Ends and Means
Ends and Means
of the Free Network Movement
- 1 Introduction
- 2 Vision
- 3 Context
- 4 Strategic Roadmap
- 5 Conclusion
An introduction to the Free Network Movement, and an overview of the key points contained in this document.
- The Free Network Movement aims to promote the free and equitable transmission of information in data networks.
- The FNM is accomplishing this aim by designing, specifying, and stewarding the emergence of telecommunications infrastructure that is owned and operated cooperatively, by those that use it - rather than by for-profit and state actors.
- The architecture of the Free Network enables information exchange that is materially, rather than logically, peer-to-peer. We call this architecture fractal mesh.
- The Free Network will be immune to censorship and resistant to breakdown - it is highly distributed and capable of operating independently of existing infrastructure.
- The FNM is part of a global movement towards digital self-determination. It contributes to, and draws from, free software projects from around the world.
- The Free Network technology stack will precipitate an array of inventive commerical applications. Its political and social import is such that it necessitates a non-profit stewardship entity whose organizational process is driven by consensus.
- The Free Network Foundation is registered as a Missouri non-profit corporation, and has submitted IRS form 1023, the application for tax-exempt status as a public charity under section 501c(3).
- We intend to raise $50,000 in our first capital campaign, for the purposes of prototyping and demonstrating a free network.
A description of the architecture and applications of the Free Network, from its component hardware and software, to its commerical, political, and social upshot.
The most salient feature of the Free Network architecture is its ability to enable communications that are materially rather than logically peer-to-peer. In logical peer-to-peer, the pattern of value exchange is peer-to-peer, but the pattern of information exchange is not. This is what we have today with, for example, BitTorrent or GNUtella, Freenet Project or I2p. Value moves from peer to peer, but the information itself passes through an intermediary (the Internet Service Provider). We are not truly engaging in peer-exchange when we involve a paid, third-party provider for bit moving. Such activity fits the peer-to-peer model only at the level of the application layer, a logical overlay, the very highest layer of the network stack.
What the Free Network enables is a radical departure from such a shallow model, towards a model that is peer-to-peer from the ground up. This means that everything from the physical medium of transmission (radio waves), to the pattern of exchange (social networking), is based on horizontal peering. Neighbors talking to neighbors directly, or via other neighbors, eventually without the need for paid bit-transit.
The Five Freedoms
The idea of material peer-to-peer captures this notion, but it does little to explain its far-reaching and appreciable benefits. We summarize these benefits in what we call the five freedoms: access, transmission, storage, authentication, and consignment.
The first freedom is access - In a Free Network, constituents would pay only the actual cost of owning and operating their share of the network. Buy and power a FreedomNode, and you become part of the Free Network - contribute to your local co-op, and make the economies of scale work to your benefit. Compare this to today's environment, where network 'consumers' pay the costs of access, plus a hefty margin, in order to lease a line that is owned by a corporation.
The second freedom is transmission - this is the ability to send bits from peer-to-peer without the prospect of interference, interception, or censorship. The Free Network achieves this aim through the use of cryptographic best practices, and by eliminating the network chokepoints where packet inspection is likely to occur.
The third freedom is storage - the FreedomBox allows people to run their own network services, such as social networking, telephony, and web hosting, and thereby enables them to maintain posession of those bits. Due to the fact that the FreedomBox is in the posession of its owner, gaining access to its contents would require a warrant or subpoena. This is not the case in the current network environment, where bits pertaining to our private lives are scattered and held in various data centers around the world.
The fourth freedom is authentication - people ought to be able to maintian an identity that is verified as authentic by others. This technology is called a 'Web of Trust,' and is built into the freedom-enabling software stack at a low level. Just as important, however, as the ability to present a verified identity, is the ability to present a pseudonym, or to remain anonymous entirely. The Free Network will make the authentication spectrum easily intelligible to its constituents, and clearly indicate whether a given session is onymous, pseudonymous, or anonymous.
