english > FOSS

FOSS: Free / Open Source Software

What, Why & How

Prof. Pedro Antonio Dourado de Rezende
Dept. of Computer Science, University of Brasilia

ITI Brasil -- Bazilian Delegation to WSIS Prepcon2 -- Feb 13 2005

What is FOSS?

Source code

Revolutionizing many social and economic activities, the Internet did it most profoundly to businesses related with ICT intangibles (such as softwares, digital contents, etc.). In particular, it has yielded new business models and modes of production, distribution, and enterprising for software, as well as new ways to mix and match them.

New models for development and licensing, collectively known as Free/Open Source Software (FOSS), have emerged and shown their worth. By exploring decentralization, commoditization, convergences and cost decreases in ICTs through -- and as -- new means of value aggregation which Yale Law professor Yochai Benkler has named “commons-based peer production”1.

The crucial difference between these new models and modes of production, and those inherited from the pre-Internet era, is the way they treat source code. Knowing this difference is essential to grasping its deep consequences. Roughly speaking, source code is the human expression of an intellectual work in a form that can function as template for packaging and distributing products.

For instance, before typewriters, manuscripts would be source code for literary publications. And after high level programming languages, the version in which a program is originally written, by a human in such a language, is source code for that program. Whereas its automatic translation (compilation) to a given hardware platform's machine language is its object code for that type of platform.

Object code is actually what runs on a computer. When intended for distribution, a copy of a set of inter operating object codes, packaged with documentation on how it may work and a license to use, becomes a software product. Note that not all software is done for distribution, and not all licenses rely on the same contract law, copyright or intellectual property protection mechanisms.

Modes of production

In fact, most software development is not for distribution, but for private use by an enterprise. In this case, whether done by the enterprise or outsourced, software and production mode are called in-house. In-house softwares are mainly developed to supply specific semiological needs, not general enough to spur demand for distributable products (“distros”) from the software industry.

While harnessing the largest share of programming resources, the in-house mode of production yet attracts much less attention from scholars, investors and the press following ICTs' evolution. One can attribute this over-focusing on modes for distribution to the fact that “distros” aggregate value and generate business chains among distinct sectors of the economy. But there's more to it.

The importance of production modes for distributable software goes beyond economics, as UC Berkeley sociologist Manuel Castells shows2. It has to do with the nature of ICTs, the dynamics of its markets and their structuring roles. Roles not only at in-house development, but also in the evolution of social practices, including governance, and trade, mostly of other ICT intangibles.

Microelectronic industries and markets operate with cost structures very sensitive to scale. So they evolved under a drive for modularization, standardization, miniaturization and product streamlining which, by the early 80's, has reached an important threshold, the so called “downsize revolution”: the effects from the drive reaching other markets and industries, through interdependencies in business chains.

With downsizing, local networks connecting smaller, cheaper computers of growing capacity, became prevalent over mainframe data centers, of centralized processing and communication. In tandem, software engineering was growing complex, pushing systems into layered and modular architectures, with standards for interconnectivity and interoperability gaining importance.

Free, as in speech

The downsize revolution affected the economy of ICTs in important ways. It allowed software and hardware businesses to decouple, pushed by new risk assessments from a milestone antitrust lawsuit filed by the US government against IBM, for its predatory monopolistic practices of tyeing in hardware rents, software licenses and support contracts, then the prevalent, monolithic model.

To flourish independently, the business of producing and licensing distributable software had to develop its own models. Thus, a decade before the potential for inter network mesh connectivity matured into a rich, diverse, layered set of universal standards, reached by today's Internet with the inclusion of the http protocol in the 90s, a set of such models was picked by the market.

For lack of a better word, and to be contrasted with FOSS, this set has been called the proprietary model, for it hinges on the premise that software enterprises ought to treat source code as business secret: thus, a guarded property, such as the recipe for coca cola. The soundness seems obvious. To translate and copy code costs but nothing, licenses to use individual copies generate revenue.

The logic is superb, except it does not factor in the dynamics of ICT markets, driven by the semiological nature of its products. Soft + ware is not soft + drink. One then has to ask: what from this picture's missing dynamics could weaken the proprietary model to the point of someone considering alternatives? An answer is best grasped by examining alternatives of proved success.

We'll do it by contrasting FOSS and proprietary models and modes, and by listing case histories. But before, we end this session by contrasting their founding premises. As early said, this is key. So, what is FOSS? FOSS is an exercise of choice, to treat source code as language, a “privately owned” commons, a knowledge base open and free (as in speech) to the able and willing.


