Our Internet!
The Internet is not something we connect to. It's not even a network like the phone network. It simply the name for the community that can connect using common protocols. Instead of thinking about networks and broadband we must look for transports that create the opportunity to connect. We should pay for copper and glass and radios and not have funding by buying service we can do ourselves. It's not about the money -- it's about the lack of opportunity.14-Dec-2006

We are the Internet

We don’t need be given “Internet” by a central authority because the defining principle is that we can and must connect from the edge rather than relying on carrier be it a Telco or “Internet Inc”.

I’m been arguing that we should own the wires (or transport) in our community just like we own the wires in our homes. In order for this to make sense you need to know not just what the current Internet is but what it can be. Today we are using a flawed prototype – the real Internet can be so much more.

I’m not going to give a detailed design – my goal is to simply give the reader a sense of what it means to take control over our infrastructure and the necessity of doing so. The goal is to facilitate cooperation rather than enforcing it. Simple conventions for packet format help and at a higher level some conventions like HTML will reach the levels of accepted standards simply because they attract a large community.

I’m writing about more than just taking control of today’s Internet. The principle goes much further and applies to the design of systems in general. It can help us understand why governance is often counterproductive by shielding our misunderstandings from the scrutiny of marketplace forces.

Executive Summary

Today’s Internet is treated just like another service in the image of the phone networks or even another channel on your Broadband Set Top Box. And to a large extent that image is reinforced by the central authority of ICANN as the dispenser of identifiers and identities.

But this isn’t in the spirit of the Internet which is defined by the user/developers outside the network. The Internet is just a name for the community of those cooperating and using common protocols. Instead of viewing the Internet as something we connect to we should view it as radiating from us and our devices. Our home networks reach out through community networks and beyond finding paths that work.

It’s akin to shifting our reference from choosing which schedule train we can take to simply driving our own cars and using whatever path we can navigate.

It’s time we took control over our local transport – just as we own the wires in our homes and the roads in our communities we must also own our local information transports.

We must not tolerate being forced to buy services from providers that have a stranglehold on our wires – whether they are physical wires or radios. Today our ability to communicate is limited by the unenlightened business needs of the carriers. This is intolerable and inexcusable.

The situation is tolerated because it seems obvious that we need someone to govern and operate the Internet.

We don’t. The expedient compromises that define today’s Internet are not necessary. We do not need an organization to hand out identifiers. we do need not an omniscient network operator to keep track of all the local networks and relationships. Most of all, we do not need a centralized directory to assign names.

Very briefly:

  • Skype and other P2P systems already provide their own stable identifiers independent of the IP address. Today’s operating and database systems already generate their own unique identifiers or GUIDs. This means we don’t need a separate system, the DNS, to provide stable identifiers.
  • By not requiring the IP to double as a stable handle we are free to use them to describe routes as the paths change. Thus removing the need for a single central network operator.

This means that we can indeed connect by simply finding a path between two identifiers even as the network changes and the end points move.

In effect we can remove the training wheels from today’s prototype and realize the basic design principle of the Internet. The Internet itself is nothing more than the community of those who cooperate outside the network – at the end points.

Once we understand that the Internet can arise purely from the edge we are then free to take control from the edge.

In light of this understanding it is difficult to defend or tolerate giving control over to an operator whose interests are in direct conflict with our need to communicate.

What now seems necessary becomes an egregious example of restraining trade and an inexcusable violation of free speech (as per the US constitution).

In the US the FCC’s mission of assuring a stable marketplace puts in the role of protecting the industry from scrutiny. Once we no longer confuse the services with the transport the FTC is free to address the antitrust issues raised by having the service providers in exclusive control of the transport. Competition is not effective if they act in concert to prevent the creation of new services. With abundant capacity we can ask why the carriers are able to exert such stifling control and why free speech has been given such short shrift.

But the biggest impetus for change will come from investors who realize they have far more to gain if the carriers were not in the position to limit opportunity.

Once we escape the Regulatorium and it’s service-framed model we can recognize the value of the transport as infrastructure and fund it directly and transparently.

Once we understand how to take control from the edge and how this dispersed control produces even more capacity than we see in today’s Internet we’ll be able to apply this principle to marketplaces and beyond.

In More Detail

Opportunity rather than Promises

It’s natural to view the Internet as if it were another network like the phone network but that misses the point of the Internet itself. It’s a real example of Janis Joplin’s Freedom’s just another word for nothin’ left to lose”.

The Internet’s designers needed to interconnect local area networks but had to make do with whatever connections they could find and couldn’t make any promises. Rather than succumbing to Janis’ despair, programmers were able to program around the constraint and in doing so found they had wrested control from those who insisted on telling us how to communicate and how much to pay.

