on additional hops that were used on a longer distance transactions that are now not used. It may be at this point that the typical bandwidth demands of an internet user are too cheap to meter. Will that be the case when thousands of California users request different high definition video streams from Europe simultaneously? Additional bandwidth on the infrastructure to support this will cost something, and someone gets to pay for it. Should low bandwidth users pay less?

    FCC Rules online at

  5. Stbalbach

    Bellheads had ATM, packet switching, as the roadmaped future before the Internet was on the radar screen of possibilities. So no, netheads have always been the winners. It was a matter of which net: the ATM network or the Internet. As it turned out, ATM is now used to carry the Internet .. they both won.

  6. Interesting side note there spiritwoman, your link in your signature block (if that’s what you can call it, not sure it’s really a “signature block” per se) is ironically not correct and yet your proposing you are from a company that doesn’t fail in this? Just found it ironic is all, nothing meant by it but a bit of humor. (The .biz is simply a .bi in the link). Figured maybe a fruidian slip of the keypad? At least until I added the z and found the site.

  7. Michael,

    Thanks to your book learnings, and pointing out that IP has had an additional 3 mystery layers onto it by other providers. Your interesting point of view is not without merit however is without basis in technical grounding. QoS is and never was a “layer” within IP nor security. It is an extension of the full TCP/IP stack which last I checked was credited to DoD for it’s creation. A slight misnomer but close enough to be accurate for this arguement.

    You state I should break the PSTN into it’s elements, and then don’t truely do so yourself short of vague and not entirely accurate comments for todays day and age. But I shall use one of your breakdowns and then expand upon such for my example. Yes typically these worlds share common backbone and interest, but first of which is the LAN which is extremely rare for the telco or other provider to “own”, MAN’s in contrast can frequently use the PSTN as a transport. WAN’s typically use a long haul provider for it’s access, of which last I checked are typically not what I would consider PSTN as they are “privately” owned (i.e. by the company providing) and while still governed by the PUC is quickly eroding into a competetive and non related industry.

    An example of this is that depending on which job funtion (yes I hold more than just a position with NSI and also hold a position with a DoD entity) I use several long haul providers. Some of which I use absolutely no “PSTN” access to get to (depending on city and availability). In any case I can define it further if you like but this should be sufficient for the rest of the arguement.

    Now on to the other “additional layers” comment, of which none of these are directly related to your mistatements in that they are merely implimentions of a layer in the TCP/IP stack. None truely unique as the model was defined around that concept and they took hold due to popularity (not because they were a new concept to start with). Not withstanding as well is your comment of “But in the core public internet WAN, development stopped there, circa 1996.” which is blatantly not true. VoIP has its birth in that time frame and was not a flop due to non research. It has failed for the same reason it continues to fail in todays businesses and homes. A failure to properly engineer the contidions it is resident in. A 56k Frame Relay circuit is in no way prepared to handle VoIP traffic for a small branch office with 2-3 calls on at any given time. Let alone the data they are using. But because it used the same infrastructure it “must be able to deal with that.” This is such a common mistake in todays businesses that it has stimatized VoIP to it’s slow growth.

    You also state, “The reality of Internet 1 was that it was an insecure, asynchronous, glorified, database lookup.” Since when did this change as well? It still IS exactly as you describe (unless you count AOL’s wonderful “firewall” feature or something silly like that.

    And lets not forget you state, “begin disintermediating the vertically integrated PSTN service provider model” in which I also ask has it only begun? If this has begun then wouldn’t it stand to reason that the status quo is not the same as it was in the 90’s when you say everything stopped? Or when you also state that it’s disintermediating the PSTN do you simply mean that you feel it’s given them a different way to carry voice instead of TDM?

    Morover you claim my assertions on broadband to be false but don’t in the slightest actually refute it. How are they false? (Another case in point along these lines is your statement that this all somehow had to do with ROI, etc. and how I somehow rejected 100 years of history)

    If you want this to be an intellectual conversation and debate I am more than happy to oblige but at least back up what you say with some data, and try to keep it to something you can back up rather than turn it into an acronym spewing session in which you and I may or may not agree on the definitions of such or other area’s. It’s usually good practice to level the field with definitions of your stance.

  8. I really enjoy the PHD’s of VoIP and see the ego’s of all that research that is done. Champion communications is all about success and accomplishing it through Network marketing. The company that HAS VoIP and has WHOLESALED it to the HUGE companies. You guys go for it.. Remember me….. when you have exhausted all the rest…… when you go through the rest of the wannabees. God Bless you all.

