Next: Libre Services: From concept Up: Libre Concept Previous: The Subscriber Services industry Contents Index
We are proposing a radically new, completely non-proprietary model for the delivery of Internet services. We call this the Libre Services model.
Libre Services are an extension of the principles of free software into the domain of Internet services. Free software allows complete freedom of action: it may be copied and reused without restriction. Libre Services provide equivalent freedom of action: they are Internet services that may be copied, modified, reproduced, extended, and redistributed in their entirety. Libre Services are:
They are a communal resource, not owned by anyone, freely available for use by society at large. Any company, organization or individual can reproduce and host any Libre Service, and deliver the service to others. Or any group of individuals can host the service for themselves, thus acting as their own service provider.
The Libre Services model exists in relationship to the proprietary Subscriber Services model of AOL, MSN, Yahoo and Google, in an analogous way to how GNU/Linux exists in relation to Microsoft Windows. Both Libre Services and GNU/Linux are open and free models, and both provide the essential freedom of action that is absent from the closed model.
The figure is not unlike the common ``hourglass'' representation of the OSI (Open Systems Interconnection) and Internet architectures, in which a high degree of heterogeneity at the upper and lower levels is bridged by a common set of protocols in the stem of the hourglass.
Figure 1.2 has a similar shape, but here we are showing how a high degree of heterogeneity among hardware platforms and user environments is bridged by a unifying set of software components. At the bottom of the figure is the hardware level, consisting of the large number of hardware platforms and architectures available for client side and server side computing.
In the open-source software world, the key enabling component is the GNU/Linux operating system, providing a complete environment for open-source software development. Linux, the operating system kernel, is the unifying interface for running GNU on a wide variety of hardware platforms.
(To place things in historical perspective, the GNU Project was founded in 1984. The earliest version of the Linux kernel was released in 1991.)
An important development in the evolution of the open-source software movement was the appearance of GNU/Linux distributions, or complete GNU/Linux software packages, assembled together for easy installation and use. The first of these appeared in 1992, soon after the first release of the Linux kernel.
Distributions play an essential role within the GNU/Linux framework. The integration of the various GNU/Linux components into a usable system is not a simple process. Distributions eliminate the need for a developer to locate, download, compile, install and integrate a large number of necessary components into a working GNU/Linux system. Instead, the complexities of system construction are handled transparently by the distribution software.
Our initial Libre Services implementations are based on the Debian distribution of GNU/Linux []. Debian was founded in 1993, and has emerged as the most practical and reliable distribution for software engineering development. Equally important, Debian fully conforms to the philosophy of the free software movement. The Debian project is guided by the Debian Social Contract [], an explicit statement of the philosophy and guiding principles of Debian.
The Libre Services Integration Platform (LSIP) is a generalized framework for developing Libre Services. All our initial implementations are based on LSIP.
LSIP is a set of tools, policies and conventions for services development and deployment. It provides a uniform, disciplined environment for transformation of software into services, integration, and service aggregation. It allows efficient integration of free software components into coherent services.
LSIP is the key technological component of Libre Services. It is the component that makes generalized, large-scale services development practical and efficient.
Libre Services are implemented as server-side software entities, typically running on servers at the service provider's premises, remote from the user. The user interacts with the service using his or her own computer, and may do so using any of a wide variety of user environments, such as EOE (Emacs Office Environment), a web browser, thin client, or KDE (K Desktop Environment). These are shown at the top level of Figure 1.2. The link between the user environment and the service is the Internet itself.
Since Libre Services are completely open, there are no proprietary restrictions on which user environment can be used to interact with the services. The Libre Services model allows any user environment to interact with any service using any hardware platform.
Libre Services bring the cumulative and collaborative development characteristics of free software into the services arena. They are open to completely unrestricted, large-scale collaborative development, and therefore have an ability to undergo complex evolutionary growth that cannot be matched by the proprietary model. In terms of richness of functionality, they have the ability to surpass the proprietary model completely.
In addition to straightforward end-user functionality, Libre Services also provide a number broader societal benefits.
In both the free and proprietary models, software is created by engineers. But the motives driving the engineering effort are entirely different under the two models.
