As the FCC ponders Intercarrier Compensation yet another time, and incumbent carriers engage in endless disputes with competitors over how much to pay each other for what, a key question remains unanswered. Formally raised about 15 years ago (the FCC does not like to make rash judgments), there is still no clarity about how should Voice over Internet Protocol calls be treated, for regulatory purposes? New rules that attempt to answer this with special treatment for “VoIP” are doomed to fail.
The reason that this question is so hard to answer is because it’s not a valid question at all. While I accept that there are generally no stupid questions, just stupid answers, there are nonetheless questions that by their very nature lead to less, rather than more, insight. These are questions whose very framing requires a certain type of response which may well be incorrect. We all know the old joke, “When did you stop beating your wife?” The question itself carries strong implications. How to regulate VoIP is indeed the same type of question, because it, too, carries a false implication, that VoIP per se is something distinct.
VoIP is like magic. It is a neat trick, and it requires a bit of skill to pull of correctly. But more importantly, magic is based on distraction. A magician distracts the audience from what he’s really doing by diverting its attention, creating, in effect, a blind spot in which the trick’s magic can be worked. When we frame the discussion about VoIP as if it were a real thing, we’ve already looked the wrong way. We’re fooled.
Thus, the only way to understand how to deal with VoIP is to change the underlying assumption. Let’s begin by saying that there is no such thing as VoIP. Of course this is not literally true from an engineering perspective. There are in fact many kinds of VoIP. And that alone illustrates why VoIP is not a single thing, and thus cannot be properly regulated as if it were one. Rather, being “VoIP’ is a secondary characteristic of many different things. The magician has a lot of tricks in that hat.
The many flavors of VoIP
As a generic type of technology, VoIP is widely used. But from a policy perspective, the different types of VoIP are not similar at all. Some VoIP-based products and services are just like their non-voice analogs, differing only in the use of the IP header for multiplexing. This is entirely internal to the network and thus should have no regulatory impact. Others are novel applications that were enabled by IP technology and thus don’t fit into existing regulatory baskets. And sometimes the former are confused with the latter, which is when things really get out of control.
Nomadic VoIP is different from POTS
Perhaps the best-known application of VoIP technology is in the provision of what are sometimes called nomadic, over-the-top services. This category was pioneered by Vonage, who won an important FCC ruling against the Minnesota Public Utilities Commission several years ago. The FCC held that Vonage is jurisdictionally interstate and not subject to state regulation. The key word is nomadic: Vonage does not provide the transmission path to its subscribers (hence "over the top”) and thus does not really know where its subscribers are. It depends on its subscriber’s existing Internet connection instead. Traditional POTS (plain old telephone service) bundles a transmission path with a switched network access port, but these are two different things. Vonage sells the network access, not the path. (Of course many Vonage competitors now do the same thing.)
Since the Internet masks geography, Vonage, and thus its underlying telephone carriers (since Vonage is not a carrier per se, but gets its calls through CLECs, especially Level 3), does not know where its customers are. A subscriber can take a Vonage adapter with him on a trip and attach to the Internet anywhere and still receive calls to the same number. Yes, they are now asked to register their location so that they can dial 911 correctly, but Vonage-assigned telephone numbers are not necessarily in the geographic rate center that the subscriber is in, so the usual phone number-based billing mechanisms don’t work. Tying E911 to billing data would still be a hell of a chore, especially for the minority of users who use it in the nomadic manner. Bear in mind that it’s the other carriers who would care if the call is local, intrastate, or interstate; they can’t read Vonage’s confidential E911 registration data.
Thus nomadic VoIP services are not simply POTS-type services that use IP headers. They are materially different in separating the transmission from voice-switching functions. Some other over-the-top (or “parasitic”) services, like Skype and VoxOx, don’t even need special hardware; they use the PC’s audio capabilities to make phone calls. What could be more nomadic than a laptop with a telephone number? The Minnesota ruling correctly recognizes this.
This type of VoIP doesn’t fit the existing rules very well; it is a real mix of Internet and telephone, and trying to pin state jurisdiction onto it would be rather difficult. Actually it could be easy, but only if the FCC abandoned one of its sacred, if foolish, principles. Nomadic VoIP rides over the Internet and meets the PSTN somewhere, and a telephone number is assigned to somewhere, too. If the FCC allowed either of those locations – the physical demarcation, or the telephone number’s rate center – to be treated as definitive, then the Internet side of the call wouldn’t matter. If you had a Minneapolis phone number and took your VoIP service to Brazil, it could still be billed as a Minneapolis call.
But that would violate the FCC’s “end to end” principle, wherein the actual location of the end users matters more than the handoff. There are exceptions to this, but it was the basis of the FCC’s ruling some years ago on ISP-bound calls. The FCC claimed jurisdiction over them too, lowering the reciprocal compensation paid to CLECs below what they’d have received for state-jurisdiction “local” calls. That ruling pleased the ILECs who asked for it, but it has come back to haunt them when applied to VoIP, lowering payments for non-local calls. The same fatuous argument cuts both ways. The FCC’s newest compromise treats non-local VoIP calls as interstate toll calls, removing the Vonage exemption from paying any terminating access charges, but exempts them from usually-higher intrastate rates.