  9. Todd,

    Thanks for underscoring the fact that IP was (and is) a 4 layer protocol stack. That’s why you had the boom and bust of the asynchronous internet. The additional 3 layers have been built by competitive market players, making IP (or whatever you want to call it) ready for primetime.

    Before you critique others, I think you need to break the PSTN down into piece parts and understand what is and isn’t competitive, and what is and isn’t tied to your “separate internet”.

    Last time I checked they ran on a lot of common elements. Ask your CIO where NSI gets it’s access to the internet. What facility do you use at home to obtain access? How were those facilities built and paid for?

    There are 7 layers to a service provisioning model (and within those 7 layers, each has another 7 layers, and another and another; you as a systems engineer should appreciate this).

    As well, there are geographic boundaries (LAN, MAN, WAN) that are context (application and market) and regulatory dependent.

    Lastly, there are essentially 4 different network topologies: telco, data, broadcast, mobility; each with its own special mix of elements.

    For you to reject 100 years of history doesn’t jive with market reality (revenues, profit margins and ROI).

    I guess my favorite example is the development of the internet itself. In addition to all the research and DOD precedents in the 60s and 70s, it was the competitive WAN/IXC forces in the 80s that led to the ILEC introduction of flat rate local access. This in turn let guys like Steve Case distribute low cost data routers and access points to provide the transport and access underpinnings of your internet. Then the addressing layers (www) were added. Then your translation layers (html) were added. But in the core public internet WAN, development stopped there, circa 1996. Individual applications and private (subscription) service providers added additional layers. The bust made made it painfully clear that these standalone apps and solutions didn’t scale without the core WAN being upgraded.

    Over the past 5 years, these core layers have begun to be added for data and voice: LCR, CDR, QoS, security, etc…

    The reality of Internet 1 was that it was an insecure, asynchronous, glorified, database lookup. Don’t get me wrong, I think it was and is great. From it, all the “IP-related” components and platforms have scaled to a point where they can begin disintermediating the vertically integrated PSTN service provider model.

    Lastly, a lot of your broadband access presumptions are flat out wrong. They contributed to the hype and the wasted $250 billion of investment. That said, there are topologies and strategies that can drop current access pricing by 95+% within 2-3 years.


  10. Looks like we got back on topic a bit but lets also not forget a few things here in that what this is all about isn’t where we were as much as where we are going.

    – Dan asks the crucial trillion dollar question — “What happens to the existing global telecom industry?” then doesn’t answer it. “It goes to zero” isn’t helpful, unless one is advocating the same solution for incumbent telcos as for the highways Dan uses as his analogy: nationalization. (I don’t think that’s Dan’s argument.)

    This isn’t a crucial question in the slightest. The analogies were an example of lesson’s learned not of repeats of mistakes. The reality is that the PSTN as it is now will no longer be. It will vanish into the past to collect dust.

    – “The forces putting the Internet in a position to displace the PSTN have nothing to do with voice as an application.” Huh? This directly contradicts Dan’s own argument. I used to pay $19.95/month for dial-up Internet access, and now I pay $45/month for my cable modem. Where is the great price reduction driven by Moore’s Law? Answer: It shows up in the fact that I pay little or nothing for the voice *application* on top of that network, and there is room for innovation because the voice offering no longer defines the network.

    No this is the most common misconception of high speed internet I know of. The cost savings you seek lie in 2 aspects. Firstly the time saved in getting those files and websites in 1/10th the time it would have taken on dial up for only double the cost. Secondly you are getting 10x the speed for double the cost. The numbers for me are obvious and for most others are as well. My time is worth money and I am willing to spend a little to save a lot.

    I also notice a comment on IP becoming seven layer, when the hell did IP grow? It has always been 4 layers, TCP/IP has been 7 layers but then my next question is if that is what was meant then when the hell was it not 7 layers? The only thing that grew was the devices carrying it to be able to actually understand and USE all 7 layers and even that is a stretch. It’s been ready for prime time for a long time, it was the networks that carried it that haven’t.