In the proprietary model, engineering acts at the behest of business. And the prime directive of business is to make profit. Though corporations like to proclaim that their number one concern is the customer, this is a fiction. The welfare of the customer is of concern to the corporation only insofar as it relates to profit; beyond that it is meaningless. Under the proprietary model, service providers can and frequently do act directly against the interests of the user.
But in the free software model, engineering does not take place within a business framework. Instead it is a collaborative effort, undertaken by many organizations and individuals in a variety of diverse environments. Therefore the dependence of engineering on business imperatives is severed. The engineering effort no longer takes place at the behest of business. Instead it is driven by fundamental, constructive engineering motives: the desire of the software engineering community to create applications of real value to the user.
The resultant software is therefore fully aligned with the usage, requirements and interests of the user. It is built to benefit the user, not to benefit business.
In the proprietary services model, the service provider and the user are two separate and distinct entities. All policy decisions regarding operation of the service are made by the provider, with little input from the user.
In particular, the user is entirely subject to the provider's actions regarding civil liberties issues such as privacy, censorship, and freedom of speech. The provider's actions are taken based on commercial considerations, and these actions may constitute serious violations of the user's civil liberties.
As we have noted, commercial providers can silence dissent and enforce censorship in order to gain access to foreign markets. As we will note in the next section, they can also cooperate in government intrusion into their users' personal and private affairs. Commercial providers have also cooperated with enforcement of censorship and freedom of speech restrictions by monitoring web logs and bulletin boards, erasing banned content, and reporting offenders to government authorities.
Under the Libre Services model, however, any group or community of people can host the service cooperatively for themselves, and operate it according to whatever policies they see fit. The Libre Services model thus breaks the separation between the provider and the user--they can now be one and the same.
Libre Services can be operated by the user, for the user. The civil liberties of the user are thereby assured.
In the proprietary services model, user activity can be monitored without the user's knowledge or consent. There are two forms of monitoring that present societal concerns:
In the case of commercial monitoring, any aspect of a user's activities can be recorded and reviewed by the service provider. This includes the content of incoming and outgoing e-mail, search queries, websites visited, products and services purchased--indeed, any service usage that is technologically available to the provider can be monitored, without the knowledge or consent of the user.
This form of monitoring is much less of an issue in the case of Libre Services because, as we have noted, the service is designed to benefit the user, not a commercial entity. Since the service is not created for commercial benefit, there is no great incentive to include commercial monitoring capability within the service.
But to the extent that commercial monitoring remains a concern, the Libre Services model can provide complete guarantees of privacy. In the case of proprietary services, based on closed software, monitoring can take place because the community of users has no way of knowing what the software is actually doing. But in the case of Libre Services, the complete openness of the software permits verification and authentication that the service is completely free from all monitoring activity. The community of users is able to know exactly what the software is doing, and that it is doing no more and no less than they wish it to do.
Much more worrisome than commercial monitoring is monitoring by the government. National governments may have very broad powers to monitor their citizens' usage of Internet services. In the USA, an agency with sufficient authorization can compel a service provider to disclose all available information about a user, and cooperate in monitoring all communications and other service usage, without the user's knowledge or consent. The FBI's controversial Carnivore system, for example, is designed to capture all e-mail traffic for a particular targeted individual. Post-9/11, the necessary authorization can be provided simply by association, at several levels of remove, with someone the government considers to be a person of interest for national security reasons.
In the proprietary services model, covert government monitoring is possible because the user has no way of knowing what the service provider is doing. In particular, the provider is under no obligation to disclose government monitoring to the user. But in the case of Libre Services, any individual or organization can prevent covert monitoring by running the service for themselves, rather than leaving it in the hands of a third-party provider.
In addition, by eliminating the separation between the provider and the user, the Libre Services model makes current monitoring practices impractical. Under the proprietary model, a government agency conducts monitoring activity by directing its compliance demands against a well-defined commercial service provider. But under the Libre Services model the oligopoly of commercial service providers has disappeared, to be replaced by numerous private self-providers.
The government can still come knocking and demand access to a user's information. But it must now direct its compliance demands against a multiplicity of individual persons and organizations. And it can no longer do this without the user's knowledge.
In both the free and the proprietary worlds, software applications and services can be discontinued. The provider of the application or service can go out of business, or may decide to discontinue supporting the application. In either case the user may be left with an investment in an ``orphaned'' application. But the dynamics of how this occurs, and the effects on the user, are very different under the two models.
