Simply Connected!

Policies such as network neutrality and minimum speeds for broadband seek to limit the ability of carriers to favor some applications over others. Well-intended though these initiatives are, they still leave users negotiating for passage while confined to the carriers’ “pipes”. In this scenario, end users remain limited by how the incumbents choose to build their broadband content delivery networks.

Treating the Internet as a utility is not sufficient because, utilities still leave us dependent upon providers whereas, infrastructure is a resource we can use as we choose. What can we do to make these resources available and encourage innovation?

The answer is simple—we can pay for the wires and radios we use for Internet connectivity like we pay for sidewalks and other such infrastructure. We pay for them as a community and are then free to use them. The big difference is that copper, fiber and radios are much less expensive than pavement.

This seems very simplistic, so we have to step back to understand what Internet-native policies are, and how the Internet is different from traditional telecommunications services.

I’ve made some of these points in the past but, what is new in this essay is an emphasis on how the Internet is different from traditional telecommunications because, bursty traffic doesn’t fit into the pipe model. I also point out how the shift from networking as a service to networking as something we do means, we don’t need providers managing the network as a whole. Instead, we can composite local connectivity to create a whole that is greater than the parts, and even better, we’d realize the promise of “one Internet”.

The Internet is neither a thing nor place. It doesn’t carry packets like a railroad carries freight. If anything, it’s more like the way we use paths, streets and sidewalks to get from point A to B. And rather than managing our ability to communicate (speech), we need a new agency that assures the ability of the facilities we use to interconnect. A Department of Connectivity would assist cities and states to manage our new infrastructure.

First Steps

Today’s public policy is centered on a debate about who controls what we do with the broadband pipes. What we need are public policies that enable a world outside the confines of pipes.

Change will likely come from the private sector with owners of apartment houses (condo boards or landlords) buying a single connection to be shared for all users in the building, just as we share a single broadband connection within a home or apartment.

We already see examples such as Castle Village in New York where the tenants pay $10/month as part of their building fees rather than paying for high cost individual connections from Time-Warner Cable.

The FCC can play a role in expanding the policy discussion between separate but neutral pipes to include the option of cities doing the same—sharing connectivity as a basic resource for the city just as they provide roads and other essential facilities.

Around the world, communities such as villages can control get of their local connectivity and pool their resources to connect with the rest of the world. Instead of just having cell phones, they can use the connectivity for all of their devices.

But first we need to understand what makes the Internet so different from traditional telecommunications services.

As an FYI, http://Connectivity.xyz will become a site to discuss how to own our connectivity. It’s currently a place-holder but you can sign-up to get an alert when the site is operational.

Internet-Native Connectivity Policies

Today we need polices that are Internet-native; policies that recognize that the Internet is not a thing but rather a way we can use any available facilities. In the 1970s, common carriage rules played a vital role in requiring that carriers provide a raw packet service. To the FCC’s credit, this occurred despite the carriers’ worries about competition from third parties using the digital technology to create competing services.

Common carriage was never a very good fit with the Internet because it only assures access to telecommunications services. An Internet-native policy would give us the ability to treat the common facilities as infrastructure we can use. I choose the term infrastructure, rather than utility, because the term utility is typically associated with metered services and a utility such as electricity. The Internet uses wires and radios, much like we use paths and roads when we walk and drive without having to pay a provider for carriage.

To understand this better, we can look at real world examples. At home, when you buy a PC, a tablet or any of a myriad of IP-based devices, you don’t need a separate account with a provider for each device—you just connect it to your home network. Contrast this with the world of cellular telephony, where you need a separate account for each device and can only use it in ways approved by your carrier.

The difference between the two, and the reason for it, is that you own the networking within your home.

This wasn’t a result of a shift in public policy, but rather a result of our ability to create our own solutions using the wires and radios within our homes. This wasn’t supposed to happen. In the 1990s, the carriers started to expand on their broadband offerings in the style of cable TV—each time you want to add another TV, you would have to have another set top box with a monthly fee. In much the same way you would pay for each PC and device connected. AT&T also expected to get paid for each purchase made over the Internet.

I’ve been online since the 1960s and in 1995 I wanted to connect multiple PCs to a single dialup connection. I learned how to share that single connection among all my PCs. I was working for Microsoft at the time, so I took what I did and incorporated it into Windows so anyone could buy their own router and do their own home networking. Anyone could’ve done it themselves but by incorporating “just works” connectivity into Windows before broadband connectivity was common, do-it-yourself networking became the norm available to all.