    (1) highly variable sound quality based on bandwidth limitations and/or network traffic issues, (2) no guaranteed Quality of Service (QoS) based on underlying limitations of TCP/IP technology. For all its failings, these are not issues that POTS has at all, particularly in the local loop

    My comment above applies but so does one other issue, you were using a provider that could not give you any QoS and as such you were a victim of your provider not your VoIP company. Don’t judge the tech based on your experience with a problem. Nobody blames the tree they ran into with their car. It’s not about the service it’s about the carrier you had to get you to the service. So the PSTN doesn’t have these same problems? What is an “All circuits are busy message” then? And these are limitations of TCP/IP? When did the technology become the problem? We can all say it’s a problem with the technology but we must also take some blame on ourselves or other aspects. If a tire blows out on our car, was it the tire manufaturer’s fault for not making a tire correctly, or was it my fault for trying to go 145 MPH for 3 hours straight?

    Last but not least:
    Many a VoIP provider requires a subscriber to access the service through an ATA. This limits many of the features that are possible because of the rich signaling protocol.

    How do you figure this? All and ATA is amounts to exactly what the acronym stands for, an Analog Telephone Adapter. This has nothing to do with limiting VoIP, it helps it. How many people have Cat 5 Cables in their house strung to everywere they have a phone? How many people want to pay to have it done? How many have the switches to terminate all that? It’s a process and the ATA’s fit the niche nicely by providing a pathway to adoption. Hore over you say that the ATA’s “rich signaling protocol” is a problem? then change it. The ATA is simply the hardware, the software loaded on it determines the protocol and feature set.

    Simply put folks I see that almost everyone here has given some GREAT history lesson’s and everyone has had some great comments and insight. I don’t mean for any of my posts to be taken as negative, and I do appologize for anyone that does, but I do get frustrated in the stigma that propagates about all of this unwarranted because of ignorance and or propaganda that is due to tunnel visionaries that see the light but only from one side of the picture.

    Lets put it in perspective and think of it as a process. In order for us to get to point C we must first travel through points A and B. The PSTN must go, that can only happen through time as adoption spreads, then and only then can it move to the top where the predominance of calls are VoIP and not traverse the “PSTN” but rather the internet in whatever form it takes.

  11. I think somehow this got off on a tangent. Packetized voice and VoIP are two different things from the perspective of the article. The PSTN is predominately packetized (we don’t think TDM is real time when it’s got time division right in the name do we?) but it’s the difference of a largely open and standards based medium carrying it versus a proprietary and tightly controlled medium. The P in PSTN should have stood for privately since the only one’s with control even today is the ILEC. The internet gave way to a medium that while predominately contingent on the PSTN still is quickly eroding that stronghold due to the technologies emerging. Cable internet for one, Powerline, wireless and others are overpowering the ILEC’s hold on your telecommunications. Make no mistake VoIP is here to stay and should laregly stay unregulated. What is constantly forgotten is that the FCC is behind denying the wishes of the telco’s to have so called “fair regulation”, but moreover is that other large companies have a vested interest in it too. You don’t really think that cable companies that are playing in that field really want to lose their advantage over the telco’s by backing legislation to charge themselves? Are we really that gullible to believe that the likes of Cisco (now vested with Vonage to provide these types of servicecs) wants to do the same and hurt the emergence of their products? So often we see one side of the story and grab hold fearing draconian measures are taking place to thwart the little guy when in reality things are quite the opposite and indeed capitalism is still in effect. Everyone needs to take a step back and think this through again. Yes the telco’s are threatened, yes they are upset, yes they have a bankroll, but so do others and last I checked capitalism thrived on the people looking after their own pocketbooks and not everyone elses. I also find it funny to see how many people are so fooled by the idea that the the PSTN and VoIP are so coupled together. Or that similar technologies are so easily implimented. The are indeed and will stay two totally different mediums until the ultimate demise of the PSTN as we know it. Does anyone really think that because their modem uses the PSTN that an IP packet inherently can traverse it as well, or that because the PSTN is packetized as is and IP packet that the phone on the other end let alone the switch carrying the call knows what an IP packet is? Wake up and smell the solder folks, the one issue that remains on why VoIP has to grow still is that at some point on the network there has to be, without question, an interface to get onto the PSTN and thus a cost incurred. But what will happen to the phone numbers we all know and love? Anyone hear of IPv6? Yes that will be your phone number (and I would predict that will happen in under 10 years) but that will suck…so what will you do? You’ll use DNS…easy answers, but not so easy to do on the spot. Moreover why would anyone want a “standard voice call” when video will be the predominate market growth area around this time frame. We are so hung up on “can you hear me now” that we forget to look to the future of “why mr Jones why on earth are you still in your bath robe during your business teleconference” that is so quickly arriving. Voice is a disposable app that shall quickly get displaced, get on the bus or get run over by it because this revolution hasn’t even begun to happen.

