The FCC posed a very interesting question to the industry. They asked if VoIP should become the only way of delivering voice service. This infers having an all-IP network that is extended out to every customer. The FCC asked this question as part of the IP trials that a few telcos are currently undertaking to see what an IP world looks like. While IP undoubtedly makes for the most efficient telco network, I think there many practical reasons why this can’t be implemented everywhere.

We can start with the FCC’s own estimate that there are something like 14 million rural homes without a broadband alternative. These are people who live in rural areas and are almost universally on old and sometimes very poor copper. These are people who can’t get DSL or cable modem and for whom VoIP would not work. There are also a ton of people in the country who are on marginal DSL service who also will have a hard time getting working VoIP. Most such people are also rural, but there are older urban networks with bad copper that also suffer from problems related to the condition of the copper.

But aside from the rural issue, it’s an interesting question. Cable companies already all use VoIP for voice, as do fiber overbuilds. Urban telcos could also give everybody VoIP, but it would mean providing a DSL connection to everybody on copper. This would cause all sorts of network problems. We found out years ago that you can’t put too many DSL lines into the same large copper sheathe or you create interference problems, and universal VoIP would put DSL on just about every copper pair. I also can’t think of any financial benefit to the telco for spending the money to put voice on DSL if that is all a line is going to be used for.

All of these issues make it hard to imagine mandatory VoIP at the customer end of the network. I’ve always envisioned that the IP transition would mean an all-IP network between carriers, which would create the most efficient network. But forcing VoIP where it won’t work right sounds both expensive and impractical. And it would likely boot millions from access to the voice network.

But there are other parts of an all-IP network that could be interesting. For instance, if the whole network was IP from end-to-end you could do away with telephone numbers. In an all-IP network each customer would be associated with an IP address, and so keeping telephone numbers would be forcing a historical structure onto an all-IP network. While we all would probably still have phone numbers, it would be just as easy to just pick somebody out of a computer menu by name and connect with them without going through the fiction that a number is required.

There are a few situations where going all-digital is a bit of a concern. Take 911. The current 911 network is comprised of a redundant pair of special access circuits between each carrier and each local 911 center. This network layout was created to greatly increase the likelihood that a 911 call can be completed. But in an all-IP world 911 traffic would probably be routed with everything else, which is not an issue of itself. But we know that Internet pipes go down all of the time. So anytime there was an Internet outage in a town or a region, 911 would go down with the Internet.

Of course, the FCC didn’t suggest this in a vacuum. They are being prodded in the whole IP-transition by both AT&T and Verizon who would like to get out of maintaining rural copper lines. AT&T has said many times that they want to cut down millions of rural lines and convert them to cellular. And so any pretense that the carriers are interested in creating a rural all-IP network is a fiction, because these large carriers don’t want to own or operate a rural landline connection of any kind. As we recently saw with a large sale of Verizon FiOS lines to Frontier I’m not sure that the two big telcos want to maintain any landline connections at all. These telcos are now mostly cellular companies who are finding landlines to be a nuisance.