Without such examples for connectivity outside our homes, we tend to view the Internet in terms of more traditional telecommunications service where watching the web seems to be like a television channel and the Internet itself seems to be a utility like electricity generated by a provider. And just as we need more electricity as we get more appliances, we need more (or faster) Internet as we add more devices.

Ideas like network neutrality, the Internet as a utility and high speed broadband are all well-intentioned but are framed in terms of traditional telecommunications rather than recognizing what is different about the Internet. The differences didn’t matter so much, as long as the Internet’s packet traffic could fit within the standard operational and business practices of telecommunication networks, and as long as they didn’t threaten the primary revenue streams. Ironically today, to a large extent, the Internet is a very profitable offering for today’s providers, because traditional telecommunications policies including so-called broadband competition limit the consumers to a choice among providers rather than giving them the ability to reinvent how we communicate.

Today these issues are coming to a head, not so much because of the overt discrimination (such as blocking voice packets), as much as indirect discrimination based on managing the quantity of Internet packets which impact the business arrangements of the telecommunications industry. We can no longer ignore the mismatch between the Internet and a telecommunication industry that manages well-defined “pipes” and makes money creating value-added services using those “pipes”.

Common Concerns

High Speed Internet?

It seems obvious that if we have more devices, we need more capacity to handle each device. This would make sense if the Internet were like electricity, and if you turn on two lights you’d need twice as much current.

The Internet’s traffic is bursty. A high speed connection would allow you to see a large web page in a fraction of a second, and then you could spend a minute reading that page. You could share your connection with 100 other people and it would seem just as fast for each of you. To put it another way, if you have a 25Mbps connection (as proposed), then you don’t need 100 times that (2.5Gpbs—gigabits per second) but instead that one connection will serve 100 users or 100 devices.

Of course the carriers understand this and will share the capacity among many, perhaps 50, customers claiming each one has up to the promised capacity. This is why the business is so profitable for the carriers. They pay once for the wires and can bill people every month long after they’ve paid for the connection. They can then upsell the customer by implying that adding additional devices needs more capacity, thus generating more revenue with little additional cost.

Conversely, this is why home networking worked so well. That single connection could be shared among all the devices with each getting the benefits of the full capacity. It’s also important to understand that the Internet is as much about local connectivity as accessing distant content. We can stream as much (video) content locally without using any of the capacity of the rest of the Internet. This is another benefit of having a home network, rather than only connecting devices to distant servers or, today, connecting only to the cloud.

Because we own the connectivity within our homes, we can choose the products that have the best price/performance. This has driven the kinds of Moore’s law improvements we’ve seen with the rest of technology, moving us from modem speeds to 10Mbps Ethernet to the 1Gbps that is becoming the common within our homes and offices. And we still have a lot more latent capacity evidenced in the 10 Gbps HDMI cables. Improvements in capacity have been more than matched by the decreasing cost of gear and the increasing availability of high performance wireless connectivity.

This is in sharp contrast to the limited capacity available outside of our homes. Moore’s law shows the power of ownership and the need for access to the physical wires and radios—not just “Internet” as a utility. The wires themselves are just materials—copper or fiber. The speed is a matter of how we use them and that, in turn, is a matter of motivation.

With the advent of packet technology, AT&T saw a future in which telephone calls would be commoditized, so they sought to get into the cable TV business using their existing wires and, in the 1980s, the same wires that handled phone calls went from carrying phone calls at up to 56kbps to multi-megabit ADSL lines. The motivation changed when they were told they had to share those copper wires with competitors, but they wouldn’t have to share any fiber connections.

The decision of whether to use fiber or copper should be an engineering decision, not one driven by regulation and marketing. In fact within our homes, the HDMI and USB cables can now go 10 gigabits per second over copper. We should take full advantage of our existing copper infrastructure (and the 25 years of Moore’s law improvements in technology), to get connectivity now at little additional cost. Over time, we can switch to fiber where it makes sense.

The lesson is clear—we need to focus on incentives and creating opportunity, rather than framing policy in terms of the Internet as just another telecommunications service. Perhaps the deeper lesson from what I did in home networking is that, giving people control over their connectivity didn’t require government intervention because people could do what they wanted within their homes. In the same way, we can extend this to private spaces like their apartment houses.

