Ryan Tool Econ 191A/B - Crawford 1 Economics of Open Source
Transcription
Ryan Tool Econ 191A/B - Crawford 1 Economics of Open Source
Ryan Tool Econ 191A/B - Crawford Economics of Open Source: The Motives and Methods Behind the Free Software that is Challenging Market Leaders By Ryan Tool June 3, 2005 1 Ryan Tool Econ 191A/B - Crawford Introduction It is difficult to open a newspaper today without finding an article on open source software. The success of Linux and the recent proliferation of Mozilla have vaulted open source software out of its geeky niche and into the mainstream. Many leading applications now have competing open source equivalents. Microsoft’s Internet Information Server competes with Apache in the webserver market. Internet Explorer contends with Mozilla, which has decreased Internet Explorer’s market share lead in recent months. Adobe’s Photoshop bred GIMP, which isn’t as fully featured but handles most photo editing tasks for a price of $0 compared to Photoshop’s $650 price tag. Open source software (OSS) is starting to be known as a viable alternative to traditional commercial software. OSS is being used in almost all fields. You have probably used it yourself without even knowing it. If you are a student at UCSD and you use campus email, then you have most likely used UCSD’s web email application, SquirrelMail, which is an open source project. Most large scale commercial applications cost thousands of dollars, but SquirrelMail is a program that is used at numerous universities and businesses in the United States and is available for anybody to use absolutely free. Anyone can download the SquirrelMail software, use it at home, and enjoy the same robust, technically sound application that is being used by large corporations. The fact that this level of application is being built and distributed completely free demands investigation. 2 Ryan Tool Econ 191A/B - Crawford How can software developed from donated time compete with commercial products developed by teams of salaried employees motivated by profits? What makes open source software different from closed software? Is one type of software preferred to another? Does closed software stifle innovation? This paper aims to answer these questions and examine the economic theories behind the success of the open source movement. We will use consumer theory to analyze the motives behind open source project participation. We will develop a utility model for open source programmers and compare this model with a typical individual to ascertain the differences between the two. Using this model we will gain insight into the motives behind open source development, both altruistic and selfish. We will examine the process of creating an application from inception to distribution in both the open source distributed framework and the closed clustered framework. Then these two methods of product creation will be analyzed using current microeconomic models to determine the welfare gained or lost to society as a whole by the different production technologies. We will look at some real world examples of open source projects and compare them to their commercial counterparts to determine what makes open source applications successful. Finally we will use these models to analyze current software trends and make some predictions about the future of software development. This paper aspires to explain the current surge of open source software through economic analysis and to provide insight to the recent success of the open source movement. While working toward these goals, this paper will 3 Ryan Tool Econ 191A/B - Crawford answer the how’s and why’s of open source software through standard economic models. Part I - Background My introduction to open source software occurred in the mid-nineties when Linux was starting to gain momentum. The idea of OSS was intriguing, a free operating system developed by a group of disjoint programmers simply because they wanted a robust UNIX-like operating system designed to their specifications for their needs. Operating systems are composed of hundreds of thousands of lines of programming code; they are huge, complex programs. Designing an operating system from scratch is like deciding to build your own car from scratch because you cannot find one on the market that suits your needs. Furthermore, if one goes through all the trouble to design and build something so complex and time consuming, you wouldn’t expect them to give it away. On basic intuition, “Thousands of volunteers are unlikely to come together and collaborate on a complex project they give away for free, which can beat some of the largest, riches businesses at their own game.” (Weber) The idea of a business not motivated by profits is a foreign concept to most businessmen. A commercial firm measures success by their profits and growth. Open source firms are typically not motivated by profits and are measured by a different set of standards. Most OSS firms are non-profits, existing primarily for legal and organizational reasons. If they have any financial backing at all it is from donations, not investment capital. Open source projects 4 Ryan Tool Econ 191A/B - Crawford are typically run on a shoestring budget with most of the development hours being donated. These firms exist not to make money, but to make software. Their user base, not their bottom line, is the measure of their success. Many debates concerning open source software stem from the idea of intellectual property. On one side of the fence you have views that, “patents and other forms of intellectual property law are the bane of the free universe, […] actually evil and should be struck down as soon as possible” (Torvalds). In contrast, the “advocates of intellectual property believe, ‘the whole world economy is driven by intellectual property’” (Torvalds). This paper will examine the idea of intellectual property through the minds of the open source developers to find the differences in ideology that drives them to offer their software creations free of charge to the public for use, distribution, and modification. We will use these ideas about intellectual property to understand the motives and methods behind open source software licenses. Part II – Defining Open Source Open source is a broad term that can be used to describe many types of projects. It would be very difficult to debate the merits of open source without a clear definition of OSS. In this paper we will use the Open Source Initiative’s (OSI) definition of open source, which stipulates that open source software must comply with the following criteria: 1) Free Distribution - By constraining the license to require free redistribution, we eliminate the temptation to throw away many long- 5 Ryan Tool Econ 191A/B - Crawford term gains in order to make a few short-term sales dollars. If we didn't do this, there would be lots of pressure for cooperators to defect. 2) Source Code – The program must include source code. We require access to un-obfuscated source code because you can't evolve programs without modifying them. Since our purpose is to make evolution easy, we require that modification be made easy. 3) Derived Works - The license must allow modifications and derived works, and must allow them to be distributed under the same terms as the license of the original software. 4) Integrity of Author’s Source Code - Encouraging lots of improvement is a good thing, but users have a right to know who is responsible for the software they are using. Authors and maintainers have reciprocal right to know what they're being asked to support and protect their reputations. 5) No Discrimination Against Persons or Groups - The license must not discriminate against any person or group of persons. 6) No Discrimination Against Fields of Endeavor - The major intention of this clause is to prohibit license traps that prevent open source from being used commercially. We want commercial users to join our community, not feel excluded from it. 7) Distribution of License - The rights attached to the program must apply to all to whom the program is redistributed without the need for execution of an additional license by those parties. 6 Ryan Tool Econ 191A/B - Crawford 8) License Must Not Be Specific to a Product - The rights attached to the program must not depend on the program's being part of a particular software distribution. 9) License Must Not Restrict Other Software – The license must not place restrictions on other software that is distributed along with the licensed software. For example, the license must not insist that all other programs distributed on the same medium must be open-source software. 