Remember that "Think Globally, Act Locally" bumper sticker you've seen on hybrids and VW buses? The folks over at Mozilla have really taken the message to heart, and it's worked out quite well for them — the next version of their Firefox browser is slated to ship with over 70 localizations, covering over 97% of the world's Internet population. Oh, and have you noticed that Firefox continues to grow market share at the expense of Internet Explorer? By the end of last year Internet Explorer fell from a 75.47% market share to 69.77%, while Firefox surpassed a 20% share for the first time ever, reaching 20.78% by the end of November 2008. John Lilly, a Chief Executive at Mozilla, called that a "significant milestone." We call it awesome, and we've got to think it's partly due to localization. As a matter of fact, their ability to engage the locals has been one of the most important factors in Mozilla's global success.
Another classic bumper sticker you've probably seen — "Wherever You Go, There You Are" — would fit in nicely on the back of the Mozilla bus. Seems to us that when it comes to scalability, it's less about being where you are and more about getting where you want to be. Which is, at least in Mozilla's case, everywhere. The problem remains, though, that even if you succeed in getting everywhere, what does it take to really be there?Ask the Mozilla team. For them, being where they are (which is nearly everywhere, don't forget) means being there in many, many different languages — very few of which are spoken by Mozilla's engineers. By building participatory local communities they've gained access to those languages as they are used by native speakers, and nurturing those communities has paid off in many unexpected ways as well. If we could boil all their experience down into one simple yet profound slogan, we'd just go out and print a bumper sticker. Since we can't, we continue our occasional series on open source community building with a case study by Seth Bindernagel about the localization process at Mozilla. What follows are some lessons in scalability from the people who have made it possible for users all over the world to gain access to a state-of-the-art, free and open source web browser that works in their own native language.
Between the release of Firefox 3 and the upcoming release of Firefox 3.5, Mozilla will add 25 new localizations to the browser's list of official versions. As a point of reference, the most recent release cycle lasted one year, starting officially after the release of Firefox 3 in June of 2008. When we started the localization process for Firefox 3.5, we had 48 localizations. It's likely that we will ship with 73. This represents tremendous growth in both our localization community as well as coverage of new languages spoken by the world's Internet population. By our rough estimate the Firefox browser now covers over 97% of the world's Internet population with a fully-localized user interface that includes translation, localized preferences settings, and a user's native language option. So what does it take to increase community contributions and locale count by over 56%? It may not be a new science for some readers, but we think that some of the lessons we've learned about scaling are worth sharing.
In case you've heard elsewhere, it's true that the Mozilla localization process is not perfect. We use a fairly unique file setup to highlight the strings needing localization, and that has had an impact on our ability to scale. With some introduction it's probably not too difficult to understand, but because we have historically chosen a framework that uses "DTD" (Document Type Definition) and "property" files, it has required localizers to learn our system, which has caused a higher barrier to entry than most other projects.
So how exactly do we lower the barrier for participation for those passionate about localization? One simple step has been to support and promote tool development for translation work that reduces the level of technical background needed. By providing tools for contribution, and helping tool authors refine and improve their software, we have started to scale with more ease. New localizers with less software development skills have become regular contributors responsible for their locale's translation.But tools are really not the only answer. Building the proper team has also been crucial. A small group of Mozilla employees focuses on localization. We call ourselves the "L10n-drivers" and provide as much technical assistance as possible when localizers need help. When a project has substantial localization needs — like Mozilla, with its multiple products and websites — the work flow can be overbearing at times, so it's helpful to have a response team to assist the volunteer community and shepherd projects along to completion.At the same time, it helps if the volunteer localizers have some technical background so they can work with us and all of our tools to make localization possible. This may sound like a contradiction to my previous point, but it's not. We walk a fine line between trying to make localization accessible to everyone who has translation interest and asking volunteers to learn a bit about our technical process, like how to check in code changes to their locale repository. Localizers also need to know how to do things like read HTML and execute limited tasks from the command line. It's far from perfect, but finding individuals with the dedication to stick with us and the curiosity to learn our process has led to a more sustainable community.
There's another critical piece that we take seriously at Mozilla: full transparency. The Mozilla L10n-drivers team never does anything that we cannot immediately share with our community. In fact, most everything we do directly involves our community every day. We created a public dashboard where all localization teams can see the status of their work as well as the work of other teams. In addition, several of our community members write active blogs that are syndicated to the Mozilla community through our Planet Mozilla service.
Communication is another factor that has helped us scale. Consistent and open communication is critical to our success because it provides channels for both newcomers and long-time localizers to ask questions, learn, and help each other. With other centralized communication channels like Bugzilla, Mozilla mailing lists, and IRC (Internet Relay Chat) rooms, we are in constant communication with the community, available nearly 'round the clock to share information and answer questions. The L10n-drivers team is located in several different time zones across the globe, so we are able to provide fairly regular coverage if something arises. A true sense of community exists on our #l10n channel on the Mozilla IRC server, where conversations are always happening. Providing open lines of communication and making sure that everyone understands that no questions are out of bounds helps us build trust and reliable relationships with our global team.Within these platforms of communication, many roles and relationships develop naturally and exist because there is a clear set of tasks to undertake if someone wants to localize Mozilla. This leads to another great reason for building and scaling a successful community. We try our hardest to outline the specific tasks that anyone can take if they choose to localize Mozilla in their native language. Without a clear outline, a localizer may begin to wonder what his or her role is. But when we define those tasks clearly, there's no room for confusion.If a particular localization has a long-established community of contributors, newcomers are often set to new, untouched tasks that need a team leader. We let our teams know that there is *always* something more that can be translated, so there's no need to worry about not finding work. Successful L10n teams in the Mozilla community have mapped out specific roles and tasks for their team members, while the L10n-drivers work hard to provide direction on where to focus. With this combination of efforts, our 100% volunteer community takes shape and participates in understandable and digestible tasks.
A few other simple but specific factors add to our success. We try hard to document everything that is critical for localization. Those documents are updated regularly and shared with the community so that people can see what needs to be done and how to do it. Of course, questions are always directed back to one of the channels mentioned above. Also — and this is a specific technical piece that is crucial for our community — we provide regularly updated builds to our community so they can see the fruit of their labors and begin testing. Since our community is fully volunteer, we have to provide them something that can be reviewed and tested on a regular basis. We generate nightly versions of Firefox so that they can see and test their work.
It's taken us awhile. Years of experiments, mistakes, and improvements along the way have shaped the strategy. But what now exists is a remarkable achievement by this community. The latest version of Mozilla's Firefox browser will be shipped simultaneously to 73 locales across three platforms: Mac, Windows, and Linux. Because of this effort, nearly 300 million people across the globe are using a free and open source web browser in a language familiar to them, making their experience on the web the same as any other locale where we ship Firefox.
Allowed tags: <a> link, <b> bold, <i> italics