As impressive as today’s Internet is, it is still a prototype on training wheels. Users are still dependent upon what I call “Internet Inc”. They have to get their access from a service provider. To participate they need to get the use of an IP address and, because the IP address isn’t stable, they need a stable name in the Domain Name System. The DNS name isn’t even yours – it is loaned to you under rules set forth by ICANN.

Freedom from the Internet

If the Internet itself arises from taking control ourselves outside the network then we needn’t be dependent upon Internet Inc!

This is not at all theory. It’s practiced every day in the P2P world. P2P is really just a rediscovery of the basic End-to-End principle of the Internet. Unfortunately that’s often confused with Womb-to-Tomb because it’s hard for people to understand how you can make promises without having full control of every element of the network. The Internet isn’t a network like the phone network – it’s just a word we use for what happens when those with nothin’ to lose and everything to gain get together. They form their own communities.

The Internet provides opportunity rather than guarantees

There isn’t any Internet Inc to make promises! The carriers don’t have enough control over enough of the network to make promises. They can’t guarantee service across the entire network. The best they can do is impose limits on the first mile of the connection where we have granted them exclusive control. Without that control they would have no bargaining power and nothing of value to sell.

The big surprise is how well the Internet works on its own terms. Once you can take advantage of any opportunity you discover you have abundant opportunity. It’s the way evolution works – if you play roulette and only one number can win then you don’t have much choice. But if you can find opportunity in any number that comes up then you’ll be a sure winner. And that’s what happens with the Internet – if you don’t have much capacity you’ll just use email and that provides enough value for the market to deliver more capacity. One day Voice over IP (VoIP) went from a novelty to something we just assume will work as if by magic – we simply crossed a threshold in capacity.

If you have enough choices the odds are in your favor. If you have only 10 molecules there’s a high probability they all leave the room at once. If you have a zillion then you can, and do, bet your life that they won’t all leave at once.

No wonder users chose to use modems rather than paying for ISDN and no wonder Verizon’s Genuity couldn’t make a profit by selling high value bits. The odds against them were and continue to be astronomical.

Don’t Confuse the Demo with the Real Thing

The network within your home is just as much “the Internet” as a broadband pipe we use to deliver “Internet”. Maybe more so because you’re in control. Your light switch can use the same protocols to tell the light bulb to shut off as you use to purchase a book on Amazon.

I’m fudging a little. Today’s Internet protocols were compromised because, after all, it was just a prototype. In the 1970’s every bit counted so you had only 32 bit addresses and computers didn’t move around much so you didn’t worry about the small amount of housekeeping necessary to keep track of the machines. Some of us were already using personal computers and objected but as usual we were viewed as unrealistic. As the housekeeping problems increased and the IP addresses became a scarce resource we created the DNS and then ICANN to manage the scarcity.

What we call the Internet today is really just a prototype or a demo. It’s normal to cut corners and fudge things a bit for a demo. At least this is working demo not just a mockup. Unfortunately people confuse the demo with the real thing and assume the artifacts are necessary properties rather than accidents.

Transcending the Prototype

Xerox didn’t use the IP address. Instead it used the machine address on the network board. If you moved a machine it still had the same address. In today’s Internet the number assigned to your network or LAN is part of your 32 bit address. In the Xerox network it was only a hint. If you moved your machine to another location the network routers would find you again. In the 1970’s it may have seemed too difficult to keep track of all these machines and the 48 bit network address may have seemed too big.

An IP (V4) address is 32 bits and written in four groups of eight bits called octets and written in decimal. In 192.55.226.1 the first 24 bits represent the network number and the last are the local portion. 169.254.1.1 has 16 bits for each part.

Yet in today’s Internet the challenge of keeping track of all these LANs is nearly intractable. Even worse, the DNS which was supposed to provide stable identifiers for linkages has gone toxic. The seeming innocuous decision to use human-meaning tokens has led to a disaster – the names are now so valuable that they are only leased and not owned! Thus the network is guaranteed to unravel as these leases expire! At the same time the naïve assumption that order only derives from a central authority has given those charged with managing the DNS, ICANN, a pretext for imposing stifling regulations. It’s as if those acting in the name of the marketplace have undermined the fundamental principles of the marketplace by denying ownership and control to the participants.

There is no exaggeration in saying that this is a disaster – if the DNS names evaporate the Internet must unravel. We see this anytime we try to use an old link – it will either fail or wind up at the wrong and often inappropriate destination. Our ability to work around problems and the fact that the leases can take years to expire make the problems less visible though no less dire. Our low expectations make us ignore the failures we do see.

The ability for today’s Internet to operate despite these obstacles is a testament to the strength of the defining principle. Yet, just as we didn’t know the potential locked up in the traditional communications infrastructure until we discovered the Internet, we don't know the price we are paying for accepting the fiction of a telecommunications industry that uses their remaining control of the transport to limit our opportunity.