Not all VoIP is nomadic or uses the Internet
IP is essentially little more than a multiplexing header. It carries no real semantics. So it tells you nothing about the call. This is one reason why nomadic VoIP can’t be billed based on IP addresses either. But then not all VoIP is nomadic. Take, for instance, PacketCable, the standard for delivering telephone service across cable TV networks. This is probably the dominant use of VoIP technology today, but it’s nothing at all like nomadic over-the-top VoIP. It’s a fixed service, wherein the cable company provides both the transmission path and the switching function. It does include a VoIP header in its protocol stack, but it never touches the public Internet. Since the IP header is thus literally meaningless (semantic-free), its presence in PacketCable is also meaningless.
So there is no real difference between a PacketCable "VoIP" line and a POTS line. Its location is fixed, and known to the provider. It is essentially POTS, and there's no valid reason to treat it otherwise for regulatory or billing purposes. This agrees with an FCC ruling some years ago on “VoIP in the middle,” where they held that AT&T’s use of VoIP in a call between two POTS lines was irrelevant. They were the long distance carrier and owed access charges regardless of the technology within their own network.
But Verizon disagrees when it is to their convenience. Thus, they have been refusing to pay intrastate access charges to some cable companies who use PacketCable, on grounds that the IP header somewhere on the cable is a kind of magic pixie dust that changes everything. They’re playing on the fallacy that there is such a thing as VoIP, and that all VoIP is the same. Vonage takes advantage of ambiguity in the rules to not pay access charges on the toll calls it delivers to competing carriers.
Another widely-used type of VoIP is the VoIP PBX, a market in which Cisco Systems has had considerable success competing with traditional telephone equipment suppliers. Sometimes these are connected to multiple sites within the same company’s network. These systems often make use of MPLS to sort out the VoIP from other IP traffic, providing the Quality of Service needed for high quality voice transmission. Again it is generally not sent over the public Internet, and since the PBX is simply buying service from carriers, its use of VoIP vs. older technologies probably shouldn’t matter. But if the PBX is connected to a CLEC using VoIP trunks, rather than TDM, then it too might be considered “customer premise VoIP” and might be eligible to get long distance at a discount. However, VoIP PBXs do not generally connect to the PSTN across the Internet, and while some of their extensions may be nomadic or at remote location, most usually aren’t. So where should they fit?
These aren’t the only potential meanings of “VoIP.” There are VoIP trunks that operate within a carrier’s network (though that’s “VoIP in the middle,” a rare case where the FCC has said that the use of VoIP doesn’t matter) and VoIP (SIP) trunks that interconnect carrier networks. Some CLECs have petitioned the FCC to require ILECs to offer VoIP trunks, rather than TDM, for the exchange of calls between their networks. That may be a perfectly reasonable request, but it leaves open the question of how to charge for them. (It’s now an open issue before the Commission.) And there are VoIP stations-off-premise on PBX systems that are connected to their carrier using TDM trunks, among other combinations of TDM and VoIP technology. In such cases, where is the end point for billing purposes?
Believe it or not, the FCC and carriers are still worried about “toll evasion!” (Mr. Cleaver is going to have a talk with Beaver about this any day now.) If the price of a phone call depends upon whether it is “VoIP” or not, and where its “end points” are, this can get very confusing indeed. A much simpler solution would have been to ignore whether a call is or isn’t “VoIP.” The reason for the exemption is the end-to-end test. Modems used by enhanced service providers (most of whom morphed into ISPs) were seen by end users as having “local” numbers, but in the 1980s, the FCC insisted that they were really interstate (toll) access circuits that were merely “exempt” from being charged as such. The tariff for a local-type circuit used for interstate access is called Feature Group A, and it dates back to the MCI Execunet controversy of the 1970s. If Feature Group A were abolished, then a call would begin and end and its PSTN demarcation, and it would be local or toll based on its phone number. No further exemptions would be needed. Instead we have rules of nearly fractal complexity, and a pending reform that merely adds to that complexity.
The magician has pointed to a shiny bright object called “VoIP,” allowing this distraction. Like all magic, however, it is all part of a larger trick of deception. The American telephone network operates under a web of obsolete rules that were designed in the last century to subsidize local service, back when the plain black telephone, manual or rotary-dial, was the consumer’s sole voice connection to the rest of the world. Arguments about jurisdiction and call classification stem from this scheme that undercharges for some services by overcharging for others. The Internet proves the obsolescence of this business model. Any wholesale intercarrier pricing that depends on invisible factors within the other carrier’s network is doomed to fail. Asking whether or not a call is VoIP is thus a proverbial stupid question, one that can only lead to stupid answers. Leave VoIP to the engineers. Billing and associated rulemaking should be technology-neutral. Thus the correct paradigm is that there is no such thing as VoIP. That is the only way to let VoIP technology evolve correctly.