    Todd Keller
    Senior Engineer
    NSI Communications
    VoIP/Business Convergence Provider

  12. Daniel:

    Thanks for responding to my questions. As a followup:

    It is a bit of rash conclusion to characterize the packet voice work at BL to be Packet over Frame Relay and dismiss it. The packetization technology could have been used on any packet network. Yes, the initial application was for trunk efficiency; still it benefited the consumers. Personally speaking, that technology increased the call completion rate to India, which in turn reduced the price considerably. From a consumer’Äôs point of view this is useful rather than a philosophical position of obsolence of PSTN.

    Many a VoIP provider requires a subscriber to access the service through an ATA. This limits many of the features that are possible because of the rich signaling protocol. For example call waiting is limited. Here is an example of some realizations of VoIP has artificially limited the features available to the users. Very few proponents of VoIP, especially those motivated by benefits to the users, have used their bully pulpit to change this situation. I would like to know your thoughts on this matter.

    I am not sure I fully understand the issues related to customer premise regarding TrueVoice. Is it that the house wiring and the local loop were mostly analog and so can carry only 3 kHZ voice? In any event, this should not have been total surprise to the project. At that time AT&T prided itself in the extensive testing that was conducted. (; Still AT&T couldn’Äôt verify their assumptions regarding the performance of CPE? Just for this, PSTN must be condemned? Also we have to keep in mind that once the access loop was digitized with ISDN, a new capability called 7 kHz voice was introduced (some 5 years earlier).

    More importantly, considering the present time, how would ATA based service provider fare? I posit that the situation is no different. Upgrading the ATA is not possible because the service provider may have access to only one side and not the other and most certainly the telephone and the wiring connected to the ATA can not be controlled. Do we have to look for an alternate transport network?


  13. VoIP can’t be stopped if ppl want to use it. Example Skype, most of my friends use it, and since I dont use the cell phone or own a fixed line there is not much use for a cell. But this is a situation at college, so once work starts things change. The point is that there are about 10 of us who use Skype and this would be the bulk of the phone bill if I was to use the phone. So how much are the telcos losing now?

    P.S. nice work on the paper :)

  14. VoIP can’t be stopped if ppl want to use it. Example Skype, most of my friends use it, and since I dont use the cell phone or own a fixed line there is not much use for a cell. But this is a situation at college, so once work starts things change. The point is that there are about 10 of us who use Skype and this would be the bulk of the phone bill if I was to use the phone. So how much are the telcos losing now?

    P.S. nice work on the paper :)

  15. Bill Blass

    Interesting article, but a little too over-hyped about VoIP. Have you ever actually used the technology for any period of time? I have, and have concluded that it is *still* not ready for prime time more than nine years after introduction.

    Here are the two biggest problems: (1) highly variable sound quality based on bandwidth limitations and/or network traffic issues, and (2) no guaranteed Quality of Service (QoS) based on underlying limitations of TCP/IP technology. For all its failings, these are not issues that POTS has at all, particularly in the local loop.

    I’ve just switched back to POTS, because VoIP on my 1 megabit uplink is far too variable to be usable. It sounds fine for a while, and then my voice turns into a robot and is unusable. Maybe in another nine or ten years [sigh] it will be useable on the Internet.

  16. Paul Zawada

    This is a very interesting article but I think it misses one point in looking back at pre-divestiture AT&T. Other folks pointed out that Bell Labs had the smarts to figure out packetized voice and deploy it if called upon to. The problem was that much of AT&T was a product of the heavy regulation it lived under. AT&T and the independents essentially had a pact with the government. Their job was to provide highly reliable voice service to as much as the population as possible at a reasonable cost; if they did so (and nothing else), the government would allow these companies to earn a modest yet stable return on investement. When the events of the late 60s and 70s threatened the return on Ma Bell’s investment (the federal gov’t, for better or worse, was reneging on their end of the deal) AT&T reacted in ways that should have been completely predictable.