The fifth and final freedom is consignment - the ability to perform exacting mechanisms of access control. In large part, this is about making it easy to see and manipulate individual privacy settings, yet it is also contingent upon storing one's data locally. When people own their own data, and are able to decide exactly who can access it, bit consignment becomes a willing, rather than unwitting action.
The Free Network is not just a leap forward - it is also a survival strategy. The current hierarchical network model will not scale to meet the demand of an increasingly networked world. The obvious fix is to keep local traffic local, and that is something that can only happen efficiently on a network that enables material peer-to-peer.
The Free Network will offer a richer network experience at a fraction of the price, and in so doing provide a technological platform for a new wave of innovation. It will enable communities to leverage the economies of scale, driving down the price of network access while increasing network resilience, and reducing our susceptibility to interception and censorship. It will allow us to meet the surging global demand for network access, and ensure that no one ever again goes hungry for knowledge.
The Free Network is unlike what has come before - it is the people's network, owned at once by all and none. The technology to build this network exists. At this point, it is a matter of making it possible for all to participate, through integration, optimization, and packaging.
The fundamental technology that underpins the Free Network is that of the mesh network. A mesh network is one in which nodes pass messages directly to one another, rather than through a central hub. This type of network topology is horizontal and decentralized, meaning that no node is a single point of failure. Mesh networks are self-healing, meaning that the network has the capability of routing around a node that fails.
The concept of the mesh network is not new - in fact, the original premise, and promise of the Internet was of a wire-line implementation of mesh principles. We view the Free Network as a harkening back to those original principles of distribution and resilience from which the network of networks was born. From the perspective of telecommunications companies and Internet service providers, at the level of the backbone, the Internet is a mesh network. The problem lies in the fact that Telcos and ISPs have themselves become points of centralization and control. By implementing a network access layer which is hierarchical, companies such as Verizon, Comcast, and AT&T have gained the ability to route packets through centralized hubs, and inspect those packets in the process. The Free Network represents a departure from that model, and a return to an Internet that is controlled by no one and every one at the same time.
As we proceed, keep in mind that the global communications network is a construction of immense scope and complexity. For any initiative with the aim of bettering that network to be successful, it must rely upon the principle of emergence. That is, the desired changes must come about in a manner that is gradual and organic. What follows is a description of our ends, but not our means. A feasible means of achieving such an end is described in detail in section 3 of this document.
Wireless implementations of the mesh topology have evolved over the course of the last two decades, from a theoretical and experimental novelty, to a field-tested and battle-hardened production technology. Metropolitan-scale wireless mesh networks exist in several European cities, and smaller networks are in operation around the world. The proposed architecture of the Free Network uses the wireless mesh as a design element, but expands upon and improves the idea.
Our innovation is called fractal mesh, and consists in the application and interconnection of mesh networks at different scales. A neighborhood mesh of a few thousand nodes (what we call a neighborhood network) is connected to other neigborhood networks in a regional, backhaul mesh. This regional network is then connected to other regional networks via a global mesh of fiber-optic and satellite routes.
In a Free Network, the people become their own Internet service provider. Instead of paying profit to those that own the infrastructure, they themself are the owner. One recieves access to the network in exchange for agreeing to provide access to others. This is possible using current technology, but it is exceedingly difficult. Our aim is to specify standards for the interoperability of free networks, and to produce networking solutions that assist people in the construction of such networks. In order to achieve these ends, we will build a reference implementation, which can be copied, modified, and improved upon.
Of paramount concern in the implementation and integration of the technologies referenced in this document is their accessibility to all users. The clearest path to a freer network is to take those freedom-enabling technologies that exist today, and make it possible for anyone to use them. That's where the FreedomBox comes in.
FreedomBox is a distribution of the operating system Debian, which will turn a headless, small-form computer into a smart router and personal server. With sensible defaults, a dead-simple user interface, and plug-and-play functionality, it will make it easy for anyone to secure, anonymize, and encrypt their communications. More than that, though, it will make it simple for anyone to participate in a neighborhood mesh network.