This isn’t to say that government can’t play a role. Today, it may be hard to imagine a time when people didn’t own the wires in their homes, but before the 1970s and the Carterfone decision, that was indeed the case. Today, you don’t really own cellular phones, because there is no clear demarcation between the cell phone and the service—just like in the days when the phone in your home was part of the phone company’s service. In the days of Carterfone, the red/green wires were a well-defined and simple interface.

Today, that interface is the Ethernet jack (or equivalent), where the carrier’s service meets the home. We need to assure that the customers can install their own equipment on premises. The FCC needs to be explicit in assuring an arms-length relationship between the carriers’ offerings, and what we do with the Internet.

This interface provides a boundary between the new model, in which the customers individually (or together as in an apartment house) own their equipment, and the world of traditional telecommunications, where the customers buy connectivity through the carrier’s broadband pipe.

We will soon see (shared) gigabit (or higher) speeds as the norm from market forces, rather than by regulation.

Broadband Competition?

The reason that “broadband competition” hasn’t yielded the rapid price/performance improvements we’ve seen elsewhere, is that competition is defined in terms of competing pipe providers divvying up a market. This is the case even in places, such as where I live, with three providers competing for my business.

Again, we can learn from the experience with home network, where vibrant competition for the users’ business rapidly lowered costs and improve performance.

To understand the problem with broadband competition, we have to look at the Internet apart from the business of delivering cable content. I have both, Verizon and Comcast as providers. If I move the jack from one to another, there is essentially no change.

There is one Internet, so why do we have the expense of multiple broadband infrastructure? We understand that electricity is electricity, and that it doesn’t make sense to have two electric grids which simply doubles the cost without improving service. It makes even less sense to have competing broadband infrastructures, because the meaning of the content isn’t contained inside the wire as it was in the heyday of telephone.

As VoIP demonstrates, we no longer need to depend on the carriers for phone calls (as AT&T anticipated). In fact, we do better on our own, because we can take advantage of the opportunity of video at no additional cost.

The problem for broadband providers is that if the value is created outside of their network, how do they make money beyond providing a dumb pipe?

The owner of a dumb pipe can make money if they control the market, so the customers have a stark choice between paying what is asked or else not being able to communicate. This applies even in the face of competition, because that competition amounts to divvying up the market.

How else can we have redundant capacity with each provider only getting a fraction of the market? If we have two providers, then we have two infrastructures, which doubles the price without providing any more capacity than one provider could provide. Capacity that benefits from Moore’s law, even without building out new infrastructure.

With a single phone network we get the benefit of the network effect (also known by the neologism—Metcalfe’s law). The same is true of a single Internet. We get this effect globally, but deny ourselves the benefits of shared facilities locally, all in the name of broadband competition. We also deny ourselves the benefits of resilience that comes from having rich local connectivity.

We don’t rely on the pipes to keep the packets in order. Each packet can take a different path, so that pooling all the capacity into a common infrastructure, improves resilience since we’re not critically dependent upon any one “pipe”.

By the 1990s, VoIP implementations demonstrated that we could do telephony without depending on the phone company. Yet, even today, some people have qualms about VoIP because there isn’t a network operator to guarantee that it always works. But that’s the nature of markets—we accept a small amount of risk so as to enable the disruptive innovation that reinvents the world. VoIP is just one of the ways we take advantage of all the opportunities around us, rather than being dependent upon what is understood by, and is profitable to a provider.

It doesn’t matter if the provider is a phone company or the city. It doesn’t matter whether we are confined to billable pipes or have access to a common infrastructure. What matters is how we fund this common infrastructure. Must each conversation be profitable to a provider, or can we look for the value to society as a whole?

So, again, we can learn from the example of any office or campus network or home networks. If we treat the wires and radios as common facilities, we are free to communicate. We just need to apply that policy to the larger community, and then revel in the abundance.

Benefits of Internet-Native

With Internet-native policies, we get a world of opportunities unfettered by intermediaries—even those with the best intentions.

We sense the possibilities when we simply access a web page, without thinking about how to reach a site that might be on another continent—something unimaginable not so long ago. As wondrous as that is, it only hints at the possibilities we are denying ourselves with policies premised on scarcity.