10) License Must be Technology Neutral – No provision of the license may be predicated on any individual technology or style of interface. Open Source Initiative (OSI) is a non-profit corporation dedicated to managing and promoting the open source definition. We have chosen their definition because it is the de facto standard for many open source projects. There are a number of open source copyright licenses in use today. “The fundamental purpose of open source licensing is to deny anybody the right to exclusively exploit a work.” (St. Laurent) The OSI definition encompasses many of the popular licenses for open source projects, such as the GPL, BSD, and Apache licenses. The main difference of these licenses is whether or not they protect derived works. Some licenses allow derivations based on the code to be sold while other licenses protect the open source code base requiring all derived works use the same license as the original open source application. Within these defined guidelines, why would a company make a product which is free by definition? Is it possible to make money off of a product you give 7 Ryan Tool Econ 191A/B - Crawford away? Yes. There are a few different business models that have been successful for open source companies. The most common business model is based on support. In the support-based model companies give their OSS away for free to anyone, and then they make money by selling support services, similar to a warranty from a car dealership (if the car were free). These support services range from teaching users how to use the software to configuring the most complex servers. Mandrake is great example of a support-based OSS company. They offer their operating system online for download absolutely free of charge, but they also offer packages for sale. These packages range in price and can include manuals, CDs, DVDs, phone support, web support, membership to online user forums, additional integrated software, and many other benefits. Another company that uses this model is SUSE, a subsidiary of Novell. They offer their Linux operating system free of charge while selling packages with additional benefits similar to those offered by Mandrake. In fact, most OSS operating systems use this model. As a side note, both of these companies’ operating systems are versions of the Linux operating system. The kernel (analogous to a human brain or a building blueprint) for Linux is another open source project. This Linux kernel serves as the base or core set of code for all of these different versions of the Linux operating system that are being sold commercially. The different versions of Linux are great examples of the forking principle (described below). This is also an example of one open source project being incorporated into another 8 Ryan Tool Econ 191A/B - Crawford open source project. This type of behavior is allowed and even encouraged under the OSS license. Another popular business model for OSS is based on offering free consumer versions and charging for commercial versions. The idea behind this model allows individuals to use the software for free for personal use, but requires businesses to pay for slightly different versions of the same software. Redhat is a good example of this type of business model. They offer a free desktop version of their operating system, known as the Fedora project, which is offered without support, and they sell a more robust, fully supported version as an enterprise (corporate) operating system. This allows them to use the consumer version as a testing ground for new features. It also allows individuals to try their product before dishing out money for the fully supported commercial versions. This model is very similar to the model of offering scaled-down or limited versions for free and charging for more fully featured versions. An interesting byproduct of open source software development is forking, where the software splits into two different pieces of software each based off the original. Forking happens when two key developers have opposing ideas concerning the direction of the project. This is different from a commercial software firm, where a decision would be made and one idea would make it into the product while the other idea was dropped. In an OSS project, if the inner circle of developers cannot come to an agreement, they split up and the project diverges into two new projects. One project typically retains the original name and idea and the other project mutates into something new. This may seem 9 Ryan Tool Econ 191A/B - Crawford counterproductive, but it is actually quite the opposite. These two similar pieces of software are now released into the wild to compete for users. While a corporation may try to manipulate consumers into using a specific product, open source developers will let users decide which product will be adopted. If users adopt both products then the two open source projects from the fork will continue to flourish. However, if users clearly favor one product over the other, typically that project will flourish and development efforts will recombine around the stronger project. A great example of forking can be seen in the various Linux distributions. There are many different varieties or flavors of Linux, each with their own special attributes. One group of developers thought security should be their main concern, while another group of developer thought ease of use and installations should be the key feature. Both of these Linux variants have been successful, establishing very different user bases. The security focused Linux is typically used for web servers and the user-friendly focused Linux is typically used for desktops. Without forking, the core Linux project would have chosen one of these paths and the other one would not exist or a compromise would have been made and the product would not excel at either task. Similarly, when useless features are forked into new software projects, those projects are rejected in the competitive market and developers are able to quickly determine what features are necessary or desired. It is easy to understand why some of these open source projects get started, but difficult to explain why they don’t turn commercial once they’ve 10 Ryan Tool Econ 191A/B - Crawford become popular or why they continue to flourish against competing commercial products. By understanding the motives behind the developers and the open source firms we might be able determine why some of these software products are so successful and why the number of open source projects continues to grow. Part III – Model of an Open Source Software Developer What makes an individual want to contribute their time and skills to an open source project? Weber finds the determinants of OSS contribution to be: art and beauty, job as vocation, joint enemy, ego boosting, reputation, and identity and belief systems. There have been a handful of studies to ascertain the motives of open source contributors (Lakhani and v. Hippel, Haruvy and Wu, Lerner and Tirole). While the leading determinant of contribution differs among theses studies, the set of determinants is similar throughout. We find the common reasons for a developer to donate time to an open source project are: altruism, future earning potential, future employment, personal enjoyment, hobby, recognition, politics, and cult status. These motives are similar to findings in previous research and are explained in more detail. One reason to contribute to an open source project is the obvious avenue to personal gain, like future employment or future earning potential. In this regard, working on an OSS project is like the unpaid internship you take in college; you put in the time without pay to make connections and build your resume for future gains. You gain valuable experience and learn to solve 11 Ryan Tool Econ 191A/B - Crawford complex problems you might not find in a classroom-learning environment. This could explain the reason students or recent college graduates contribute their time, but it does not explain why veteran programmers work on these projects. With this thought in mind one might assume only rookie programmers work on OSS projects, but this is clearly not the case. The contributions from experienced programmers must be explained by other reasons. They could contribute because they love to program and working on an OSS project is a hobby for them. Solving a complex coding problem is like scratching their person itch. Some individuals like to solve crosswords or puzzles in their spare time, while computer-savvy programmers gain the same enjoyment from solving problems in computer code. They might contribute because they have political motivations that are aligned with the open source project. One such political motivation could be a dislike for large corporations, which would impel them to support a smaller, grassroots project like some open source applications. Another political motivation could be their views on intellectual property. As