Voice over IP has been hyped as a breakthrough technology for at least ten years now. According to the hype, VoIP will make the telephone network obsolete, and that soon, every telephone call will be free. People were saying the same things ten years ago. This utopian world where every call is free is an alluring idea, but it will never happen. Here’s why…

I have been working with computer telephony and VoIP technology since the early 1990s. It is interesting stuff, and it enables you to do things that were expensive or impractical with older circuit switched telephone systems. Unfortunately, many people outside of telecom have a fundamental misunderstanding that somehow the Internet and the telephone network are different things, that the Internet is the shiny new toy, and the telephone network is the decrepit old dinosaur.

The reality is that there is a great deal of overlap between the two, especially in the last mile from the Internet backbone to end users. How do you receive your Internet connection? Via a DSL line (connected to your local telco’s central office)? Via a wifi access point (probably connected to a DSL line or ISDN line)? Via a T1 connection? Via a cellular phone that is communicated with a tower that is wired to a terrestrial network? Most types of internet connections touch what we think of as the “telephone network” at some point. If you are running a peer-to-peer wireless LAN with no connectivity to the outside world, you might be operating in a pure IP world, but most real-world users are leasing circuits or wireless access from telephone companies.

Everybody loves to bash telephone companies, but the fact is that it takes a lot of work to build and maintain a telephone network, and to provide the level of reliability people expect. Unlike writing software, building a telecommunications network involves a lot of physical labor and civil engineering. While much of this work was done decades ago (which makes it easy for people to think that one can wish things into existence with a few keystrokes), the telephone network is an essential utility that requires continual maintenence and investment. The guys who repair the wiring when a tree gets blown down in a storm aren’t moonlighting for stock options.

Skype and other VoIP providers are right that metered rate (per minute) billing for telephone service is going out of style (for landlines). However, what they don’t mention is that total spending for telecom services has actually increased in the past decade. Add up your bills for basic telephone, DSL/cable internet, mobile phone and other communication services (mobile email, unified messaging, etc). I’ll bet you’re paying more today for this bundle than you were ten or so years ago, when you had POTS and basic cable and nothing else. These companies have to recoup their investment in new technologies (xDSL for landlines, “3G” for mobile networks). They aren’t going to give services away for free. Instead, they find ways to bundle things so they look cheaper (see above).

The basic problem I have with consumer VoIP services, and the hype that surrounds them, is that they are largely chained to fixed line broadband connections (most wi-fi access points simply front end a DSL line or similar). If both parties are close to an access point, this works fine. The problem is that landline use is flat or decreasing, while mobile use is growing. I rarely use my landline any more, except as a backup when I have poor cellular reception. Many of my friends rely exclusively on cell phones. As much as I’d like to talk to my friend in Germany over the Internet, he’s rarely in front of his computer.

The problem with VoIP in a mobile environment is many-fold. First, there are technical issues which make VoIP over cellular troublesome or flat-out impractical in many situations. Second, bandwidth is finite in mobile networks, and that makes it impossible to give service away for free. Third, mobile communication is about convenience and ease of use. People are willing to pay a premium to have the convenience of a mobile phone, so price per bit is not the only factor. People will pay extra to have access to a network with wide geographical coverage.

The technical issues of doing VoIP well on handheld devices are formidable. First of all, mobile networks are inherently unpredictable, moreso than fixed line connections. This makes all of the usual problems with VoIP (dropped packets, delayed packets or latency, variable delay or jitter, even worse). 3G networks solve many of these issues, but they are not free. If you are using older 2G service such as GPRS, your experience will be similar to VoIP over dial-up, in other words not very good.

Another major issue is that most mobile phones are not designed to run CPU intensive applications such as VoIP software. So even if you had a fast enough mobile connection, getting the VoIP client to run well on a typical mobile phone, a J2ME handset for example, will be a challenge. I’ve been using smartphones for about five years. I have not had many good experiences with software for these devices. To say they are buggy is an understatement. Plus, most consumers prefer cheap handsets that provide basic telephony and text messaging features, and not much else. The market for high-end devices is quite small relative to the mass market (e.g. a few million smart phones compared to two billion total handset sales). Sure some people will figure out how to run a SIP phone on their Treo 650, but I suspect that is an extremely small minority compared to the number of people who will take the latest Nokia El Cheapo handset they get at the local phone store.

However the real roadblock to VoIP in the mobile world is economics. Whether you like them or not, the mobile phone companies provide a service that people want. They’ve invested billions of dollars in infrastructure that does not maintain itself. Providing generally reliable coverage that spans an entire country is not a simple job. They also have a finite amount of bandwidth, and cannot accomodate an infinite number of subscribers. The phone companies want to maximize their return on investment, but they don’t want so many people blabbering away that the network becomes congested. Pricing is a very effective tool for managing demand.

