"In our view, at this early stage research and theory building on open source software development should be receptive to debates in several fields and attempt to make linkages with important work done in other disciplines."
Georg von Krogh & Eric von Hippel
Open source theory and practice is inherently interdisciplinary. Viewing the challenges faced by open source communities, businesses, and contributors through the lenses of different disciplines can yield novel solutions. This article reviews select lessons from the diverse fields of fashion, gaming, and scientometrics. It examines the way these other industries have addressed issues that are of relevance to the open source community and suggests ways to put these lessons to good use.
Lessons from the Fashion Industry
In a recent talk at TEDxUSC, Johanna Blakley shocked a largely technically-focused audience with an ingenious examination of the impact of the fashion industry's lack of copyright protection on its ability to rapidly innovate. She argued that, "because there is no copyright protection in the fashion industry, fashion designers have actually been able to elevate utilitarian design, things to cover our naked bodies, into something we consider art. Because there's no copyright protection in this industry, there's a very open and creative ecology of creativity."
Much like the practice in open source software communities, fashion designers can take ideas and designs from their peers and incorporate them into their projects, reusing them in novel ways to serve their own purposes. This rehashing gives designers a broad palette to work with and can act as a springboard for advancement since new innovations can be built upon existing foundations of style. In some areas of innovation in fashion, the practice is not unlike the development of extensible stacks of open source software, working together to serve a purpose. In one instance, the purpose is more artistic, and in the other, more functional, but the commonalities are sufficient in the innovative process itself that it deserves a further analysis.
Blakley explained that one of the most useful side effects of a culture of copying was the establishment of trends. These trends are hashed out of the remixing that naturally emerges when people try to define their own clothing styles. Very few people will wear clothing by one designer alone. Instead, they will wear pants by one designer, a shirt by another, shoes by yet another, and so on. Their style is as much defined by the individual articles of clothing as the combination. Different people have different needs and value different things in clothing design, from visual appeal to functionality, on a sliding scale depending on their personal preference. This model is nearly identical to that of the open source bazaar, where different users have different needs and they value different things. Very few users have a computer that has software from one vendor alone. Trends emerge when users find a useful feature or design idea in a piece of software that another vendor picks up and incorporates into their offering.
Perhaps the most stunning supporting evidence of the impact of the creative liberty of the fashion industry was a comparison of the gross sales of goods of major industries that have copyright protection and those that do not. Blakley showed that there are many industries that thrive with low intellectual property protection for designs, including the food, automobile, furniture, and fashion industries. Gross sales in these industries greatly exceed those in industries with copyright protection, such as films, books, and music.
Blakley wraps up with a cautionary note on creative industries and the evolution of the legal frameworks that govern them: "The conceptual issues are truly profound when you talk about creativity and ownership; we don't want to leave this just to lawyers to figure out. They're smart, but you want an interdisciplinary team of people hashing this out, trying to figure out, what is the kind of ownership model, in a digital world, that's going to lead to the most innovation. Fashion might be a really good place to start looking for a model for creative industries in the future."
This TEDx talk can serve as a lesson for open source communities as it demonstrates that many older and larger industries have struggled with the timeless challenges of encouraging innovation, creating and maintaining competitive advantage, and deciding whether to protect intellectual property or not. These industries have found ways to integrate the realities of their cultural and legal environments into their processes and corporate structures. Open source communities, and companies that have traditionally held their intellectual property close to their chest, may want to look closely at whether these strategies could help increase their revenue or competitive advantage.
Lessons from the Gaming Industry
In a recent article for Forbes, Elliot Noss, President and CEO of Tucows discussed what he has learned from playing video games, and how they have shaped him into a better leader. Tucows Inc. is one of the largest domain name providers. It is actively involved in Internet governance issues and is a strong supporter of open source principles, practices, and communities. Noss noticed the similarities between the workplace and player-organized events in World of Warcraft. He explains that when people are coming together to achieve a common goal, such as defeating a difficult dungeon, it is "really easy to see how valuable are skills like managing the social dynamic, making sure there is the right level of preparation, and making sure that there is a clear hierarchy in terms of who is performing what roles. Each action, even a small task has a purpose and fits into a broader framework." He noted the analogy to the management of Tucows, where each employee has a job to do that feeds into the whole, supporting the company's goals. By promoting an open dialogue with his employees around the way the company is run, its history, and its challenges, he found that employees have a better understanding of what they do and have greater job satisfaction. Feedback suggests that "it helps people feel they are part of something bigger."