In the free software world, application extinction occurs because of migration of the community of users away from the application towards other, better applications. Extinction occurs because of a process of user-driven convergence, based on the genuine merits of alternative solutions.
In the proprietary world, applications are left orphaned not by the actions of the users, but by the actions of the provider. And these actions may be taken for reasons that have little to do with the actual merits of the application, but may be based on purely business considerations.
Because of these differing dynamics, application orphaning is a gradual and organic process in the free software world, whereas in the proprietary world it can occur suddenly and without warning.
Thus in the free software world, continuity of applications and services is much less of an issue than in the proprietary world. Applications persist based on their merits, and where they do not persist, this is to the ultimate betterment of the industry and the user.
But to the extent that service continuity is of concern to the user, the Libre Services model provides guarantees of continuity that are completely absent from the proprietary model. First, since the services are a communal resource, the user is not tied to any particular service provider. The effect of the Libre Services model is to decouple the service functionality from the service provider. If a provider goes out of business or discontinues providing the service, a user can simply go to an alternative provider, and be assured of receiving a functionally identical service.
The same consideration applies to the availability of technical support for the service. Again, since the service is a communal resource, the user is not tied to any particular vendor for technical support. Under the Libre Services model, technical support remains readily available for as long as the service itself remains available.
Finally, under the rather implausible scenario in which an entire Libre Service inexplicably disappears, but an organization remains fully committed to the orphaned service, the organization still has recourse. Since Libre Services are implemented entirely in free software, the organization has guaranteed perpetual access to the software. If necessary, the organization can reproduce and operate the entire service for itself.
As indicated in Figure 1.2, Internet services work by communication over the Internet between a client application running in the user environment, and a corresponding server application running within the service.
In the proprietary model, a particular service is tied to certain specific user environments. The service can be accessed only via one or two user environments, typically a web browser, and possibly also a dedicated client application provided by the service provider.
Under the Libre Services model, however, there are no proprietary limitations placed on integration between the User Environment layer and the Libre Services layer in Figure 1.2. Since the service is completely transparent, the dependence of the service on any particular user environment is severed. Thus any user environment can be integrated with any Libre Service.
Furthermore, a much more complex level of integration is possible. In particular, free user environments (i.e. user environments based on free software) can be integrated with Libre Services. And since both the client and server sides of the service are now completely transparent, this permits a highly complex level of integration between the two. This allows the development of Internet services with a power and versatility that far exceeds what exists today.
For these reasons we believe that the free software movement as we see it today is just the beginning. Today, free software exists at operating system and application level. The Libre Services model brings the power of free software to the Services level, the User Environment level, and the integration between these two levels. The result will be a complete transformation of the Internet services industry.
The Libre Services model also brings important benefits to the providers of Internet services.
Under the proprietary model the Internet services industry is dominated and controlled by a few large providers. These dominant players actively stifle competition by means of restrictive business practices, such as the use of proprietary protocols, highly aggressive patent assertion, and other practices based on ownership and control of intellectual assets.
Under the Libre Services model, however, there are no intellectual property barriers to business entry, and any company that wishes to host services can do so. This has major business consequences. The effect of the Libre Services model is to open the entire services industry to free market entry. This will result in unrestricted engineering collaboration and business competition, and will catalyze enormous industry growth. Libre Services will transform the closed industry of today into a truly open industry, in which all participants can compete on a level playing field.
In the proprietary model, small service providers can also be marginalized on the basis of service quality and functionality. A small proprietary service provider cannot compete with the resources of the large providers in their ability to develop new and better functionality.
Libre Services, however, are based on the large-scale cumulative and collaborative development mechanisms of free software. Any development contribution made by any engineer, anywhere, becomes immediately available to the entire constituency of service providers. In effect, the Libre Services model permits global pooling of engineering development resources.
This provides a level of cooperative development capability that far exceeds the resources of even the largest proprietary provider. Eventually the Libre Services model can surpass the proprietary model entirely in terms of service quality and functionality.
As we saw in the case of Instant Messaging, small proprietary providers can also be marginalized in terms of representation in industry decision-making, for example in establishing technical standards and protocols. This is because a service provider's voice in such decision-making is based ultimately on the provider's size, in terms of number of subscribers. The major commercial providers are thus able to exert a dominating influence over industry standards and policies.