creators of information (their programs) they might feel that information should be free for everyone, and with such a disposition they would be inclined to develop and promote open source projects. In this respect they act much like academic researchers, publishing their work for anyone to digest and build upon in hopes of solving a bigger problem or developing better software. Altruism is another reason to donate time to an open source project. Developing OSS is the computer geek equivalent of giving blood; it is a selfless 12 Ryan Tool Econ 191A/B - Crawford act that is done solely for the benefit of the computing community as a whole. Open source software is a true public good: it is non-rival and non-excludable. (Johnson) Almost all software is non-rival in that it is not consumed during use, so one individual using it does not stop another individual from using it. It is nonexcludable by design as non-excludability is part of the OSI definition of open source software and built into the licensing of all open source projects. Any time spent contributing to an open source project could be considered altruistic because open source software is a public good by definition. Developers could contribute to open source projects for less altruistic motives, such as self-esteem or prestige. The title of open source developer can be an important one. Being a developer on one of the larger OSS projects is very impressive within the development community. This causes some open source projects to have a sort of cult status. Key positions (the decision making positions at the top of the hierarchy) in important OSS projects are scarce, causing them to become more desirable. Being a key developer on a successful open source project is like going to an Ivy league school, there is a certain honor or appreciation associated with the experience. Using the aforementioned reasons as the basis for a developer to contribute to an OSS project, we can start to develop a utility model for an open source developer, but first it is important to distinguish the differences between a computer programmer and the average individual. The average salary for developers is significantly higher than the national average, so developers are able to afford ‘average’ consumption with money to spare. The annual average 13 Ryan Tool Econ 191A/B - Crawford salary for a software engineer is $78,025 while the national average income is only $36,520. (US Bureau of Labor Statistics) The fact that developers have higher salaries implies they have more income to spend on additional goods or luxury goods. Contributing to an open source project can be thought of as another good, which developers are able to afford due to their healthy salaries. With higher salaries these developers may be able to work less and still afford the same levels of consumption as the average individual. In other words, these developers are able to work fewer hours to supply a necessary level of income to satisfy their needed utility through consumption and spend the extra time working on OSS projects in pursuit of additional utility.1 They may choose to spend less of their time working for income and spend the time saved seeking utility from other sources. Interpretation of these ideas leads to the hypothesis that developers may have excess time or excess income that can be used on other utility generating endeavors, like open source projects. Computer programmers tend to exhibit different personalities that attract them to the career of software development. These programmers are predominately male, introverted, and comfortable with a one on one relationship with their computer. Furthermore, participation in an open source project is like a cult. These developers may form a subgroup of the typical computer At a salary of $40,000 per year with a 40-hour workweek, I cannot afford to work any less as my necessary utility requires about $40,000 for consumption. However, at a salary of $80,000 per year with a 40-hour workweek, I could choose to work 30 hours a week for a salary of $60,000, which would yield more than enough income to obtain my necessary utility through consumption and leave me with extra income and time to be used for other utility generating activities. 1 14 Ryan Tool Econ 191A/B - Crawford programmer. Many OSS developers pride themselves on the projects to which they contribute. Within computing circles, being a member on a successful open source project is like being a movie star. It’s important to understand the way these developers think to understand their attractions to open source and model their behavior within the common models for consumer utility. To model utility of open source developers we use current consumer theory. At the core of consumer theory we have the principle that consumers maximize their utility or happiness. This utility function is subject to a budget constrain, typically income. Another way to think of the budget constraint is to use time as the constraint instead of income. A consumer can exchange all their time for income, which would be just like leaving the budget constraint in terms of income and spending that income to generate utility, or they can decide to spend a portion of their time for money and a portion of their time for added utility. Open source developers are different from typical consumers; they are able to achieve maximum consumption utility without exhausting their income. In terms of the time budget constraint, this means OSS developers will not exchange all their time for money. They will instead maximize the amount of utility they can derive from money and consumption based utility, and then they will use the remainder of their time chasing additional utility. This additional utility can be obtained from open source projects. The list of previously mentioned gains from open source contribution only strengthens the idea that developers would want to contribute some of their time to such a project. In terms of the strictly monetary budget constraint, open source developers have a much higher 15 Ryan Tool Econ 191A/B - Crawford budget constraint than the average consumer. This could allow them to afford additional goods or luxury goods. Within this framework, open source projects could be considered an additional good or luxury good. In either case a developer would gain utility by spending a portion of their excess income on open source projects. So far we have only considered the supply side of open source to determine the motivations behind the developers, but we can also look at the demand for open source software. (Prasad) With the success and widespread use of some open source projects it is easy to see there is a strong demand for the software. From a developer’s perspective there may be additional demand. Many open source developers work as software developers for commercial corporations and many of these corporations use open source software for their business needs. For example, 70% of all webservers use Apache, an open source webserver application. (Netcraft) The chance that an OSS contributor also works on open source code at his day job is fairly high. For example, Yahoo! uses FreeBSD for its servers, so they must employ a number of developers and engineers to maintain these servers. These developers are paid to support and maintain the company servers which run FreeBSD. They can look to the FreeBSD community for help with difficult implementations and they typically give back to the community when they make improvements to the software. There are a number of companies using open source projects for their business needs which creates a demand for open source software. This corporate demand for OSS translates to developers being 16 Ryan Tool Econ 191A/B - Crawford paid to develop and contribute to open source projects. If the added utility from developing OSS is not enough motivation to contribute to a project, there are a number of developers that earn a living by using and supporting open source projects, which benefits the open source community as a whole. Some companies promote open source efforts by offering prizes for specific open source solutions to computing problems they may have. This allows the company to obtain a product it needs for a fraction of the cost it might take to develop the product in house all while paying developers to create a product that will be given back to the community as soon as it is completed. Computer Associates (CA) recently employed this method with great success. CA wanted a product that would migrate existing databases to their database format, so they offered a $1 million challenge to the open source community, “The Challenge was designed to leverage the global talent pool of the open source community, while providing substantial financial compensation.” (Computer Associates) The Challenge was a huge success, yielding a couple open source projects that successfully migrate data to CA’s database format. From these examples we can see there is demand for open source applications, which will lead to more open source applications being developed. 