Why open source?

At this point, one may ask the obvious question: why open source code? If compilation into object code is automatic, and source code is made openly available, this doesn't jive with user licenses for individual copies as a source of revenue for a software enterprise. Free as in beer? some jump indignantly: How can programmers earn a living, or the boss pay salaries this way? How, if their intellectual work is to be treated as a commons, therefore available to competitors?

The answer lurks under the word “competitor”. How competition is transformed in and by cyberspace. Software is not soft drink. FOSS licenses do not place code in public domain, their authors still control the use of their work. But not by attempting to control access to individual copies of object code, installed in computers around the world. Rather, by controlling how its human readable expression, source code, can be legally used by others, through copyright law.

Instead of competition and piracy in the marketplace, FOSS licenses make a U-turn from the proprietary view, to aim control at cooperation in production modes. FOSS distinguishes right to access from right to reuse (to make derivative works from), and authorship from property of code. It does not treat software as soft drink, for the latter is a rivalrous good, and the former isn't. The question then becomes, how can this U-turn pay off, and to whom?

Before delving into case histories to answer, we'll step back and go by stages, for one's view of what's going on will get shuffled. Better keep an eye on the card deck. First, note: the opening question actually begs for its opposite, more revealing. Why was software's source code closed in the first place? For it only happened when its business decoupled from hardware's, by the 80s. Second, keep in mind: most software development is in-house. In these cases, the proprietary-FOSS dichotomy need not apply, since there is no software distribution, no general user license.

“In house”, how source code is to be treated becomes a private matter between programmer and who contracts him/her to write software, responding to the enterprise which will pay and use it. Since it's no one else's business, programmers may sign whatever in-house contracts they like, there being FOSS in cyberspace or not. But he/she will not program in a vacuum. Other softwares will play some structural role, particularly software fit for distribution. Just as ICT markets, with the role of its products and services in today's society. But could he/she distribute?

Why open standards?

What makes a software business wise fit for distribution is, basically, the scale of demand. Some projects may plan to build demand by marketing for new needs (demo versions), but most types of software have the demand scaled by the niche it can occupy in the “ecosystem” of ICT intangibles, as softwares become ever more modular, layered, complex, interdependent and bound by standards, besides ubiquitous.

From architectural layering, patterns emerge. The closer to hardware, the broader the potential demand for a given type of software. Basic ones with the broadest potential, such as operating systems (OS) and network stacks, on platforms going from servers to desktops, followed by middleware, such as databases, managing tools and virtual machines (for platform independence, distributed computing, etc), followed by applications.

In the application layer, demand patterns again diversify. From the pervasive office suites, email and browsing clients at the desktop, to more specialized types, such as desktop publishing, video and programming tools, with in-house apps at the higher end. And, as demand at lower layers stagnates to market growth and hardware upgrading, the strategic importance of standards grows.

So, a good reason for markets to have picked the proprietary model at an early stage of software business independence from hardware's may be its capacity to unfetteredly channel technical and semiological net effects into economic ones, towards monopolism. In a deregulated, emerging market it sure happened, as we saw the largest accumulation in the history of capitalism. Control of the process of setting up and licensing standards from a broad base became, then, key to power.

While a desktop broad base and standards process emerged from the OS market (with Windows), the Internet was being built on open and free standards, set up by meritocracy, collaboration and consensus. Softwares for Internet stacks were FOSS from the start, there was even a FOSS OS from late 70s, but which lacked key ingredients to breadth of market. So the monopolistic strategy became EEE: Embrace (open standards), Extend (to break interoperability) and Extinguish (competition through net effects), a bedrock for vendor lock-in.

Why collaborate?

This OS is BSD, a branch from Unix, the first OS fully multi platform. BSD is UC Berkeley's OS enterprise, branched from Unix's source code under license by its owner (AT&T), then a regulated telecom monopoly barred from doing business with software. One reason BSD may have failed to become a broad base for FOSS standards, led today by Linux, is its developing model (somewhat closed). Which brings us to why and how FOSS models work or don't.

Today's softwares are complex works, requiring many programmers to build. To successfully compete with proprietary equivalents, a FOSS project has to find a way to compensate for lack of revenue from user licenses. Savings from not having to manage compliance of every installation of a copy is not a cash flow. Although both regimes yield revenue from support and services, the streams from FOSS projects are open to any. The way is open collaboration, the key to success is motivation. Thousands of coders refuse to be passive victims of ever nastier vendor lock-ins (one only needs to read a proprietary End-User License Agreement, EULA, to realize what that is).