To use one simple example—connected health care. The technology for connected health care is already available. Where the market functions, we’ve seen an explosion in the number of health and fitness devices. But often these devices are prevented from communicating directly with a monitor with a doctor, a clinic or a hospital’s computers. Failure is the default because each step along the path must be set up just right, and every billing relationship has to be in place. This makes ad-hoc the exception—each relationship must be set up and maintained manually, as in pressing “agree” or assuring that the provider’s fees have been paid.

With any connectivity to the rest of Internet along any path, we can exchange packets. It’s that simple—everything can become a resource.

Cable, Wireless and Pipes

The term “broadband” was associated with Cable TV which started as the business of relaying over-the-air broadband TV signals. Perhaps one of the most telling inconsistencies of the proposals for network neutrality, is in ignoring the amount of network capacity dedicated to broadband cable content.

If the Internet is the common infrastructure, why do we need a separate infrastructure just of legacy broadcast television content? This is the extreme “fast lane” preference, and any consistent policy wouldn’t treat some content as special.

Public policy also treats wireless bits separately from wired bits. In effect, wireless frequency bands are treated as pipes just like the wires and fibers of landline connectivity. But ultimately there is no difference between those bits and any other bits. Wi-Fi calling is a powerful demonstration of how much capacity is available at no additional costs over other uses of connectivity. Ultimately whether we use radios or wires, should be an engineering decision.

Current trends are already offering alternatives to traditional cable with YouTube, Netflix, Skype Video and now Sling.com, demonstrating how quickly we are moving away from traditional cable. Sling.com is interesting because it’s giving people the ability to subscribe to basic cable channels via the Internet.

The increasing use of Wi-Fi calling demonstrates a rapid shift away from traditional cellular telephony. Wi-Fi calling also allows me to treat LTE connections as a Wi-Fi medium, rather than paying a carrier for Voice over LTE as a special offering.

Behind all the retail offerings is the business of telecommunications companies each with their own infrastructure and complex settlement relationships to balance out the (imputed) cost of carrying bits. I say imputed, because those costs are determined by accountants, as there is no direct path from any particular piece of gear to the cost of carrying the packets.

As the amount of video traffic increases, it becomes more difficult to pretend that all the traffic is in balance. Hence the argument that companies, like Netflix which are generating much of the traffic, should be their (imputed) “fair share”.

We can’t address neutrality without impacting these arrangements. More to the point, however, if we are no longer dependent upon pipes, why do have so many competing pipes?

Rather than creating new policies based on separate but neutral pipes, shouldn’t we be moving towards a common infrastructure so we can get the benefits of synergy?

We can work with the market forces by paying for common infrastructure locally, and then expanding those areas of infrastructure over time until it is the norm and we are not hobbled by being confined to the narrow twisting and winding passages of legacy telecommunications.

And then we get the benefits of “one Internet” rather than having to navigate between providers’ domains. Connecting to my neighbor using a different provider, incurs transfer costs and creates delays. It’s as if I had to take the trolley downtown and back instead of just walking 20 feet.

Towards a Deeper Understanding

When trying to understand what is hard to understand, I run into some common (mis)understandings as outlined below.

Language

It’s hard to communicate if we don’t speak the same language, and even if we speak “English”, it doesn’t follow that we have a shared understanding of what a word means. It is obvious that the word “work” in Physics has little to do with the “work” people are paid to do.

So far, I’ve been using words in their common use. But words have baggage. For example when we talk about high speed Internet, that implicitly assumes we can speed up the pipe in the same way we can build a highway with higher speed limits.

This is why I tried to use the word capacity instead—we can increase capacity any number of ways, be it by gigabit fibers or by making all the facilities available as a common resource.

The term “broadband” has a lot of history. It started as a technical term but came to be used for the business of a provider owning a pipe, and using that pipe to sell services. Today the term has been used broadly for any high capacity delivery system and the services delivered using that pipe.

This becomes a problem when we call the Internet broadband, because the Internet is something we do with that pipe rather than a service with content provided from a carrier. More to the point, the Internet is a different way to use the physical medium—the broadband wire. Instead of divvying up the capacity into fixed bands (or lanes), we use packets. This allows bursts to use 100% of the capacity rather than being limited to a fixed portion.

This is just another example of how the Internet gives us the benefits of sharing a common facilities, whether via a single cable or the entire infrastructure. This is possible because each packet is tagged with a destination, so it can find its own way. You can think of it like driving cars which can each find their way to their destination unlike railroad cards which depend on tracks to guide them.