The analogy holds for open source communities as well. Applying these lessons to community development efforts could help promote involvement, improve communication, promote a sense of belonging for developers, and possibly even reduce community fractures. Research by Bonaccorsi and Rossi has shown that open source participants frequently contribute to projects to promote a sense of belonging in the community. This parallel from the gaming community supports this notion and prescribes a means of improving that sense of belonging. Community managers could learn how to motivate participation and focus effort by observing the top guild leaders. Perhaps most importantly, approaching open source development like one approaches a game could lead to participants having more fun. As Lakhani and Von Hippel observed, open source participants who enjoy what they are doing may contribute more readily to projects, leading to benefits for the whole community.
Another lesson from the gaming industry comes from the Humble Indie Bundle project. The project's goal was to bring together independent game developers and charities to offer a game bundle in exchange for donations. The bundle consisted of five games and gamers could donate any amount of money above one cent to get all of the games. In addition, all the game developers agreed to release their games under open source licenses at the end of the project. This strategy is comparable to one of Frank Hecker's open source business strategies: "sell it, free it" and provides a good case study of this model in practice. In Hecker's model, the product is initially offered for sale and is later released as open source at the end of its life cycle. The strategy in the Humble Indie Bundle project is somewhat different in that the open source release occurred one week later, while the products were still commercially viable.
The project was a great success with nearly $1.3 million donated in just over one week. Over 130,000 people contributed an average of $9.18 each. Self-reported Linux users donated an average of $14.49 each, while self-reported Windows users donated an average of $8.05 each. The donation system allowed contributors to choose how their donation was distributed. Approximately 30% of the total donation amount was allocated by users to the following charities that are well known in gaming communities: EFF and Child's Play Charity. The remainder was allocated to the five game developers, who split the amount and each received over $160,000. The project demonstrated that the "sell it, free it" model can be an effective open source strategy. It also highlighted that creativity and community involvement are essential tools for success using this revenue model.
Lessons from the Field of Scientometrics
Scientometrics focuses on quantifying and qualifying scientific achievement. Since the early days of academia, people have sought to measure achievement, and these measurements have a broad range of applications. Measures of achievement are often considered when determining promotions, awards, funding, tenure, and recognition of contribution to a field as a whole.
Quantifying the health of an open source ecosystem is a challenge faced by many communities. There are many ways of measuring the individual contributions of participants, but it is not clear which ones are most closely correlated with ecosystem health. The communities and the projects upon which they build are so diverse and their goals are so distinct that it is a challenge to find uniform measures. This challenge is similar to that faced in scientometrics, where different scientific fields are so different that uniform assessment is complicated. Yet, over the past century, there have been many advances in the field, and measures have been developed that address these challenges. Open source communities could benefit greatly by applying these lessons to the assessment of contributions to their projects and ecosystems.
The major step in the task of measuring achievement is attempting to distill a unit of measure from the array of types of contributions. In science, numerous measures have emerged, including numbers of papers accepted to peer-reviewed journals, number of citations by peers in the field, number of projects or students supervised, number of patents granted, number of keynote presentations at conferences, impact of research, number and amount of grants received, number of chapters or books published, complexity of problems solved (especially in mathematics), commercial viability of research, and institutional involvement. These many measures have emerged due to the diversity of scientific research, and certain measures are more applicable in particular situations than others.
By contrast, in open source ecosystems, the number and types of measures of contributions are still fairly limited. The common measures include number of lines of code contributed, number of bugs reported, number of features coded, and perhaps amount of development on a project, especially for older projects where seniority is valued. These measures, while useful in some contexts, fail to capture the full range of types of contribution to open source ecosystems. They focus almost exclusively on contributions made by programmers. What about the contributions made by community organizers, evangelists, users, complementary projects, retailers, researchers, artists, editors, reviewers, and some of the many other important roles in open source communities? These roles are essential to community health and growth, but are not well measured. Open source communities could learn from the scientometics research to find ways to better quantify these types of contribution and the overall health of an ecosystem.
The Eclipse Foundation has been researching ecosystem health measures for some time. Donald Smith, the Director of Ecosystem Development for the Eclipse Foundation, deduced that there are three major, relevant measures of ecosystem health: i) productivity, or how much value is being created in the ecosystem; ii) robustness, or how readily the ecosystem can adapt to external events; and iii) niche creation, or the ability to expand the ecosystem with meaningful diversity. These measures are extremely complex and cannot be easily assessed. They also focus on the community as a whole, as opposed to individual contributions to that community. The question remains how to measure individual contributions to the goals of increased productivity, robustness, and niche creation.
