How Do Mozilla Practice Open Innovation?

Interview with George Roter, a Director on Mozilla's Open Innovation team


Formed in 1998, Mozilla Firefox is the second most used browser on the globe, with a usage share of around 30%, and has been translated into nearly 90 languages. Ahead of our Open Innovation summit, which takes place in London this April 25 & 26, we sat down with George Roter, a Director on Mozilla's Open Innovation team, to talk about how they practice open innovation.

Based in Berlin, George is responsible for a company-wide crowdsourcing function and developing global community initiatives that engage thousands of contributors. He is an Ashoka Fellow, was named as one of Canada’s Top 40 Under 40, and was awarded the Action Canada Fellowship on public policy. He has been recognized with four honorary Doctorate degrees, along with completing a Bachelor of Applied Science from the University of Waterloo. He has also been presented with the Meritorious Service Cross (Civil) by Canada's Governor General.

How do you encourage open innovation at Mozilla? Do you have any tips for successful open innovation?

In many ways this is easier inside Mozilla as we were pioneers in open source software development, crowdsourcing and contributing to standards bodies. So the cultural barriers are pretty low.

For us the challenge is moving beyond well-worn methods and practices, helping people update their understanding of what is possible. We’ve taken two important steps in the past 18 months.

The first was to set the frame for this work moving forward through a substantial strategy project involving internal research, external case studies and analysis. This was led by a newly formed Open Innovation Team reporting to our Chief Innovation Officer, so immediately we had some mandate for this work already built in. We built in broad stakeholder engagement to this project -- resulting a set of ~50 allies in influencer roles across the organization -- including executive sign-off. As part of this project we had a large internal communications effort to establish narrative and shared language for this work, “Open by Design”. We bolstered this internal communication through external communication, launching a blog, speaking engagements and some media outside.

With that in place, our second step was to build-out proof-points for this strategy through a series of pilot projects. One of these was Common Voice, which I’ll talk about during the Summit. The keys for these was that they:

Were fully resourced by our Open Innovation Team, so the only commitment we asked from internal “customers” was time.

Touched enough different parts of the organization.

Started building out new capabilities, specifically in crowdsourcing and innovation challenges.

How has open innovation contributed to the development of your machine learning speech recognition software?

In two ways: First, our speech recognition software project (codenamed DeepSpeech) is open source, inviting co-development from others who have an interest in software of this sort. For us, and many other companies, it makes sense for this part of the speech interface “stack” to be freely available, as all of the value-creation for users is built atop this basic function of speech-to-text. In that way, it has the starting conditions for an excellent open source project.

Second, any machine learning project requires quality data, and for speech recognition this means paired voice clips and transcripts. It turns out this kind of data is not publicly available at-scale even if we wanted to buy it. We’ve had to be creative in gathering this data, at first reaching out to other organisations who might have data, who are aligned with our overall aims and can also benefit from open source speech recognition software.

We also launched Project Common Voice, as a way to crowdsource voice clips from people around the world. We positioned this project and custom crowdsourcing app an opportunity for “speech recognition to understand me” and to provide alternatives to the dominance of the large internet plays. In the first 6-months over 20,000 people contributed nearly 500,000 voice clips.

Are there any other advances you put down to open innovation?

Mozilla was built on open innovation practices. From being a leader in open source co-development, to using crowdsourcing methods for localizing Firefox into 90+ languages, to driving coalition efforts like the Alliance for Open Media ( to develop alternative video encoding technology.

What are the challenges of adopting open innovation practices?

While our organization is pre-disposed to adopting open innovation practices, we come across predictable challenges.

What we hear the most is “I don’t have time” for that approach. Of course, there’s a reality to that challenge, which is that doing work through an employee or standard contracting method can be faster. So clearly the time pressure people feel trumps the benefits they might see.

The second challenge we’re seeing is conservatism around using new practices. This requires upfront education and also bringing down the “cost” of trying new methods. For example, we’re running a data science competition in parallel with solving the same problem using a contractor, to compare the approaches.

Do you think established organisations should be aiming for a startup mentality in order to embrace innovation?

I think that one characteristic startups tend to exhibit is agility, and I think that is crucial for innovation.

In the case of our team, we borrow heavily from Agile software development methods for most of our work, even if it’s not software related. This emphasizes learning and iteration, delivering and sharing work products regularly, not falling in love with solutions, forcing prioritization conversations, and collaboration.

How can innovators prepare their business for a transformed future?

There are three things that I think are crucial:

1.Developing the capability to solve problems through a multiplicity of methods, so that there’s choice.

2.Teams will be set-up and torn-down faster and faster, and will have members who are more distributed and remote. So we need to invest in “teaming”. We need to be great at quickly forming teams, getting them to high performance, and then learning as they are disbanded.

3.This is probably obvious, but every single business needs to fundamentally understand software. This needs to be at every organizational level and highly distributed. Every single industry and business model will eventually be “disrupted” by software, and it’s likely better to be anticipating or driving this than having it catch us by surprise.

Hear more from George at our Open Innovation Summit in London, April 25 & 26.

Lightbullb small

Read next:

What Makes An Environment In Which Innovation Can Thrive?