17 Ryan Tool Econ 191A/B - Crawford Part IV – Model of an Open Source Firm There are over 100,000 registered open source projects and over 1,000,000 registered users at sourceforge.net, the world’s largest open source software development website which provides free services for open source projects. (Sourceforge) The majority of the projects listed on sourceforge.net are very small consisting of only a few developers, but some of the larger projects have hundreds or even thousands of contributing developers. For this paper we will only consider open source firms with more than 10 developers. Open source firm size is determined by popularity or usefulness of the project. A project with a large user base, such as Apache, will generate a much larger developer base and be larger in size. Widely used projects are more prestigious for developers. Predictably, developers will choose not to contribute to a project they view is headed towards a dead end. Developers want to be members on successful and useful projects, so firm size will be correlated with the usefulness and popularity of the project. An open source firm is not much of a firm in the traditional sense. An OSS firm is typically a collection of developers that come together, intellectually not physically, to create a specific piece of software. The firms are not seeking profits; they exist only to promote the growth of the software. The goals of the firms are just a reflection of the goals of the developers. Every developer has a stake in the firm because their name is associated with the project, so the firm’s motivation is determined by the individual motivations of the developers. 18 Ryan Tool Econ 191A/B - Crawford The bottom line of an open source firm is not found on an income statement. The general goal of these firms is to produce a viable product. The specific goals of these firms can vary greatly from creating a super secure operating system to designing an easy to use application. Open source firms are not driven by profits, so they measure success in different ways. For some projects, creating a working piece of software is success. Larger or more ambitious firms measure success by market share or user base. Market share is measured as the proportion of industry use within a software category such as web browsers or operating systems. User base is the total number of users currently using the software. Some of the larger OSS projects measure success by gains in market share while other projects measure success by the size of their user base. For example, Mozilla’s Firefox reached a milestone last month when it surpassed 50 million downloads. (Mozilla) The development model for open source projects is very different from that of their commercial equivalents. Commercial software is developed in companies with a team of programmers all working for a single manager. This team works together to develop software to the specifications of the project manager who makes all the top-level decisions. On a large project there can be layers of managers and teams, but it almost always comes down to one leader making the key decisions and delegating the programming tasks to teams of developers. These developers work on a project from 6 months up to 3 years before releasing it. After the product is released the development team fixes 19 Ryan Tool Econ 191A/B - Crawford bugs and solves problems as users find them, so major updates occur at about the same frequency as the initial release. In an open source project hundreds or possibly thousands of users develop parts of the code and submit it for review. Updates are made quickly and new versions of the software are released very quickly, sometimes daily. This causes the software to evolve very quickly. This rapid evolution is one of the key advantages open source has over closed commercial software. (Raymond) Problems with OSS are usually discovered and corrected much more quickly because the user base that is testing the product is larger and more diverse (in knowledge and hardware configuration) than the product test group at a commercial firm. Commercial products are tested in house to work out bugs and then they are released. Bugs that are not caught in the initial testing phase are reported by users and corrected by developers for subsequent releases. The users that report the bugs are your typical computer users, so they often cannot explain in detail the true nature of the problem or bug. On the other hand, open source projects are tested in the wild; you might say they are released prematurely and full of bugs. When a new version or update is completed it is immediately available for download and use by thousands of users. The number of users to quickly adopt the new software is advantageous because as Linus Torvalds once said, “with enough eyeballs all bugs are shallow.” (Raymond) The developers of OSS are typically the first batch of users that test the software in their own environments, thus they are the 20 Ryan Tool Econ 191A/B - Crawford ones reporting bugs to the development team, of which they are a part. Since developers are reporting the bugs, they are able to explicitly describe the problem, which leads to a faster solution by the other developers. With a larger user base, open source projects can be tested across many different hardware and software configurations which allows the software to evolve quickly. Another advantage to the rapid release model is that bugs are fixed quickly; one developer sometimes fixes bugs before another discovers them. This is different from the commercial method where a bug might be discovered and fixed, but the software is not updated until the next commercial release. Open source firms are organized differently than their commercial equivalents. Developers at open source firms are not liable to investors; they are liable only to each other. The hierarchy of open source firms is generally streamlined for efficiency and productivity; it is not convoluted with board members, C-level positions, or any of the other structural hierarchies found in commercial firms. The structure of larger open source firms has evolved to include necessary leadership yet still be efficient. We will examine three successful open source projects and the structural models they use to define their hierarchy. The first model is a pyramid hierarchy. The Linux kernel is an example of an open source project that is organized by the pyramid hierarchy. In the pyramid hierarchy model, there is usually one sole creator who makes the final decisions and sits atop the pyramid. Directly below him there may be a few powerful developers who manage large portions of the source code and help 21 Ryan Tool Econ 191A/B - Crawford influence the overall direction of the project. Below them there will be a larger group of individuals who are each responsible for specific systems within the software. The next group would be developers that contribute little snippets of code that make up parts of the different systems within the software. This model is similar to commercial firms with a top down management style. The next model is a concentric circles model. FreeBSD uses the concentric circles model. In this model there is a core group that makes important decisions regarding the direction of the software and major changes or updates. Below this group lies the next circle of committers who are responsible for small changes and corrections to the software. The final circle is composed of developers who submit code to the project but are not allowed to make any changes directly to the software. (FreeBSD) Another model relies on a meritocracy, or govern of merit. Apache uses this model, electing members only after they have shown excellent and sustained contributions to the project. The meritocracy has different roles building in importance: user, developer, committer, PMC (Project Management Committee) member, and finally ASF (Apache Software Foundation) member. (Apache) Being an Apache Software Foundation Member is a very prestigious position. These are just a few examples of the various hierarchies that have come about through the creation of open source software. It is important to mention that each of these three products was developed by members from very different geographies, so these decision-making hierarchies not only successfully manage 22 Ryan Tool Econ 191A/B - Crawford very complex software projects, they also span countries to allow management of a very distributed group. Some open source projects compete directly with commercial firms, but