Basically, a FOSS project requires free license, open standards and a community of developers. But understanding how collaboration yields positive balances for software development, enterprising and consumers is a complex task3. Most projects which scaled up to global success, such as GNU, Linux, BSD, Apache and Sendmail (respectively web and mail server global market leaders), Mozilla, PHP, MySQL, KDE, Debian, etc. had each to develop ad-hoc strategies to face ominous challenges of different nature while going from small voluntary groups to global cooperatives with complex managing, paid workers and large corporate contributors. The economic bottom line is efficiency, only bad for entrenched monopolies.

To large corporate contributors or to small independent ICT shops, the payoff is measurable in demand for services and support, at a truly free market where competence is regulated by openness. IBM, for example, which maintains hundreds of full time programmers dedicated to maintaining the Linux kernel (fixing it and making it evolve), besides others to the Apache project, has ended 2004 with an estimated backlog in service orders of US$ 111 billion. The part collected from Linux-related services alone, twice the revenue in licenses from its own (and world's largest) ICT patent portfolio4.

For governments, the bottom line is political. Reasons for preferring FOSS to conduct its actions, ever so more dependent on ICT, include


How good?

The first obstacle to FOSS benefits is understanding what's at stake. Monopolistic stakeholders, and those in positions to benefit from them, get out to spread fear, uncertainty and doubt (FUD)about FOSS and about what may follow from its choices. Politicians have gone from denouncing it as child's play, a romantic experiment by lunatics, to the latest communist threat to capitalism. FOSS is ideological, since it's a choice for a more socially balanced evolution for ICT and the Information Society. But also a set of models for today's software market, of growing efficiency.

Meanwhile the proprietary model, the best the market could pick before the Internet, only lose efficiency as the Internet revolution unleashes, poised to double licensing costs in two years5 . Monopolies, as they age, tend to get abusive searching for the fountain of youth (have you read your latest EULA?). To sustain their positions, ITC monopolies need to create artificial scarcities of information and knowledge. For that, they promote a radicalization of Intellectual Property regimes, calling it “harmonization”, bound to asphyxiate new modes of commons-based peer production, such as FOSS12.

To defend themselves from IP radicalism, FOSS enterprises need to be overcautious and zealous with copyrights, to which its transparency is a boon. Meanwhile, as source code gets more abundant and programming tools better at code reuse, since people don't need to reinvent the wheel they can better concentrate on collaborative ideas for new business experiments. The more FOSS gets created, the lower the threshold for potential demand from which collaborators to successful projects can be drafted, and therefore, more at the application layer supplied.

How are we now? Judging from 2005 LinuxWorld's agenda, quite well. Keynotes include6

Brazil, for its part, has taken a lead among developing nations in affirming its right to choose FOSS, and its citizen's rights to express and access knowledge in an Information Society. In an article by journalist Terry Wade, Reuters opines that Brazil's president Lula “has reshaped the debate on intellectual property rights to reflect the needs of poor nations....Brazil is now at the forefront of what may be a global shift in how knowledge is produced and distributed. It has spurred a debate about what inventions should get patents, becoming intellectual property. Lula believes software, science and art should be governed by open-source laws7.

How in Brazil?

Law professor Eben Moglen, from Columbia University and council for the Free Software Foundation, declares: "Brazil is now the case study. It will play a major role in intellectual property talks and is going to provide an alternative example." So, how is that example? It comes from an ITC policy of choice for FOSS, with actions on migrations from proprietary systems and digital inclusion projects, each facing a different set of challenges and obstacles.

On migrations, besides public criticisms for spending to change what's working (lock-in costs) among the 1.2 thousand local media stories on the government's ITC policy, and internal problems from lack of trained staff (38%), tangled legacy lock-ins (33%), resistance from end users (19%), from technical staff (8%) and from upper management (4%) on 182 cases, the federal government had reached, in two years, 14.3 thousand desktops running on free OS, 46 systems migrated, 37 agencies with ongoing FOSS development projects, 54 FOSS systems under development, with U$ 10 million saved from non-renewal of proprietary EULAs8.

On digital inclusion, there have been several municipal and state projects already implemented, providing low income residents of large urban areas with Internet access, training and self-organized IT-related educational and cultural activities, through community centers known as “telecentros”, where computers run special GNU/Linux distributions with apps tailored for educational purposes. The federal government is about to launch its own, “Casa Brasil”, modeled after the one from the city of So Paulo.