More subtle, perhaps, is the word “network”. It is a very general word that we use, both networking as something we do, and networking as a service. Or, to put it in more familiar terms, social networking with arranged marriages.

With traditional telecommunications we depended on a telegraph company or telephone company to insure that our messages were carried intact from one place to another. That was the “network” of its time. With the Internet we don’t need a carrier (or a “network”) to do more than “best efforts’ transport of packets. Best efforts means that we are not asking the carrier to take responsibility for carrying the packets.

In order for a carrier to take responsibility, it needs to control the entire path between the two end points, or cooperate with others that do. The Internet doesn’t require a carrier take responsibility, thus there need not be a network owner nor does networking need to be a service. Instead we can have individual entities’ best efforts composite into a whole service.

This is why we can composite connectivity from the edge, rather than depend on a provider to deliver Internet.

Purpose

The implicit assumption is that a network serves a purpose. A phone company’s responsibility is to make sure that phone calls work. A cable company builds a network suited for delivering video content.

We tend to interpret what we see through our current understanding. The Internet happens to be able to mimic traditional telecommunications networks, so we assume that we need to build those applications into the network. Yet, in doing so, we lose the soul of the Internet—its ability to create opportunity for us to discover what we can’t even imagine.

To put it bluntly, the Internet is not a content delivery network though we can use it to do content delivery. In the same way, I can do engineering but am not (just) an engineer, and I can play a musical instrument but I am not (just) a musician.

Our understanding is further distorted by the fact that we’ve repurposed a video delivery network (broadband) for doing “Internet”, and thus shouldn’t be surprised that it is suited for video delivery.

This retro-framing comes at a price because the implicit assumptions of the telecommunications framing can work at cross-purposes with our ability to create our own solutions. TCP allows users at the edge of the network to share a common infrastructure, even when there is limited capacity because the protocol detects congestion and adapts to the limitations. Traditional networks increase apparent capacity by buffering packets in an attempt to insulate the users from congestion, and in doing so, in effect, lies to them about the capacity of the network.

As a result, the network seems to have a lot of capacity until the buffers are full but the user may have to wait a long time for packets to reach the other end as the buffer slowly empties. Even on a gigabit network, each gigabyte buffer could add ten seconds of delay making the network appear to be very slow to users who (and applications that) would normally adapt to the actual capacity.

This is a good example of the fundamental difference between networking as a service, with the carrier knowing what we want, versus the Internet in which we can think outside the carrier’s framing and outside their business model.

These conflicts are pervasive, and simply requiring telecommunications to be neutral with regard to each pipe, results in treating users as passive consumers of content rather than empowering them to be participants and contributors. More reason why it is so important to have Internet-native policies.

Social Policy

One major source of confusion is the counter-intuitive idea that so much of what we associate with the Internet is because social policies are arms-length from the actual infrastructure of the Internet. This frees people to create their own solutions, rather than simply choosing among what is offered to them.

The web and Facebook (to name but one) are just ways we use the common infrastructure, but are not built into the actual plumbing of the Internet.

The End of History

As much as I write about how wonderful the Internet is, we do need to remember that the current implementation is very much a work-in-progress.

Even as the FCC is defining broadband as 25Mbps “down” and 3Mbps up, I’m finding myself increasingly using applications that just assume symmetric connectivity to the “cloud”.

Dropbox, OneDrive, Google Drive etc. are now expected to be as responsive as a local disk drive and, in fact, they do work that well for those of us with symmetric connectivity. In fact I’m writing this document using OneDrive. I also backup my files to remote servers. This is another reminder that we’re not talking about “content delivery”, but rather peer connectivity.

This symmetry is even more appropriate when thinking about peer connectivity among things—the so-called Internet of Things.

The Internet is still undergoing rapid evolution and innovation and we need to avoid freezing engineering decisions in regulations. Designers need to be free to use the appropriate technology based on their merits, rather than trying to work-around laws that have different rules for glass bits.

An Internet-native approach is centered on aligning incentive so that we aren’t dependent upon providers who make money selling content. Instead, communities can work together to hire people to assure they provide generic best-efforts packet transport rather than building-in each service.

The wonder of the Internet is that from such a simple constraint, we’ve achieved world-wide connectivity. If we are too proactive in trying to help, we will find we’ve broken the dynamic and lost the Internet.

For more depth…

For further discussion of Internet-Native and related topics, please go to Further Reading.