do they face the same costs as a commercial firm? How do open source firms deal with the costs of creating and distributing a product? To evaluate how OSS firms cut costs, it is important to understand the costs that a commercial software firm faces. A typical commercial software firm is motivated by profits. Profits are defined as revenues minus costs. After the code has been written, the cost of producing and distributing software is very low. CDs or DVDs can be made for pennies and packaging costs little more. When developing a commercial software product the largest portion of costs are commonly development costs and marketing costs. These development costs are used to pay programmers salaries and cover general support costs. Support costs can be thought of as all the general business costs that are not developer salaries, like rent, supplies, and management salaries. The very first copy of a new software program is very expensive to make, but once it is made, replication is easy. With an automobile, the cost of physical inputs is a large and absorbs a significant portion of the price, but with software the only physical input is the media the software is distributed on. Due to the cheap cost of media, the marginal cost of producing software is practically zero. The average cost of software is constantly decreasing as more software is produced, which makes this a decreasing cost industry. 23 Ryan Tool Econ 191A/B - Crawford When a commercial firm sets the price of its new software product, it tries to set the price in such a way that the revenues from sales will cover their total costs. With an OSS project there is no revenue, so unless someone is willing to “lose” money, there must also be no costs. How are OSS projects run with zero costs? Open source software projects are capable of operating with little or no costs. This is possible by a number of different reasons. This first and most obvious is due to developers donating their time. Developers have higher than normal average salaries, so paying a team of developers can get expensive very quickly. Even the most trivial computer programs take hours to write, test, and then fix, so having volunteer developers is crucial to the success of an open source project. Another big reason open source projects can escape costs is the growth of a recent invention known as the internet. The internet is one of the most influential contributors to the advancement of open source development. The internet is the single biggest factor that allows the decision-making hierarchies of open source projects to function across varied geographies. The internet has allowed some employees to telecommute, or work from home over their phone lines via computer connection to their office. Students and contributors to open source projects have also discovered the benefits of telecommuting. Instead of connecting with a corporate office from home, OSS developers are capable of connecting with each other over the internet. These developers no longer need an office or boardroom to hold meetings or make decisions. This lack of 24 Ryan Tool Econ 191A/B - Crawford confinement to a physical office allows users from very different geographical areas to work together on a single project. This produces two benefits: the overhead cost of rent is reduced to zero, and transportation costs of human capital are also zero. Furthermore, the ability for developers to work whenever and wherever they choose may increase their actual output when compared to commercial developers being confined to spend eight hours each day in a cubical. Another cost a typical commercial firm faces is the cost of equipment. Current computers are needed to run current software, so a software programmer must use, or at least have access to current computers. Commercial firms provide their employees with an office and a computer to work on. This is not the case for an open source firm. OSS firms rely on developers working from home on their own equipment. Software developers are typically tech savvy people, so we can assume they will have a computer at home. Furthermore, having your own computer is a necessary condition for being an open source developer, so the cost of capital that a commercial software firm faces is again reduced to zero for an open source firm. A final cost that should be considered is the cost of distribution. CDs and packaging may be small parts of software costs, but they are still costs. The recent proliferation of the internet allows open source firms to offer their software for free via download over the internet, thus avoiding the common costs of distribution. Any user can download the current version of the software as a CD image file and burn that image onto a CD, or install the software directly from the 25 Ryan Tool Econ 191A/B - Crawford internet without using any media. The cost of distributing software for an open source firm is only internet bandwidth, and most open source firms don’t even pay for that. Open source projects can be hosted for free at sites like sourceforge.net, which pays for its bandwidth by donations. Also, universities and corporations sponsor many of the larger open source projects that have many users and require large amounts of bandwidth. These companies provide download mirrors so the open source files are actually downloaded from the universities and corporations themselves, thus using the universities’ “donated” bandwidth. Hosting a mirror for an open source project is an easy way for a nontechnical company to support that project. It’s also a great way to show gratitude and promote the growth of the open source project. The motivations of open source firms are aligned with the focus of each specific project and with the developers that contribute to the project. Open source firms are able to operate with minimal costs or no costs with a volunteer labor force, new methods of product delivery, and innovations in management and decision-making hierarchies. Many of these cost cutting measures are available due to technological advances from the internet. The structure of the development model allows the software to evolve rapidly by finding and fixing bugs quickly. The benefits of open source software are being realized across many industries; as the usage of open source software grows, the number and size of open source firms will continue to grow. 26 Ryan Tool Econ 191A/B - Crawford Part V – Current Trends and Future Predictions Now we will compare and contrast a few of the prevailing open source projects with their commercial counterparts to determine the causes of their success and make predictions concerning the success of future open source projects. We will examine the following software categories: internet browsers, desktop operating systems, server operating systems, email servers, and webservers. These software categories can be divided into to subcategories: used by consumers and used by firms. A desktop operating system, Windows for example, is the program which all home users have installed on their computers. It can be thought of as the brain of computer software with all other applications being the appendages. One of those appendages is an internet browser which is used to surf the web, or browse the internet. Desktop operating systems and internet browsers are used by both consumers and firms although they are typically thought of as consumer products. Server operating systems are used almost exclusively by firms. They are the software brain which controls the expensive, complex hardware that firms use for their operations. The additional software or appendages of these server brains are services that are needed in today’s business world, like email and a website. A webserver is the software that publishes a company’s website to rest of the world. An email server is the software the manages a company’s email. It can be thought of as the software equivalent of a brick and mortar post office. The webserver and email server software is installed along with server operating 27 Ryan Tool Econ 191A/B - Crawford systems, so these three products are used by almost all firms in today’s technology-influenced business world. Internet browsers are a good place to start for our comparison. The ability to browse the internet is a key component of the majority of the computers in use today. Most all new personal computers sold in the United States have internet browsers installed on them. The current market leader is Microsoft’s Internet Explorer (IE). IE is packaged with every version of Microsoft Windows, so it is the default browser for all Windows users. If Windows users want to use a different browser they must download and install it themselves. A popular browser to switch to is Firefox. Firefox grew out of the code base from another browser which was released