The reference hardware for the FreedomBox project is Marvell Technologies' DreamPlug. The DreamPlug has an ARM CPU running at 1.2GHz, 2GB of flash memory, 512MB of RAM, two ethernet ports, and a single 2.4GHz radio. All that needs to be added in order to use this rig as a node in the Free Network is a pair of 5GHz radios. We call this setup - a DreamPlug running FreedomBox with the additional radios, a FreedomNode or FNode for short. The total bill of materials for an FNode could be as low as $150, if we procured contracts with original equipment manufacturers. The 2.4Ghz radio would be used to distribute connectivity to client devices inside the home or business, and the 5GHz radios would be used to communicate with nearby FNodes and the neighborhood FreedomTower.
An FTower is owned and operated cooperatively by a neighborhood network, roughly the size of a census tract. It has several 5GHz radios for communicating with FNodes, and 3650MHz radios for long-range links to other FTowers. It is important for the FTower to be visible to a significant portion of the neighborhood network. Line of sight dramatically improves the quality of radio communications links.
Within any group of 200-300 FreedomTowers would be a single FreedomLink, serving a population of up to a million, connecting them to an Internet backbone. The FLink would communicate with the towers over 3650MHz, and maintain a mutihomed fiber connection to the Internet Protocol core. It would, again, be owned and operated by the community, allowing them to participate in the actual Internet by speaking BGP to other networks.
With community-owned fiber and satellite routes between FLinks, the picture is complete, and the constituents of the Free Network would be able to purchase network access at cost.
This may sound like no small feat, and that's certainly true, but the Free Network Movement has a truly practicable plan for making this vision a reality. Our strategic vision is explained in detail in Section 4 of this document.
An overview of major stakeholders in the global network, including state and corporate actors, followed by a rundown of key initatives in the distributed networks problem space.
The Free Network Foundation is an American organization with global ambitions.The United States is not only our base of operations, but it is also, in many ways, at the heart of the network. As such, this stakeholder analysis focuses heavily on aspects of American enterprise and regulation. The telecommunications space is highly complex, but can be roughly divided into three types of networks, performing some combination of three different services.
Networks are Tier 3, Tier 2, or Tier 1, with each network providing some combination of access, transit, and peer services. Let's look now at each type of network, and get a feel for the services they provide.
Tier 3 Networks
Tier 3 networks are essentially resellers - traffic does not cross a tier 3 network, but originates or ends there. A tier 3 network purchases bandwidth from an upstream provider at a Point of Presence, and delivers that bandwidth to end users. They do so either by building a network, or by leasing the requisite lines from an incumbent operator. This type of operation is termed access, or last mile.
Conventional last-mile models for broadband delivery are cost effective only in areas with population density above a certain threshold. Those in unserved rural areas must resort to expensive, high-latency satellite solutions, or settle for dial-up speeds.
Access operations are those elements of the communications infrastructure with which the customer comes into contact. These are the lines that branch out from Central Offices into each home or business. In the existing model, these circuits are exclusively vertical - that is, they allow connections only from a client to a service provider, but not from client to client. This prevents the exchange of information in such a way that the service provider does not function as a paid intermediary.
Of the Autonomous Systems that make up the Internet, the vast majority are Tier 3 networks. There are upwards of twenty thousand such networks in operation today, generally outside of the United States.
Tier 2 Networks
In addition to engaging in the access operations described above, Tier 2's have regional reach, and trade or sell bandwidth to other networks. Network to network traffic is classified as either peering or transit.
Tier 2 networks are large enough that they are able to peer with some other networks for mutual benefit, but not so large that they are able to completely avoid paying for bandwidth from a more widely connected network. When networks peer, they agree to exchange traffic without the need for monetary settlement. When a network purchases bandwidth from another provider, it is said to purchase transit across that network, or simply to buy transit.
To build a Tier 2 network requires significant capital investment, even in the hundreds of millions or billions of dollars. In addition to being connected to end users and entire access networks, Tier 2 networks often connect to Internet Exchange Points, where Internet Service Providers can openly exchange IP traffic.
The roughly three thousand networks of this type do the majority of bit moving on the Internet. They are much fewer in number than the Tier 3's, but they are much larger, on average, and have a much greater aggregate capacity.