The IP address is the instrument of control – it has to serve as both the path and the identifier and can’t serve both roles.

Relationships not Networks

The Xerox approach was far better. Your identifier didn’t change just because you moved your computer to a different LAN thus no need for the DNS.

I didn’t learn about XNS until after David Reed and I talked about viewing the Internet in terms of relationships between end points rather than in terms of the packet transport itself. In cleaning up I came across a Xerox design document and appreciated the insight of separating the name from the location.

Today we can do far better and have real examples. We regularly coin our own identifiers. Globally Unique Identifiers or GUIDs are the basis for today’s information systems. We can also maintain the databases that allow us to find the end points. Skype is just one example. Think about the challenge of tracking cell phones.

The identifiers themselves are simply numbers that are only meaningful within communities. Not only don’t we need the DNS, the idea of a single global database doesn’t make sense.

People have already learned to look up names rather than assuming you can use JohnSmith.Names to find the one John Smith but we still have the illusion that the DNS can do just that. Humans are very good at ignoring dissonance.

Once we separate the path from the identifier we are no longer dependent upon an omniscient monolithic “Internet Inc” operating a single backbone. Tracking is done outside the transport thus greatly simplifying the task of routing -- but that’s another topic.

What we can observe is that today’s phone network took more than a decade to get cellular handoffs to work. Skype does all of this outside the phone network so it can “just work” no matter what path we use. There needn’t be the concept of a dropped call – your conversation always exists but you may have periods when you can’t find a route such as when you’re incommunicado in an airplane. There is no excuse for this isolation. Boeing’s Connexion demonstrated that we can indeed be connected even while in flight.

Once we recognize that we don't need to depend on the backbone of The Internet as a layer we are ready to move beyond the idea of interconnecting LANs. Instead we will think in terms of the relationships between end points wherever they may be. If you visit another company you should be able to safely share the local transport.

Even if we are isolated within an airplane we can still communicate with others in the same aircraft (modulo superstitious fear of radio waves).

The Internet Redux

The Internet is not just about going to web sites far away, it’s about making connections between local end points too.

The home (or office) network is indeed part of the Internet community but we haven’t even started to take advantage of its potential as a basic architecture for making connections.

Today we have a plethora of special wires defining relationships. Printers are directly connected to PCs, a screens (and TVs) are directly connected to video sources. Each connection has it’s own special kind of wires because each analog signal decays in its own way and we need a wire that attempts to preserve the appropriate property for the particular video and audio and even data being carried.

With digital systems we no longer need special wires – we just need to preserve bits and can even correct errors if there is a problem. Or we can ignore lost bits if we don’t have the time or capacity to do so. We can share a common transport because the relationships are defined by end point identifiers or IP addresses (using today’s protocols).

Imagine replacing all that complex cabling with a single wire or even no wire at all. Today’s home networks can easily carry a gigabit per second per wire and even without a wire we are increasing the capacity which is already over a hundred megabits per second.

The relationships we define aren’t confined to the house. We can have a medical monitoring device that is connected “directly” to a doctor’s office. Naturally this raises concerns about privacy and security but that’s a topic in its own right. The short answer is “topology”. In the same way that the relationships aren’t tied to particular paths, we can use cryptographic techniques to make them invisible to everyone outside of a community and there can be any number of “local” communities. And we define these communities ourselves. This is true empowerment.

As we gain control over our relationships we are not limited by the simplistic assumptions of today’s Internet. We needn’t be visible unless we choose to be. Our email address is no longer our “true name”. We can choose whether to make ourselves available and manage the relationships.

From the Edge

If we extend the idea of owning the wire, or transport, in our homes, what happens when the packets have to leave home? How do they get to the destination? How do we even know where that is?

We can greatly simplify the problem if we can assume abundant capacity. If you have a very wide river you needn’t worry about adding a few more cups of water. You’ll only have a problem if you have to combine two rivers. In keeping with the opportunity dynamic those applications that use only a small portion of the capacity will not have a problem. Someone attempting to add another river may have a problem but we’ll still be able to add our cup of water.

This is very important because it allows us to dispense with imposing policies on individuals’ behavior and that means we can indeed form the network from the edge.

With abundant capacity we needn’t worry about optimizing the path – we just need a path that works and then deal with the failures and occasional congestion. With a mesh architecture as opposed to a distribution architecture we can indeed route around failures.

If we distribute responsibility we’d want what I call “way stations” rather than routers. A way-station can take local responsibility. For example, it can sit between a city’s transport and regional transport and take care of finding a path to edge of another city. It can also take responsibility for dealing with failures. We needn’t quibble about terminology but I want to make a distinction between a router in a top-down architecture and a way-station that has more ability to act as an intermediary along the route.

Because the Internet isn’t a single thing we don’t have a single entity that takes responsibility for all of the details. We finesse the complexity of maintaining today’s Internet.