to the community as open source in hopes that the code would be embraced and improved. Firefox is the default browser on most Linux distributions. IE is the dominant market leader but Firefox has been gaining popularity in recent months, generating a flood of publicity. “Internet Explorer is seeing its commanding market share erode a bit, thanks to inroads from Firefox.” (Sarrel) US Browser Usage Share Browser 6/4/04 12/3/04 2/18/05 4/29/05 Internet Explorer 95.48% 91.80% 89.85% 88.86% Firefox 4.06% 5.69% *3.53% Non-Firefox Netscape and 2.83% 2.47% Mozilla browsers Other 0.95% 1.25% 1.90% th *June 4 figure includes all Netscape and Mozilla bases browsers (source: WebSideStory) 6.75% 28 2.23% 2.06% Ryan Tool Econ 191A/B - Crawford While Firefox is a long way from dethroning IE, its current momentum places it in a position to challenge the Microsoft in the near future. Because Firefox is the default browser on most Linux distributions, its user base will also grow as the Linux user base grows, which brings us to our next software category, desktop operating systems. Windows XP, the current desktop version of Windows, is the most popular desktop operating system in use today. Windows currently controls the majority of consumers desktops with a market share of 94%. (Roush) It is preinstalled on almost all new computers sold in the United States. At the time of this writing it retails for $299. (Microsoft) The underdog in this battle for the consumer’s desktop is Linux. For many computer users, Linux is typically the first thing that comes to mind when someone mentions the words, “open source.” There are different versions of Linux that have been developed for desktop computers and servers. Linux is the software that brought the idea of open source to the masses; it is free, versatile, and secure. Linux is the second most popular desktop operating system behind Windows. (Roush) Linux currently has 3% of the desktop operating system market and that number is expected to double over the next three years. (IDC) Open source software still has a lot of ground to make up on commercial desktop operating systems, namely Windows. Linux is just now getting to the point where it can be easily installed, a feature that Windows has arguably possessed for a few years. As open source gains popularity in the server operating system market, firms should begin to test open source projects on their desktops, which 29 Ryan Tool Econ 191A/B - Crawford will increase market share and may give Linux the chance to challenge Windows for the consumer desktop. Combining studies for browser market share and desktop operating system market share we can make some interesting inferences. Windows has about a 94% of the desktop market, yet it has less than 90% of the internet browser market. IE is the default browser for Windows and it is the only browser installed on almost all new Windows desktop systems. The fact that Microsoft’s browser market share is smaller than their desktop market share shows that Windows users are switching to other browsers, most notably Firefox. The switch to an open source browser could be a gateway into other open source projects. The server operating system market is where open source starts to become a viable contender. In a survey of 225 CIO’s conducted by Morgan Stanley, 29% said they own Linux servers, 25% said they are considering buying Linux servers, and 46% said they are not considering Linux. (Ricciuti) When measuring which operating system is used on internet webservers2 you can count the number of physical servers or the number of domains, which is different because webservers are capable of running multiple domains on the same physical computer. When measuring by physical server, Windows Server is the market leader running on half of all webservers. 2 Webserver operating systems are quoted in many studies because they can be easily counted, whereas corporate servers that are not used for public services are more difficult to measure. 30 Ryan Tool Econ 191A/B - Crawford (Source: Netcraft) When measuring by domains, Linux edges out Windows Server by a slight margin. (Wheeler, Zoebelein) Since its conception, Linux has gained momentum slowly chipping away at Windows server’s lead. Based on current growth, Linux is in a position to overtake Windows as the dominant server operating system in the near future. A survey of server operating system market share in 2005 might show Linux much closer if not ahead of Windows in number of physical webservers. Linux is in a perpetual cycle: as it improves as a server 31 Ryan Tool Econ 191A/B - Crawford operating system its market share grows and the number of developers using Linux grows, which in turn improves the product and leads to more growth.3 Another measure of the success of open source server operating systems are their use in large scale, mission-critical environments. Google, the world’s largest and highly used search engine uses Linux for its cluster of servers that handle web searches. (Netcraft) Yahoo! another enormous search engine with many additional services, uses FreeBSD for all of its webservers. Traffic on the Yahoo! global network increased to an average of 2.4 billion page views per day as of March 2004. (Yahoo!) Most websites don’t get 100,000 page views over their lifetime, but Yahoo! serves billions of pages per day, a mammoth amount of traffic, all from open source systems. Amazon.com, the world’s largest book retailer saved millions when it switches its servers from Windows to Linux. (Shankland, Kane, and Lemos)4 There are also informal, less scientific ways of measuring the popularity of operating systems, such as the Operating System Sucks-Rules-O-Meter. (http://srom.zgp.org/) This is a system of operating system quality and approval based on a periodic AltaVista search for each of several operating systems, directly followed by "sucks", "rules", or "rocks". The current results have Linux 3 Linux is open source, but not all versions of Linux are free. There are commercial distributions of Linux that are used for server operating systems. While these commercial server versions affect our comparisons within our definition of open source, they are offset by some open source BSD distributions that we are not including in our comparisons. 4 The American Economic Association, publishers of American Economic Review, The Journal of Economic Literature, and The Journal of Economic Perspectives, uses FreeBSD, Apache, and PHP for its website. (Netcraft) 32 Ryan Tool Econ 191A/B - Crawford with a net metric of 38,600 and Windows with a net metric of –30,475. Obviously, these measures can not be used to ascertain which system is more widely used, but they do show the frequency of search queries on operating systems with the associated positive and negative terms. Email is a very important technology that is widely used in all forms of business. It is hard to imagine a successful company today that does not have an email address. All of the email that circumnavigates the internet is managed by email servers. When a user opens their email program or logs onto to their email web portal, they are connecting to a remote email server. Each of these email servers runs a server operating system and email server software. There are three popular choices for email servers: Sendmail, Qmail, and Exchange. Sendmail is an open source project originally developed by a student at Berkeley. Qmail and Exchange are both commercial email servers. A survey of one million random IP addresses found that Sendmail is the market leader with 42% of the market. Microsoft’s Exchange email server was second with an 18% market share. (Wheeler, Bernstein) So, for a mission critical service such as email many firms depend on open source. Another important market segment where open source clearly dominates is the webserver market. Below is the market share for webservers across all domains from September 1995 to June 2005. 