In a recent special feature, Nature investigated the use of metrics for quantifying scientific contribution. Richard Van Noorden discussed the emergence of specific measures such as the h-index to quantify the impact of a particular researcher's contribution to the field. The h-index uses a two-dimensional assessment that has number of publications on one axis and number of peer references as the second axis. The result is a numerical score. An author that has published 10 articles that have each been referenced in another article 10 times would have an h-index of ten. This measure is useful for balancing number and relevance of publications, but has many limitations. One limitation is that the measure can remain static over time as the research presumably becomes stale and less relevant, that it can never go down, and that it focuses only on publications and ignores other types of scientific contribution. The analogy of this measure in open source is the number of lines of code contributed along one axis, paired with the number of times those lines of code are reused in other open source projects. Such a number could be useful for assessing the impact of a particular programmer, but may be much less useful for quantifying other types of open source contribution.
The field of Scientometrics has also borrowed lessons from webpage ranking algorithms, such as those used by Google for its search result rankings. Filippo Radicchi and his colleagues analyzed the entire publication archive of the journals published by the American Physical Society, comprising more than 400,000 papers. Their end result was the Phys Author Rank Algorithm (http://www.physauthorsrank.org), which ranks the diffusion of credits across journals in a field. Such an algorithm could be developed to track the diffusion of code snippets across projects hosted on SourceForge and rank the value of such contributions. Yet, again, such a measure would focus on code.
Most recently, the challenge of quantifying different contributions has been examined by analyzing social media. Mentions of research are tracked across various social media platforms, such as Twitter, Facebook, and blogs to provide a real-time picture of research activity. It is difficult to separate passing mention by someone who is not an expert in the field from reflected commentary by a peer in the field, yet it is exactly this sort of measure that may be the most useful for quantifying non-code contributions to open source communities. For example, it may be well-suited to measuring the contribution of a user who popularizes a particular open source project by posting a story on an influential website. This effect is frequently observed on news aggregate sites such as Slashdot, Reddit, or BoingBoing. Someone who has never written a line of code can bring instant fame to an open source project, directing tens of thousands of potential participants to the project by simply crafting a story that explains its use and submitting it to a popular news site. It is obvious that this sort of evangelism can be very important to an open source ecosystem, yet conventional measures would not value this contribution.
Johan Bollen, a researcher at Indiana University, concluded in recent research on the generality of scientific impact measures: "The notion of scientific impact is a multi-dimensional construct that can not be adequately measured by any single indicator, although some measures are more suitable than others." His research suggests - and the lesson for open source communities is - that clearly defining and understanding that which we want to measure is the most challenging task. Metrics can provide useful information, but if it is not clear what is being quantified, metrics can give the illusion of accurate representation of an effect that does not exist.
Finding Further Interdisciplinary Lessons
The examples of described in this article illustrate the power of looking outside of one's own field of study for inspiration. In many cases, these lessons arrive by serendipity. But how can interdisciplinary lessons be actively sought out? The amount of information and the sheer number of fields from which to draw lessons is prohibitively large. A carefully considered approach should be used to find the most relevant lessons, which can be broken down into thought exercises that narrow down the focus to the most salient options:
1. Clearly define the objective. Is it to generate more revenue for the company, create more features in the product, promote the product to a broader audience, receive a grant, strengthen complementary assets, or increase user satisfaction? The objective frames the search for a solution and sets the stage for generating ideas to find it.
2. Generalize the problem. Nearly all problems have existed in a similar form elsewhere. It is highly unlikely that the problem is so unique that a completely novel solution is required. Creating an abstraction of the core problem by removing detail specific to the context makes it easier to see similar problems in other fields.
3. Look for patterns of similarity across fields. Active observation can happen during exposure to other fields during daily activities, such as reading and networking. Often, similar problems - and different approaches to resolving them - can be found where they were not noticed previously. Actively looking for patterns increases the chances of finding relevant lessons.
By examining the lessons learned in other fields, open source communities can adopt strategies to help improve their innovation, social development, and revenue. They can also learn to better measure contributions to the health of their ecosystem. Open source champions and community development managers should embrace broad perspectives and consider looking through the theoretical lenses of other disciplines when facing the challenges of growth, relevance, and sustainability of their ecosystems. Uncovering the specific interdisciplinary lessons that are most relevant to a particular community's current obstacles can be challenging and requires a considered approach, but the potential for reward is great.