So whether they do this by selling flat-rate service (Metro PCS in San Francisco offers all-you-can-eat voice service for $40 per month) or by selling voice minutes in bulk, or by offering simple metered rate service, they use price to keep usage in line with capacity, and to fund expansion of their facilities. Even when VoIP over 3G becomes practical, the cellular phone companies will make sure that you pay for your calls, one way or another, probably by paying $20-60 per month for the convenience of an untethered high-speed Internet connection.

That isn’t to say that there are not fairly substantial niche opportunities for consumer VoIP services (such as dual mode GSM/WiFi phones that freeload at hotspots, peer to peer service such as Skype). The great thing about telecom is that it is such a gigantic industry that even the niches are big opportunities in their own right. However, building and maintaining a public telephone network that covers a wide geographical area is not easy, and while technology will improve and make this job easier, it will never be free.

VoIP has thrived in markets where monopoly telcos force users to pay several times fair market value for telephone service. In competitive telecom markets where a typical call costs a few cents per minute, or where telcos offer flat rate products, VoIP offers only a modest savings compared to a normal call. This is one reason why VoIP has not been a huge success in the United States. You can get flat rate unlimited calling from SBC for land lines. It’s not as cheap as “pure” VoIP service, but it is also zero hassle, with no special equipment to install, no broadband service required. In the US, VoIP is just another call transport method, and the less than spectacular uptake is large part due to the fact that, for most users, the savings from VoIP are marginal because the US telecom market is already highly competitive. If you spend a lot of time on the phone with colleagues in Estonia, VoIP is a killer app.

In monopoly markets, by contrast, the savings can be quite dramatic. Skype’s biggest success has been in Europe and Asia, two regions with well developed infrastructure and fairly extortionary voice toll charges. VoIP is worth the hassle if it enables you to avoid paying $30-60 per hour to talk to someone. In the long run, as most countries open their telecom markets to real competition, prices will drop and the technology used to complete a call will become unimportant to the user. In this sense, consumer VoIP is an arbitrage technology that is forcing telcos to be honest. That’s what I like most about VoIP. It is forcing telcos to drop their prices to a fair markup above wholesale cost. Once prices drop to a certain level, usually a few cents per minute, VoIP loses its advantage, and customers will use whatever works and seems like a fair price.

This is why I thought the eBay-Skype acquisition was a dumb deal. Skype is really nothing more than ICQ plus voice. It’s a great service, but in terms of utility and convenience, doesn’t remotely compare to cellular telephone networks with their near universal accessibility. I am glad that VoIP is in vogue again. It’s forcing telcos to cut prices to realistic levels.

The problem with the “Skype obsoletes the telcos” hypothesis is that all the telcos need to do to render it irrelevant is to drop the price they charge for calls on popular routes to a few pennies per minute. The telcos can easily do this, and as soon as Skype and company start taking up more than a few percent of overall voice traffic, they’ll cut their prices. Except they won’t be stupid about it. They only need to cut prices to a point where the service is cheap enough. In most markets, that means a few pennies per minute for fixed lines, 10-20 cents per minutes for mobiles. They’ll also start offering flat rate plans, similar to what telcos in the US have been doing for several years now.

Here is a plausible scenario for how things will unfold. Fast forward a couple of years. Skype and Gizmo have grown to account for 20% of voice traffic in Europe, enough to scare the incumbents. One of the more innovative European mobile operators decides to counter this threat by copying what Metro PCS is doing in San Francisco today. They launch an all-you-can-eat pan-European plan that offers unlimited calling to European and US phones for $40 or $50 per month, with decent per minute rates on other routes. Where does Skype figure in this picture? Well, since they don’t own any cellular infrastructure, they don’t.

I’ll wrap this up with an observation. One of the reasons VoIP gets so much attention is due to the “gee whiz, I’m talking to someone in Kreblechistan, for free!” factor. I remember having the same response to email in the early 90s. It is really cool to be able to talk to people all over the world for free. However, the people pushing VoIP repeatedly get blinded by the gee-whiz aspect of the technology and ignore the proven fact that telecom, especially personal telecom, is a fundamentally local business. People buy their phones from a local provider, and mostly talk to people who live near their home city. The percentage of people who spend hours and hours on calls to far flung countries is quite small. Most customers whittle their time away talking to nearby friends about prior or upcoming drinking binges, who’s sleeping with who, who’s going to win the next game, etc. For them, the phone is a way of continuing in-person conversations. In this domain, the mobile phone reigns supreme. Competition among carriers is also local, a battle that is fought city by city.

I suppose this has turned into a bit of a rant, but it gets tiresome listening to people talk about VoIP as if it is the path to utopia when, for example, people are anonymously working on much more difficult technical challenges in the mobile industry. When is the last time you read a breathless article in Wired about the state of the art in cellular phone antenna design? It’s always been this way in telecom though. If you’re working on components or infrastructure, you’re an old world company with an old world valuation (translation: a fair value for the work you’re doing). If you put the word “Internets” into the business, you can be massively overpaid for building a thin layer on top of over a century of other people’s investment and work.

Sour grapes? Definitely, but at least I don’t lose sleep about being unfairly overpaid for the work I do…