33 Ryan Tool Econ 191A/B - Crawford Market Share for Servers Across all Domains Sept 1995 – June2005 (source: Netcraft http://news.netcraft.com/archives/web_server_survey.html) Apache became the market leader in 1996 and has gained market share ever since. Apache is open source and it is freely available to all. Internet Information Server (IIS) is Microsoft’s webserver with the second largest market share. IIS was once sold as a stand alone product, but it is now bundled with Windows Server Editions, so it is available to everyone running a Windows Server product. Microsoft has spent an enormous amount of time and money to develop a competent webserver, yet their product still lags behind Apache. The fact that Apache is robust and completely free makes it a hard product to beat. There are a number of reasons why open source has been adapted for use by firms but has yet to become widely used end-users, or consumers. Open source projects are more likely to develop and flourish in more technical market 34 Ryan Tool Econ 191A/B - Crawford niches. Technical users are likely to test and adapt cutting edge technologies such as open source. Tech-savvy users will try different solutions to solve their problems. For example, a web designer or web developer is accustomed to writing in various web programming languages. Such a designer can “learn” the language used to build an open source application and tailor it to solve their specific needs. They can then use the open source software to develop a web page with code specifically designed to fit their implementation. The more technical in nature that a specific software category or niche, the more likely it is to have open source projects. This idea is easily supported by looking at a few internet technologies that are vital to internet communications and implemented by very technical engineers. Whenever a user goes to a specific webpage, they type in the address to that page in the form “www.somename.com.” Address servers on the internet then translate this address to the numerical address where the website resides. This is similar to looking up a person’s name in a telephone directory to find their phone number. These address servers are called DNS servers (Domain Name Service) and they run specific software that allows them to perform their jobs. The program which currently has 70% of the market share is BIND, an open source project. (Moore) Some of these websites are static, like a billboard, changing a few times a year with updates. Other websites are dynamic being recreated with the latest information every time a user visits the website. The majority of these dynamic websites are created with a server-side scripting language which pulls data from 35 Ryan Tool Econ 191A/B - Crawford a database on the server and creates the webpage at the time of the user request. The most widely used scripting language with over a quarter of the market share is PHP, an open source project. (Netcraft, Wheeler) Many administrators need to connect to computers remotely to configure or update various services. This is a common task for all IT related support and administrative staff. The Secure Shell Protocol, or SSH, is widely used to securely connect to remote computers and control them. The most popular implementation of SSH with a market share of 67% is OpenSSH, another open source project. (OpenBSD) From these examples it is easy to see that technical IT employees trust and rely on open source software. IT workers trust open source because they can see the code inside the program which is not possible with closed commercial software. Also, many of these open source internet technologies were developed by academic and developer communities as they built and improved upon the internet. We can infer that the likelihood that an application will have open source alternatives is highly correlated with the technical nature of the application. This is intuitive as technical users will be able to successfully use the software even though it may be lacking the documentation or ease of use of a comparable commercial application. The trend of using open source software in more technical settings is evident with the use of open source in the backbone services of the internet. This trend is similar in many corporate environments, with Windows being used on the majority of end-user desktops and a split of open 36 Ryan Tool Econ 191A/B - Crawford and commercial operating systems being used for corporate servers. This could explain why Linux has 3% of the desktop market and roughly 30% the server market. Another reason open source flourishes in more technical applications is total cost of ownership. Every webpage on the internet resides on a server which runs webserver software. Creating and managing a website is a technical task, so most companies hire a webmaster to perform these duties. The webmaster and his salary are both necessary, but the cost of the webserver software is not if open source alternatives exists. If OSS and commercial software are perfect substitutes for some tasks, then open source is the clear choice lowering costs for the company and increasing profits. Open source software may not be a perfect substitute when it requires more knowledge to implement, but when you have to hire a webmaster anyway then you have the option to use OSS for your websites. This is another reason that open source is widely used by firms in technical settings which require a technical staff, but it is not widely used by consumers. The savings from using open source over commercial software can be substantial depending on the size or the firm and the firm’s computer infrastructure. For example, Amazon.com cut expenses by about 25% and saved millions when the company switched to a Linux operating system for its servers. (Shankland, Kane, and Lemos) In Mississippi, public law firms and counties saved 70% by basing their new jail management system from open source software instead of using commercial alternatives. (Pickoff-White) The best part of this story is that they released their new jail management system 37 Ryan Tool Econ 191A/B - Crawford back to the community under and open source license. Other agencies can now use the system and won’t be required to recreate the development efforts or the development costs. Open source projects have made their mark in the more technical segments of computer software and they are just starting to break into the less technical, end-user segments. What kind of predications can we make about the future of open source software? The future of open source software looks promising. A survey by Information Week found 67% of companies use some open source software while another 16% expect to use it in 2005. (D’Antoni) If current trends continue, all firms will adopt OSS for at least one of their computing tasks. As open source projects continue to prove themselves in technical applications, they will begin to be used in less technical market segments. Open source software used successfully on firm’s servers will trickle down to the firm’s desktop operating systems. Once firms start to adapt open source desktops, those OSS desktop operating systems will be viewed as stable and they will start to penetrate the consumer market. Another big push for open source will come as developing countries begin to open up to computing technologies and the internet. In many cases, less developed countries will choose open source projects which are reasonable substitutes for commercial equivalents due to the decreased cost. This can be seen in Brazil’s choice to switch 300,000 government computers to Linux. (Goldmark) Brazil and China have both developed their own Linux distributions, 38 Ryan Tool Econ 191A/B - Crawford signaling that they see open source as viable solution. Developing countries are not the only one using open source. The US government is beginning to use open source for many of its computing needs. (Pickoff-White) Open source software is growing in use. OSS projects are market leaders in a few software categories and have notable market share in many other software categories. Open source is widely used by many firms for their server operating systems and server applications. Open source alternatives exist in almost all technical natured software categories. OSS has taken a small foothold in consumer software categories as well. As firms and government agencies begin to embrace OSS for all of their computing needs, open source will gain credibility and market share in consumer software categories. All of these observations add up to one giant prediction: Open source software will play an eminent role in the future of the information age. Conclusion Open source software is a complex public good created from communal efforts. The existence of open source software creates welfare for society; it is non-rival and non-excludable. A software developer may contribute to an open source project for a number of reasons, including: altruism, future earning potential, future employment, personal enjoyment, hobby, recognition, politics, and cult status. Software developers earn twice as much income as the average worker, which allows them additional resources (time and money) to contribute to open source projects. Each of these motives for contribution permits the 39 Ryan Tool Econ 191A/B - Crawford developer to gain additional utility from his or her contributions. This added utility ensures that contributions to OSS will be supplied. Furthermore, firms may solicit open source development for their needs, employing developers to create and improve open source projects. These firms create demand for open source developers and open source projects. Open source