    So let’s get something straight… Bell Labs and the Bell System did an excellent job at what they were tasked to do. I think the author of the original article completely discounts/underestimates how large of a task AT&T had. Wiring up an entire nation the size of the US for telephone service at a reasonable cost was an incredible feat. (Can you imagine receiving an “all circuits busy” message on a regular basis or having to send a telegram to someone because they couldn’t afford telephone service?) The problem was that once AT&T’s mission of Universal Service was largely accomplished, society as whole changed their view of what the role of regulation should play in “naturally monopolistic” indutries and the company was not equipped to adapt. One could even infer that AT&T’s makeup virtually guaranteed their downfall once Universal Service was achieved. By being kept out of “data processing,” there was no where for AT&T to go. (And the future of voice was data processing!)

    VoIP insurrection is not about taking action against monopolistic communication providers; it’s taking action against highly regulated business environments. Compaines like {pick your favorite ILEC} push hard for regulation because that’s their heritage and what they know; regulation is what fosters their existence. However, the regulated model came about from the ugly competition in the early part of the 20th century. Forcing the regulatory pendulum to swing all the way back to where it was during that period of history could very well introduce some of the same problems. (Lack of universal interconnectivity, dominant players in the industry strongarming the little guys, etc.) The trick will be to use VoIP technology and leverage its benefits without setting us back in the area of basic voice communication between any two points at reasonable cost. If VoIP is to replace the PSTN, it must do so by continuing the mandates of the PSTN, most notably universal service and E911. It needs to do so in a way that keeps us out of heavily regulated environments (that do not encourage innovation) yet insure everyone’s basic communication needs are met. I think it can be done, but there are devils in the details that need to worked out.

  17. First was the Computer Inquiry II of 1982, which took effect 1/1/83. We lived in limbo for a year at AT&T as our 3B20 Simplex (a knockoff of the 3B20 Duplex that ran 5ESS) could not be sold with an FPU, since that would be a computer. We sold a few, though I don’t recall whether they ever had an FPU. Ours didn’t, though it was only a pre-production unit.

    DEC thrived partly because its minis ran our earlier ESS systems, and of course, Unix was born there.

    (Seeing Phil Karn’s name again reminds me of the Usenet controversy over compact disc technology. A number of the Labs people had a high-spirited discussion of whether digital music was worth the trouble. What do you think now, Phil?)

    In truth, telephone switching and computers have been building on each other for their entire history. but as Glenn pointed out, the Telecomms Act of 1934 put up a wall there.

  18. Kevin Werbach

    I’m not sure what Dan’s saying here other than emphasizing what many of us have noted for several years — VOIP is the future of telecom. It’s great to feel good about VOIP, but that’s not enough. A couple examples:

    – Dan asks the crucial trillion dollar question — “What happens to the existing global telecom industry?” then doesn’t answer it. “It goes to zero” isn’t helpful, unless one is advocating the same solution for incumbent telcos as for the highways Dan uses as his analogy: nationalization. (I don’t think that’s Dan’s argument.)

    – “The forces putting the Internet in a position to displace the PSTN have nothing to do with voice as an application.” Huh? This directly contradicts Dan’s own argument. I used to pay $19.95/month for dial-up Internet access, and now I pay $45/month for my cable modem. Where is the great price reduction driven by Moore’s Law? Answer: It shows up in the fact that I pay little or nothing for the voice *application* on top of that network, and there is room for innovation because the voice offering no longer defines the network.

    Look, I’m as big an advocate of VOIP as anyone, and a fan of Dan’s work. But if we want this insurrection to succeed, now is the time to face the hard questions.


  19. Three digital waves generated by long-distance, datacomm and mobility competition crashed over the info-media landscape in the past 20 years. Digitization was a mechanism by which service providers balanced cost, coverage, capacity and clarity with ubiquity, usability and unit cost. Competitive service providers and the capital markets learned that getting, keeping and stimulating demand on rapidly obsoleting capital bases drives income statements and balance sheets.

    In the aftermath of these market-driven waves, the 100-year-old, heavily regulated PSTN and the 80-year-old, moderately regulated media segments were joined by the entirely new, relatively unregulated, two-way datacomm and wireless segments. The result was growing chaos, which the Telecom and Cable Acts were meant ’Äî but failed ’Äî to resolve.