The Libre Services movement, however, provides a unified voice of advocacy for all its subscribers. In effect the Libre Services model permits global pooling of the entire community of Libre Service subscribers as a single constituency for industry representation.
Libre Services are the right way to deliver Internet services to the world. As well as providing a number of vitally important societal benefits, they are also the proper basis for a healthy, thriving and egalitarian services industry.
Our goal is nothing less than the creation of an entirely new industry: the Libre Services industry. Much as others established the free software movement twenty years ago, we are establishing the Libre Services movement today.
Possibly the free software movement might have come into existence on its own, in some spontaneous organic way, even without the actions of the Free Software Foundation. We only get to experience one history, so we will never know. But at the very least the Free Software Foundation greatly expedited this process by explicitly formalizing the principles of free software. And quite possibly, without the Free Software Foundation or some other entity taking this initiative, the free software movement might never have come into existence as a coherent movement at all. In any event, 20 years later, the free software movement now exists as a viable alternative to proprietary software. Society is surely better off for having the choice.
Similar speculations can be made about the Libre Services movement. One could question whether there is any need for anything so grandiose as a ``movement,'' with a ``blueprint'' and a ``manifesto.'' Possibly it too might arise spontaneously, without requiring any explicit formalization. Possibly everything we are trying to achieve is destined to occur anyway, as a natural consequence of the already established free software movement. One could argue that services are just another form of software, and the existing free software movement is already sufficient to create its own non-proprietary presence in the services arena.
It is true that in a purely reductionist sense, services are just another form of software. But from a holistic viewpoint they are no more ``just'' an extension of software than biochemistry is ``just'' an extension of chemistry. Internet services have a richness and complexity beyond that of general software, and they have a set of dynamics that are not apparent within the general software arena.
The creation of a coherent service is a complex process, presenting its own set of technical challenges. This process involves the transformation of software components into a service-oriented implementation, integration of software components, and service aggregation.
The existing technical conventions of free software are not well adapted to these requirements. These requirements have been partially met by the appearance of hosting platforms, but these have been implemented on a specialized, ad hoc basis, and the integration of software into services remains inefficient and costly.
In the longer term a much more general framework is required. This framework must include a coherent set of tools, policies and conventions for efficient integration of free software components into services, and for consistent service aggregation. The Libre Services Integration Platform (LSIP) we have developed is one such framework.
In addition to these technical issues, managing collaborative development on services presents another set of challenges. For example, collaboration on services is much more vulnerable to software division (the tendency of open-source projects to split into rival projects) than collaboration on individual software components. For this and other reasons services require a set of collaborative policies and methodologies that are significantly different from the general free software model.
Furthermore, the motivations driving development of free software and Libre Services are quite different. The initial ``consumer'' of free software was the engineering community itself--the community of software engineers who recognized the need for non-proprietary software tools such as editors, compilers, debuggers, etc. Free software was developed by engineers for engineers, and so the benefits of free software translated directly into the necessary action to create it.
By contrast, the ``consumer'' of services is the end-user, and a completely different set of motivations and dynamics must come into play to create Libre Services. Though the benefits of Libre Services are very real and very far-reaching, it is not clear that these benefits translate directly into forces and action to create them. Under these circumstances an explicit movement is required to provide the necessary motivation.
In the end, the best anyone can say is that there is uncertainty about the future. One could be complacent, take no action, and accept whatever default history will hand us. But the stakes are very high. We believe it behooves us to take positive action to ensure the future, rather than complacently assume that history's default will be the one we desire.
The time to do this is now. All the necessary requisites for Libre Services now exist. The key enabling components are the Linux kernel, the GNU operating system, and the Debian distribution. These components are now sufficiently complete and mature to make Libre Services a reality.
The window of opportunity for this is unknown. It is possible that the proprietary services may become so entrenched that they become impossible to dislodge. The number of proprietary subscribers may become so large and so tightly bound to their service provider, that a different services model can no longer gain credibility. In this case, the opportunity to establish Libre Services, if not acted upon quickly, may be lost.
We would like to ensure that, 10 years from now, society will have a non-proprietary alternative to the proprietary Internet services. As in the case of free software, society will surely be better off for having the choice.