firms are different from commercial firms in methods of development and measures of success. Open source firms are able to operate with little or no costs due to donated labor and a number of technological breakthroughs that allow developers to collaborate across geographical regions. Innovations in decision-making hierarchies have allowed large, complex pieces of software to be developed by disparate software developers. These changes in development models from traditional software firms have allowed open source projects to evolve rapidly and secure a place in today’s technology driven world. Open source firms measure their success by the viability of their product not by profits. The goals of an open source firm are generally a reflection of the goals of the individual developers. This causes the fundamental motivation behind an open source firm to be quite different from a commercial software firm. Open source projects have measurable success in technical software categories. Analyzing current market share across server-based products shows that open source is competing directly with commercial software. A reduction in total cost of ownership is one determinant behind this trend. The likelihood that an open source application exists for a specific task is highly correlated with the technical nature of the task. The more technical a software category, the more 40 Ryan Tool Econ 191A/B - Crawford likely open source alternatives exist and are widely used. Open source software is just starting to be used in consumer-based software categories. We predict that as open source continues to flourish in server-based software categories, it will trickle down into consumer-based software categories. As open source projects thrive, more developers join the projects adding to their success. This recursive growth may allow open source software to become a dominant force in the computer industry. Open source software has taken the computing industry by surprise, evolving out of nowhere onto computers everywhere. Some open source projects have displaced market leaders while other projects struggle to survive. There will always be arguments whether open source or commercial software offers the “best” solution, but one thing that can no longer be argued is that open source software is here to stay. 41 Ryan Tool Econ 191A/B - Crawford References Baake and Wichman. “Open Source Software, Competition and Potential Entry.” Jan 4, 2004. Bessen, James. “Open Source Software: Free Provision of Complex Public Goods.” Working Paper, August, 2004. Bitzer, Jürgen. “Commercial Versus Open Source Software: The Role of Product Heterogeneity in Competition.” Economic Systems. Vol. 28, 2004. Bitzer and Schröder. “Bug-Fixing and Code-Writing: The Private Provision of Open Source Software.” German Institute for Economic Research, July, 2002. Bonaccorsi and Rossi. “Why Open Source Software Can Succeed.” Research Policy. Vol. 32, 2003. Computer Associates. “CA Announces Winners of the Ingres Million Dollar Challenge.” Computer Associates International, Inc. May 18, 2005. <http://www3.ca.com/Press/PressRelease.aspx?CID=69484> D’Antoni, Helen. “Open-Source Software Use Joins the Mix.” Information Week. Nov. 1, 2004. <http://www.informationweek.com/story/showArticle.jhtml?articleID=51201599&ti d=5979> Ferguson, Charles. “How Linux Could Overthrow Microsoft.” Technology Review, June, 2005. Forge, Simon. “Open Source: The Economics of Giving Away Stuff, and Software as a Political Statement.” The Journal of Policy, Regulation and Strategy for Telecommunications Information and Media. Vol. 2, No. 1, 2000. FreeBSD. <http://www.freebsd.org> Goldmark, Alex. “Brazil Makes Move to Open Source Software.” NPR. January 31, 2005. < http://www.npr.org/templates/story/story.php?storyId=4471963> Goode, Sigi. “Something for Nothing: Management Rejection of Open Source software in Australia’s Top Firms.” Information & Management. 2004. 42 Ryan Tool Econ 191A/B - Crawford Grand, v. Krogh, Leonard, and Swap. “Resource Allocation Beyond Firm Boundaries: a Multi-Level Model for Open Source Innovation.” Long Range Planning. Vol. 37, 2004. Gutsche, Joerge. “The Evolution of Open Source Communities.” Topics in Economic Analysis & Policy. Vol. 5, Issue 1, 2005. Hahn, Robert W. Government Policy Toward Open Source Software. Washington: Brookings Institution Press, 2002. Hawkins, Richard E. “The Economics of Open Source Software for a Competitive Firm: Why Give it Away for Free?” Netnomics. Vol. 6, 2004. Hertel, Nieder, and Herrmann. “Motivation of Software Developers in Open Source Projects: An Internet-based Survey of Contributors to the Linux Kernel.” Research Policy. Vol. 32, 2003. Johnson, Justin P. “Open Source Software: Private Provision of a Public Good.” Journal of Economics & Management Strategy. Vol. 11, No. 4, 2002. Krishnamurthy, Sandeep. “A Managerial Overview of Open Source Software.” Business Horizons, September-October, 2003. Lakhani and v. Hippel. “How Open Source software Works: ‘Free’ User-to-User Assistance.” Research Policy. Vol. 32, 2003. Lakhani and Wolf. “Why Hackers Do What They Do: Understanding Motivation and Effort in Free/Open Source Software Projects.” Working Paper MIT Sloan School of Management. September, 2003. Lerner and Tirole. “The Open Source Movement: Key Research Questions.” European Economic Review. Vol. 45, 2001. Lerner and Tirole. “The Simple Economics of Open Source.” NBER Working Paper, March 2000. Meyer, Peter B. “Episodes of Collective Invention.” BLS Working Papers, August, 2003. Moore, Don. “DNS Server Survey” May 23, 2004. <http://mydns.bboy.net/survey/> Mustonen, Mikko. “Copyleft – The Economics of Linux and Other Open Source Software.” Information Economics and Policy, Vol. 15, 2003. 43 Ryan Tool Econ 191A/B - Crawford Mustonen, Mikko. “When Does a Firm Support Substitute Open Source Programming?” Journal of Economics & Management Strategy. Vol. 14, No. 1, 2005. Myatt and Wallace. “Equilibrium Selection and Public-Good Provision: The Development of Open-Source Software.” Oxford Review of Economic Policy, Vol. 18, No. 4, 2002. Netcraft. <http://news.netcraft.com/archives/web_server_survey.html> Nosko, Layne-Farrar, and Swatz. “Open Source and Proprietary Software: The Search for a Profitable Middle-Ground.” March, 2005 OpenBSD. “ScanSSH.” < http://www.openssh.org/usage/index.html> Open Source Technology Group. <http://www.ostg.com> Pickoff-White, Lisa. “US Government Agencies Turn to Linux.” MacNewsWorld. April 11, 2005. <http://www.macnewsworld.com/story/42048.html> Prasad, Ganesh. “Open Source-onomics: Examining Some Pseudo-economic Arguments About Open Source.” Linux Today. April 12, 2001. <http://linuxtoday.com/infrastructure/2001041200620OPBZCY-- > Raymond, Eric S. The Catherdral & the Bazaar: Musings on Linux and Open Source by an Accidental Revolutionary. Sebastopol: O’Reilly & Associates, 2001. Ricciuti, Mike. “Open Source: Rebels at the Gate.” CNET News.com <http://news.com.com/2009-1001-961354.html> Roush, Wade. “An Alternative to Windows.” Technology Review. September, 2004. Sandred, Jan. Managing Open Source Projects: A Wiley Tech Brief. New York: John Wiley & Sons, Inc, 2001. Sarrel, Matthew D. “Browser Wars Redux.” PCMag.com. May 16, 2005. <http://www.pcmag.com/article2/0,1759,1815941,00.asp> Shankland, Kane, and Lemos. “How Linux Saved Amazon Millions.” CNET News.com. October 30, 2001. < http://news.com.com/2100-1001275155.html?legacy=cnet&tag=owv> Sorceforge.net. <http://www.sourceforge.net/> 44 Ryan Tool Econ 191A/B - Crawford St. Laurent, Andrew M. Understanding Open Source & Free Software Licensing. Sebastopol: O’Reilly & Associates, 2004. The Apache Software Foundation. < http://www.apache.org> The Mozilla Foundation. < http://www.mozilla.org > Torvalds, Linus and David Diamond. Just for Fun: The Story of an Accidental Revolutionary. New York: HarperCollins Publishers Inc, 2001. Välimäki and Oksanen. “The Impact of Free and Open Source Licensing on Operating System Software Markets.” Telematics and Informatics. Vol. 22, 2005. Wayner, Peter. Free for All: How Linux and the Free Software Movement Undercut the High-Tech Titans. New York: HarperCollins Publishers Inc, 2000. Weber, Steve. The Success of Open Source. Cambridge: Harvard University Press, 2004. WebSideStory. <http://www.websidestory.com/products/webanalytics/datainsights/spotlight/05-102005.html?lpos=Promotions&lid=Promotions+-+Position+4> Wheeler, David H. “Why Open Source Software / Free Software (OSS/FS, FLOSS, or FOSS)? Look at the Numbers!” May 9, 2005. <http://www.dwheeler.com/oss_fs_why.html> Williams, Sam. Free As In Freedom: Richard Stallman’s Crusade for Free Software. Sebastopol: O’Reilly & Associates, 2002. Zhao and Elbaum. “Quality Assurance Under the Open Source Development Model.” The Journal of Systems and Software. Vol. 66, 2003. 45