Tier 1 Networks
Tier 1 networks have global reach. These are well-connected bit moving platforms, worth many billions of dollars. From the perspective of a Tier 1, all other networks are either big enough to mandate a peer relationship, or are willing to pay for transit.
Building such a network entails laying thousands of miles of fiber, accross oceans or continents or both. The majority of Tier 1 networks are based in the United Sates, even if their core network includes points of presence overseas.
There are a very limited number of Tier 1 providers - and perhaps none that truly do not engage in settlement at all. Still, there is a group of ten or so networks that are well understood to qualify as Tier 1.
Tier 1 carriers have made significant capital investments in recent years to improve and expand their capacity. Still, global data exchange is increasing at rates that outpace the ability of major carries to provision new infrastructure. A paradigm shift is needed in order to sustain the network's growth.
Peer-to-peer, decentralized, and distributed systems have been the topic of much research and development in recent years. What follows is an examination of some initiatives that have emerged in the problem space, and the ways in which those initiatives complement one another.
Federated Social Web
In the past few years, many initiatives have emerged that aim to build a social web built on open standards, where a federation of servers, rather than a single behemoth, stores user profiles. Notable efforts include GNU Social, Friendika, Buddycloud and Diaspora. Not all of these projects are interoperable as yet, but they are increasingly converging on a set of standards.
Of note, however, is that all of the federated social platforms mentioned above, regardless of other design parameters, are intended to run on a server. The ideas motivating these efforts are good ones: to encourage decentralization of infrastructure, and help people take some control of their own data. Yet, existing solutions have not accomplished this outcome, because the barriers to entry and technical knowledge required to operate today's servers are prohibitively high. Thus the need for nodal computers to take part in the federated social web.
Eben Moglen's vision of a low-power, headless home home server that just works has spread like a shockwave through the freedom-loving computing community. The ramifications of the idea are clear - it would allow ordinary users to own their data, secure their communications, and maintain their privacy.
There are number of large, outstanding challenges in the effort to create a home server that's easy to use. While the community has been more-or-less able to converge on elements of the server backend, little progress has been made in the way of a user interface, or in provisions for high fault-tolerance and reliability. If the FreedomBox project is successful, there is no telling the importance of the role that these boxes will come to play in their owner's lives. To achieve the goal of widespread adoption, it is of critical importance that the box be able to run continuously, without major interruption, for years on end. It will need to be securely and automatically backed up, so that a system failure doesn't represent the loss of one's entire social graph and media archive.
These problems are challenging, but by no means impossible. Like any product or project, the FreedomBox will take years to reach maturity. As the nodal platform crystalizes and the userbase grows, it is time to begin exploring what freedom-enabling systems could be constructed using the no-fuss, always-on home server as a building block.
It is not enough to settle for communications that are encrypted, but peer-to-peer on the logical level only. The advent of the nodal computer represents an opportunity to change the nature of our communications in a more fundamental way. Yet, in order to do so, we will need a naming system that is decentralized, and mesh networking technology that makes configuring a node as easy as turning it on.
Distributed Social Networks
There are also social initiatives which operate in a more P2P-oriented style with all social computation and modeling happening in a software running on the user's device itself. Social Swarm, a working group of the German digital rights foundation FoeBuD, is researching such solutions as they bear the very favorable feature of enabling end-to-end encryption between all communication members, leaving no unencrypted data on server nodes. In contrast to the federated social web, this solution would require no installation of nodal computers in people's homes.
Distributed Global Names
There is a classic problem in network theory, commonly referred to as Zooko's Triangle. It states that, at best, a name can have two of the three following characteristics: secure, distributed, and meaningful to humans. Names that are globally unique and not controlled by a central authority generally end up being ugly strings of bits, such as IP addresses or [[|Tor|.onion addresses]].
The existing solution to this problem is the widespread use of names that are secure and meaningful, but controlled by a central authority. This system is called the Domain Name System, or DNS, and is ultimately under the auspices of the United States Department of Commerce National Telecommunications and Information Administration - the NTIA.