The city of So Paulo has received the largest inclusion project run on FOSS in the world, serving roughly 1 million people in about 1 hundred telecentos. The city of Porto Alegre, where the World Social Forum was first held in 2001, this year with more than 150 thousand attendants from 136 countries, has another interesting experience with telecentros run on FOSS. They are managed by the community of local users, with the city government covering only the utilities.

Among other important projects, the auditing and hardening of a special GNU/Linux distribution for sensitive government use, led by Brazil's Research Center on Communication Security (Cepesc), a hardware cryptographic module for high security PKI installations running entirely in audited FOSS, developed jointly by the Brazilian Navy, National Institute of Information Technology and some Brazilian universities (projeto Joo de Barro), and the recent upgrade of the software for tax income declaration, to multi platform, running now also on GNU/Linux.

How else?

And where FOSS serves as paradigm, Brazil's minister of Culture, composer and musician Gilberto Gil, is active. He was one of the earliest world renowned artists to release a song under a Creative Commons (CC) license9. Any intellectual work expressible in digital form can have a CC license assign upon release, without costs or lawyers, from a menu where the author can chose the freedoms he allows to licensees and the rights he reserves to himself, regarding his licensed work.

In fact, Gilberto Gil arranged to pick and chose rights and freedoms for creating a new type of license, now in the CC menu as the “remix license”. Another famous personality to get involved was an important ITC executive, CEO of Sun microsystems Jonathan Schwartz, who recently wrote, in his personal blog and in Portuguese, a message of praise and support for Brazil's President Lula, for his ICT policy of choice for FOSS and his initiatives on digital inclusion run on FOSS10.

Digital inclusion projects with FOSS are particularly advantageous, for, besides there being no software licensing costs, there is also no penalty cost for migrating from a legacy system tied up in tangled locked-ins. On the other hand, being also where a desktop interface hatch into digital literacy for new markets, these FOSS projects tend to concentrate the fiercest of the oppositions, with the most lopsided arguments about market-friendliness imaginable.

Brazil is actively engaged in the international scene, cooperating with Portuguese speaking LDCs and participating in debates regarding Intellectual Property at international agencies, such as the World Intellectual Property Organization (WIPO). So far, two telecentros have been installed, in So Tome e Prncipe and Cabo Verde in Africa11, and a draft for a “development agenda” was jointly proposed with Argentina at WIPO, seeking to prevent radicalizations with Intellectual Property regimes from trampling upon human rights to express and access knowledge in our Information Society.

The social cost to preserve current trends, driven by monopolistic monoculturalism, is tangible. It can be felt in the noise and insecurity of cyberspace, against what a single-pronged legalistic approach based on radical strengthening of esoteric Intellectual Property constructs and enforcement may not be the wiser, and shall not be accepted as the only option, based on purely speculative assumptions. It seems appropriate for Brazil, and other friendly developing nations holding similar views, to act in collaboration to offer alternatives. Vowing to protect the future of FOSS is a good start.


1- Benkler, Y.: “Coase's Penguin The Yale Law Journal, Vol. 11. Yale University,

2- Castells, M.: "Innovacion, Libertad y Poder en la Era de la Informacion" ,
World Social Forum, Jan 2005,

3- Weber, S.: "The success of Open Source", Harvard University Press, 2004.

4- IBM 4Q04 Quartely Earnings. 4q04/4q04earnings.phtml

5- Thomson, I.: Software license fess to double in two years: Gartner warns of predatory pricing news/1159603

6- Stafford, J: “Open source rules, SCO fades and apps abound”, 09 Feb 2005 originalContent/0,289142,sid39_gci1052513,00.html

7- Wade, T.: “Brazil reshapes debate on Intellectual Property” Reuters, 02 Feb 2005 ?type=ourWorldNews&storyID=7512457

8- Internal Report, Comit Tcnico de Implementao do Software Livre

9- Creative Commons Project.

10- Schwartz, J.: “Caro presidente Lula”, 08 Feb 2005, 20050208#caro_presidente_lula

11- Cata Preta, A.: “Andre Stefani: Consultoria GNU/Linux” ?acao=noticias&wnoLink=./noticias/not0053.php

12- Barr, J.: "HP memo forecasts MS patent attacks on free software" ?tid=110&tid=147&tid=132