    To rationalize and counter this chaos, the markets embraced the concept of convergence, epitomizing the ’Äúall in one’Äù approach in vertically integrated CLECs (PSTN) and horizontally oriented application service providers (Internet). Unfortunately, vertically integrated service providers could not scale all layers of their operations and investment effectively across a demand environment where everyone and every organization wanted its converged bundle put together differently. And while ASPs appeared to scale more effectively along those lines, IP, as a four-layer protocol, was prone to poor QOS and costs that actually snowballed in a world of distributed processing. In the end, more than $250 billion of promise and hype got washed out to sea.

    Since then, all four segments have followed the immutable laws of Moore and Metcalf in their supply evolution, while demand has continued to evolve at a rapid and varied pace. At the same time, IP has grown up as a ready-for-prime-time, scalable seven-layer protocol stack and represents the foundation for the fourth and final digital wave.

    We see that wave developing rapidly, but believe it will come from a different direction. Most expect it to start in the migration from TDM to IP at the customer premises and in the Class 5-to-softswitch conversion process. In reality, IP and VoIP have scaled in the WAN and across horizontal layers of the stack over the last five years.

    The growing wave will crash against vertically integrated service providers and undo monopoly bandwidth bottlenecks. Today, one megabyte of synchronous, high-QOS MAN bandwidth for commercial applications costs about $200 a month in developed countries and $500 a month in less competitive markets. When contrasted with actual hardware, software and operating costs in the LAN and WAN today, those numbers should be closer to $10 and $20, respectively. Furthermore, the bulk of the monopoly cash flow actually derives from the terminating, not originating, side of calls or sessions.

    There is a need to break down the likely developments that will lead to a final and precipitous collapse in access pricing and develop the revenue and demand upside. Our crystal ball says it’s a pretty good outcome, notwithstanding a lot of wrenching change. Didn’t that happen three times previously?

  20. That’s the most insightful perspective on the bigger picture that I’Äôve ever read, fantastic stuff. At Loose Connection in Brighton, we’ve seen a huge increase in usage of VoIP-esque apps (such as Skype) in our free hotspots. We’re also leading some of the first WiMAX trials in the UK and I absolutely believe in the disruptive nature that these technologies will have against the incumbent providers; VoIP is a key benefit especially with distributed businesses such as corporates or local government. When each base station can pump 16gbps wirelessly over 30miles who needs PSTN?

  21. I wonder how much you CAN regulate VoIP? I suppose there is some network latency, but if I don’t like the rules in North America, I could get a VoIP provider in Vanatu or Tonga (or wherever)? This would be my voice equivalent of an Email server/address/phone no.?

    As long as that guy can connect to PSTN locals around me, I have local service. As long as local VoIP providers will interconnect with me, I can talk to my friends for free. What are they going to do, block all IP from Vanatu?* Then the Russians will forward VoIP to Vanatu… It just gets sillier.

    The advantage of over-regulated government-monopoly PSTN is that it was pretty much a standard interconnect. Will we be reduced to a mish-mash of n-factorial bilateral interconnect agreements – “I’ll forward my VoIP to your people if, vice versa”?

    If we opt for the anarchy model – everyone their own VoIP supplier – then the local PSTN connect, and the suppliers of DNS equivalent “phone numbers”, become the bottleneck, the gonad-grip control points. Otherwise, how does grandma with her black dial phone call you?

    (If she has to dial Vanatu at Bell rates, she’ll dial and hang up; then I’ll look at the caller-id and call her back for free.)

    Of course, with a critical mass that becomes unimportant. Most of the people you want to talk to will also be on VoIP.

    (* IIRC, at one point, the NorthWest Territories phone company – Bell Canada? – tried blocking cheap long distance companies’ incoming calls. Rates were asymetrically high locally. This was a service where you call an 800 number, enter the called number, and wait for their computer to call you and the other party. First they blocked that number. Then they had to block any of several hundred numbers, and eventually whole area codes. Then they finally admitted it was impossible for PSTN to beat a computer; they had to adjust their usurious rates accordingly.)

    and how do I get paragraph breaks here?

  22. AT&T was not an ‘East Germany’ guarding the Berlin Wall between computers and Telecom. They felt they were bound by a 1933 Concent Decree to stay out of the computer industry. That is one reason they accepted the 1984 breakup, they realized that the automated switches they depended on were now general purpose computers with specialized software and possibly in violation of the 1933 Decree.

  23. To answer JD above, the one who will collect the “lion’s share”
    will most definitely be Microsoft (go read about TCPA, for instance). Our only hope that this doesn’t happen lies in Free Software.