Yet, there is something new on the horizon, predicated upon a relatively new technology that is just reaching production-level maturity. The Distributed Hash Table is a mechanism for storing a set of key-value pairs across many separate machines. By itself, it does not solve the problem of Zooko's Triangle, but were it to be coupled with a mechanisms for pet-naming and access control, it could form the basis of a human-usable, globally distributed naming system. Such a naming system would help devices such as the FreedomBox find one another, regardless of ISP policy. It would also allow for seamless integration between material and logical forms of peer-to-peer communication.
Wireless mesh networks, as mentioned before, are networks where nodes are connected to one another horizontally and redundantly. Mesh nodes can connect to one or more of their peers, and not just to an upstream hub. Wireless mesh networking enables local communications without the need for a paid Internet Service Provider. Mesh technology has come a long way, but still has some severe limitations: there is no user-space utility for easy construction and management of mesh networks, and those mesh networks that do exist are used almost exclusively as means of accessing the Internet.
At present, there are two leading algorithms in the arena of mesh routing - Optimized Link State Routing, and the Better Approach to Mobile Ad Hoc Networking. These protocols have been employed and applied by a number of community networks and research groups - notably FreiFunk, FunkFeuer, The Serval Project, The Village Telco, The Commotion Project, and Project Byzantium.
Optimized Link State Routing, or OLSR, is widely utilized. It has been deployed by community networks such as Austria's FunkFeuer to great effect. Though recent iterations have decreased CPU usage, and improved throughput, OLSR's primary drawback is heavy CPU usage, especially in discovering and repairing routes.
The Better Approach to Mobile Ad Hoc Networking, or BATMAN, emerged from the German FreiFunk community. Its latest iteration, BATMAN Advanced, works at a lower level of the network stack than other mesh implementations, and has now been incorporated into the mailine linux kernel. It has been utilized by The Village Telco, in the creation of their turnkey mesh router, the Mesh Potato, and by the Serval project in the creation of the Batphone Android application. Batphone allows the user to engage in mesh-based telephony using ordinary phone numbers.
Other significant mesh networking initiatives include The Commotion Project and Project Byzantium, both based in Washington, DC. The Commotion Project is an effort to integrate and standardize the use of existing mesh technologies on a variety of hardware platforms, and is part of the New America Foundation's Open Technology Initiative. Byzantium is being developed by a group of enthusiasts from HacDC, with the aim of building a Linux LiveCD that supports materially peer-to-peer versions of various network applications (web, telephony, chat) for use in emergency situations.
Wireless mesh technology has progressed over the course of the last decade to the point that it can be reliably deployed in production environments. Still, such deployments must be carefully planned and administered. The key outstanding challenge is to make it easy for anyone to build and run such a network, and to do so in such a way that users are encouraged to take advantage of the opportunity to route traffic locally. This could be accomplished by integrating mesh technologies into a nodal platform that includes sufficient radio hardware.
A stragetic roadmap for the development of technologies that support material peer-to-peer and the deployment of fractal mesh infrastructure. An exploration of how humanity can achieve co-ownership of network infrastructure, utilizing existing routes where possible, and provisioning new ones when necessary.
The path to network freedom begins with the advent of sovereign computing. Nodal platforms will allow any user to host their own network services, perhaps without even fully understanding that they are doing so. It will just work, because it has to, in order to be adopted. We term such usage sovereign computing, because it allows users to maintain control of their identities and their data.
The nodal server will assist in the transition from centralized and unsecured communications to ones that are distributed, encrypted, and logically peer-to-peer. There is no question that this represents a major improvement from the status quo. Yet, the long-term evolution of the network demands a more radical approach, and the nodal platform presents the perfect opportunity for such a departure.
A communications platform incorporating publishing, messaging, status updates and telephony will be an attractive prospect to early adopters. A streamlined user interface and high reliability will drive wider adoption.