By eliminating the notion of global responsibility we’ve greatly simplified the Internet itself. You solve most problems by adding local or spanning capacity. You no longer have the responsibility for knowing about every little detail of a monolithic Internet. We also avoid asking governments to implement complex policies—they only need to assure sufficient capacity along spanning paths. In the same way we can add local capacity as needed.

This is a very efficient way of solving problems because it’s very inexpensive to add capacity and we avoid arguments over complex policies.

Finding Things and Places

“Finding Things” is a very general statement. Technically we would call this binding. If you want to get to my site you lookup www.frankston.com in the DNS and get an IP address. You that you should use the name not the IP address because the name is more likely to stay valid. But how do you know it’s Frankston.com in the first place? If I want you to reach my site I’ll just tell you the path or where to find it. Of I can be listed in directories or choose to be found by search engines.

It’s similar to traveling to visit someone in another city. You need to be told or to lookup the address and then you can find the path. Or you might be told to ask a secretary where the person is at a particular time.

You don’t worry about all the details – just the major highways and local roads.

At a technical level there are many ways to facilitate maintaining a path without introducing delays. We should also be encrypting all traffic since we don’t know whose transport we are using. Encryption has the added benefit of protecting us when the destination has moved—the message is only meaningful if we reach the intended destination.

There is no sharp distinction between the places as destinations and things (people, conversations, documents) as destinations. Unlike today’s Internet where we can only address machines and then have to resort to artifacts such to reach particular points within a machine (such as an email server) we don’t have a single hard edge. The edge is basically a point of loosely coupled relationships.

Plus ça change, plus c'est la même chose

Pardon my French as I mangle the metaphor. Things are really changing but we can maintain the illusion that they are not.

We can continue to use the familiar Internet by simply using current IP addresses as destinations and we can maintain the illusion that the IP address isn’t changing by using techniques such as tunneling and meshing.

An IP mesh is simply a group of access points that share a common set of addresses. An access point is simply a connection to the rest of the Internet.

Spoofing works both ways, we can connect using the existing Internet as just another transport in the same way we tunnel over our broadband pipes.

We don't need to change everything – we only need to take advantage of existing opportunities while creating new ones.

And Everything Changes

Today’s Internet has given us a taste of what is possible when we can simply connect to end points without being dependent upon the accidental properties of the path. Imagine what happens when we go beyond merely connecting computers.

Perhaps the biggest impact will be on those things that seem too mundane for technology.

Why do we have so many remote controls when all we need are simple protocols and relationships? If we use open protocols we can do more than just replace remote controls, we can use software to act on our behalf.

If a city wants to turn on and off street lights then each one should have a tiny server in the light bulb and just send messages. The bulbs can also report their own failures either explicitly or by failing to call in. Putting a process into a light bulb is not new – Philips already has bulbs that implement a simple policy. If you turn them on they stay lit for half an hour – if you do it twice they stay on. New LED light bulbs already have circuitry so adding an additional capability shouldn’t add much expense but gives us control over our physical infrastructure.

Without the concept of “access’ there is no place for a tollbooth. Of course we still pay for the physical infrastructure like we pay for any shared facility. Without the need to charge by the path we’re free to extend wireless coverage everywhere. Thanks to the protocols anyone can extend the access.

This will have major economic impacts. Owning a pipe no longer has value – be it within a city or between countries and thus the telecommunications business would not exist as such. Instead we have a business in offering services using the transport and in maintaining the transport.

Perhaps less obvious, the concept of radio stations goes away because we no longer have frequency bands but we will have successors that offer particular content or services such as choosing content for you. Note I didn’t say television – that’s already history. Without the broadcast industry it would be difficult to impose stifling controls such as the broadcast flag. In fact, the entire FCC’s attempt to enforce policies and censorship would be moot.

We will all be winners in being able to solve our own solutions and share them with others.

But the concepts themselves will have the greatest impact. For healthcare it’s obvious that we can then do more home medicine thanks to the ability to monitor and even take action. But all the devices within a hospital can be connected by simply establishing relationships. Even if you go to a distance center for imaging you the images are as available as if you did them in the next office.

But for now the goal is to provide the opportunity. The major lesson of the Internet is that these disparate efforts to composite into a coherent and dependable whole. And this has social implications that go well beyond the technology itself. The very idea that meaning is only at the edge and not imposed is also the essential idea of Darwinian evolution. Biology is only one example.

I don’t want to end on a naively idealistic note. Any ecology has critters that are considered problematic if not loathsome. But they too are part of the ecology. It might be counter-intuitive but without enforced governance it will be easier to deal with many of them because we are no longer forced to be publicly available and exposed. We will gain control over our availability. But ecologies evolve and the new threats will be, well, challenging.