The Neighborhood Network
Long-term planning at the present juncture will pay dividends when sovereign computing becomes widespread. Including three b/g/n radios in the nodes will ensure that neighbors can find each other and establish robust and reliable routes of communication. An integrated software stack means that user communications will automatically be routed locally, whenever it is possible. The economies of scale dictate that users will pool their resources, and purchase Internet access collectively. This will drive further adoption.
While it will be possible to engage in bandwidth sharing using nodes alone, larger scale cooperatives will want to invest in a FreedomTower containing powerful radios capable of communicating with mobile devices, nodes, and other towers. Community-owned towers, in addition to serving as a logical connection point for communal outbound connectivity, will allow neighborhoods to communicate directly with adjacent neighborhoods, and in so doing extend the reach of material peer-to-peer connectivity.
Neighborhood Networks will purchase access from Tier 3 and Tier 2 network providers in a manner similar to what small and medium sized businesses do today. The price on a per-node basis could be as much as an order of magnitude below current residential levels.
Just as the adoption of FreedomNodes will drive the adoption of FreedomTowers, the adoption of FreedomTowers will ultimately give rise to regional FreedomLinks, and just as the construction of a neighborhood tower will drive the adoption of nodes, the adoption of a link will drive the adoption of towers. These links will sit in Internet Exchanges and Colocation centers, participating in the regional radio mesh, and in the global network of networks as a peer, rather than a client.
The population served by a single link will be quite large - perhaps as many as a million individual nodes. These regional networks will still have to buy upstream connectivity from an Internet Service Provider, but they will finally be able to do so in the same way that a Tier 3 network purchases transit from a larger network.
In operational terms, these regional networks will constitute Autonomous Systems. They will be able to peer with other networks, driving down the cost of connectivity even farther. It is at this stage that the constituents of the regional mesh truly become their own Internet Service Provider.
Backbones of our Own
Still, regional Autonomous Systems will be localized, and radio frequency communications are not capable of serving as ISP-grade backbone links. FreedomLinks will initially be connected to one another via upstream fiber networks, so stopping at this point would leave our ability to communicate with one another in the hands of a few for-profit entities, whose terms of service may not be agreeable.
In order to achieve true freedom, regional networks will have to either build their own fiber lines to neighboring links, or purchase existing lines outright from established providers. As these lines are procured, existing pockets of material peer-to-peer will grow together, and larger federations of free networks will emerge. These federations will grow in size and operational disposition from Tier 3 to Tier 2, and eventually to Tier 1 networks. That is to say that when the federated networks grow large enough, they will be able to provide transit to other networks, and eventually engage in settlement-free peering on a wide scale.
To the far stretches of the federation, a user would be able to send information using infrastructure of which they themself would be a stakeholder, a participant, an owner.
A Human Right
Our freedom alone is not enough. The constuction of a global Free Network is not the end of our struggle. The end of our stuggle is to ensure that every member of humanity is afforded access to such fundamentally transformational technology. Once we have built our networks, it would be wise to help others build their own.
At this stage, the ideas contained in this document are just that - ideas. Yet, all over the world, individuals are reducing these ideas to practice. They are itemizing the tasks that lie ahead, and they are executing those tasks as quickly as possible. There are many obstacles to success, but the stakes are too high to take no action.
We hope that you will support this cause in any way that you can. If you are a systems engineer, a programmer, a packager, or a hacker of any variety, we can use your technical assistance. Yet building the free network technology stack is only a part of our work - we face face social, political, and philosophical hurdles as well. You can contribute much through activism, advocacy, and discourse.
The release of this document marks the commencement of our first capital campaign. We aim to raise $50,000 for the purposes of specifying and prototyping the first set of FreedomNodes and FreedomTowers. This funding will be used to procure necessary equipment, and to enable developers to devote more time to this important work. More information regarding our finances is available on our website, under 'Donate.' We rely on public support to do our work, and donations to the Foundation are tax deductible. Please give whatever you can afford, be it your time, your money, or your expertise.
Join us in our struggle to build a freer network. Our very ability to communicate is at stake. If we value our freedoms, it is imperative that we act to preserve them. We are taking action to build a better network for all. This is our struggle - we welcome you.