In a guest solo, Meta’s VP of Internationalization, Product Quality, and Product Experience, Iris Orriss, joins SlatorPod to talk about the company’s mission to give people the power to build communities, bringing the world closer together.
Iris recounts her journey to joining the language industry, from training as a geologist to working in big tech, from Microsoft to Meta. She also reveals her motivations for working in the field of internationalization.
The VP shares Meta’s strategy to remove language and cultural barriers through technology and the localization of major languages not well-represented on the Internet, such as Amharic, Khmer, and Sinhala. She discusses the scale of internationalization operations at Meta with more than 100 languages on the platform and millions of words translated every week.
Iris announces the end of their Community Translation program, where translators helped Meta make the Internet’s major languages available on the platform. The shift comes as Meta plans to migrate to a supplier source process, where they will work with single-language vendors (SLVs), specifically in African and Southeast Asian countries, to build Meta’s translation infrastructure.
The Meta VP rounds out the episode with what internationalization looks like in the Metaverse. This includes the shift from text to audio, personalizing avatars, and creating your own home.
Transcript
Florian: First tell us a bit more about your professional background. What drew you to the industry and how did you come to work at Meta?
Iris: I have an unusual story, but maybe it is a little bit more normal for my generation in localization because localization as an industry did not really evolve until somewhat recently, especially with degree programs in localization. I trained as a geologist and I have always loved languages. I have worked and lived in four different countries on two different continents and I came to the US because my husband is American. I came to the US in the late nineties and I thought, I do not want to work as a geologist. I loved studying it, but it was not what I wanted to do. I am in the San Francisco Bay Area, let me see what I can do, so I typed German, French, Spanish into the San Francisco Chronicle classifieds online drop site search and this localization testing job came up and they were looking for somebody who could test software in German and French, maybe Spanish. I applied to this technology company. It was a little startup called ABT and half the interview was, but you do not know anything about computer science, you do not know anything about programming, and I said, look, I have a STEM degree. This must be doable, but did you know how many years it took me to be fluent in all these languages. This is hard work. They liked me and they hired me at a low level and they said, we are going to take a chance on her, and then I ended up taking some programming classes at the local college to just get some more training while I was working there as a tester. Eventually, after about a year when I had more programming under my belt, I switched to localization engineering and I was a localization engineer, working on tools, on builds, in all the other languages for many years.
Then I was at Autodesk and then Microsoft for 10 years where I moved more into people management and I found this passion that combines my passion for languages and cultures and also technology. In 2012, I was approached by Facebook because the Director of Localization was exiting at the time and he knew me. He approached me and he said, maybe you want to think about this. I started researching Facebook and I started realizing that even their mission at the time was to give people the power to share and make the world more open and connected. I thought, with a passion for internationalization, with a passion for languages, this is the biggest playground in tech. This is where you get out of bed, go to work inspired and motivated because you want to solve these problems, these challenges, and figure out how to make the world more open and connected because you cannot do that without removing the language barrier and that is how I ended up at Facebook. It has been great. I joined in early 2013. We have scaled tremendously. We have gone beyond the Facebook product, there is Instagram, there is Reality Labs. There are many other products that we support today and it is an incredible place to be because the momentum in this company around anything international, around languages is very strong because it is center stage to its mission. The updated mission that we have now is close to the original one, which is to give people the power to build community and bring the world closer together and you cannot do that without bridging the language barrier and bridging the cultural barrier.
Esther: What would you say motivates you to keep working in the field of internationalization? Whether that is at Facebook and Meta specifically or more generally.
Iris: I have developed this framework for myself. I call it the three P’s: purpose, passion, and people. I really try to pivot myself along those three P’s and when I think about purpose, I think about the company that I am working for. Where do I put my time? Where do I go every day or five, six days a week, depending on how busy we are, usually five days a week? Where do I spend my time? What do I want to work on? When I was at Microsoft, the vision was a PC on every desktop and that inspired me because I felt that everybody in the world should be included in joining the knowledge economy, in joining the digital age and that is how I ended up there and very similar with Facebook and now Meta. I am very inspired by its mission. I feel like it has a purpose, to give people the power to build community and bring the world closer together has a purpose and resonates personally with me. Then the second piece is passion. I told you a little bit about my passion for languages and this came from me being an only child and having German parents who traveled the world everywhere. I do not know whether you have noticed, but if you go to the end of the world, you will meet somebody German and this is usually true. This is what happens to me. As an only child, I was dependent on being able to find other children to play with and to interact with and communicate with, and so very early on I wanted to learn as many languages as possible. I wanted to learn every language so I could communicate with everybody.
Of course, it is not possible to learn every language, but that instilled my passion for languages and it instilled my passion in trying to remove the language barrier across this gap and remove the cultural barrier. I spent time in three countries outside of Germany working and living there because this drives me. This gets me out of bed in the morning and so this is the passionate element of my role where I get to work on the language barrier. I get to work on global enablement. I get to work on the cultural barrier as part of my job to serve this mission and that is the second piece that is important to me. The third P, people, how can you do good work without having a great team? The people you work with are critical to how much you can get done. They are critical to your mental wellbeing in a way. You need to have fun at work. You need to have great people that support each other. You need to have a great culture. You need to be able to do your best work. You need to have diversity. You need to have inclusion to really accomplish those things that we set out to do. For me at Meta, I find that the people I work with every day are inspiring me and are amazing and so those are the three P’s that I always look for. Purpose, passion, people.
Florian: How important of a piece is language in Meta’s broader mission? Recently we came across some announcements around MT and generally it feels like it is being taken very seriously. Elaborate a little bit on that and how it fits into the broader purpose there.
Iris: Again, the current mission is to give people the power to build community and bring the world closer together. There is no way to bring the world closer together, the world that speaks 7,000 different languages with about half of them having writing systems unless we bridge the language gap and also the cultural gap. I talk about this in three silos when I talk about how internationalization is structured at Meta. The first piece is the technology side which I call global enablement. This is, can you process different currencies? Do you have the right time formats? Do you have the right date formats? Do you have the right number formats? Imagine somebody who is buying something or paying for something on Facebook today and the number formats are not lining up. That would be very bad and so this is the basics of global enablement. It also includes something like finding friends, so if you are in Japan, there are different ways to write your name. They have different writing systems, sometimes alphabetic, and then they have Hiragana, Katakana, and Kanji, and depending on how you write your name, we have to be able to convert these things in the background and make sure you find your friend. This is global enablement and that is more the tech side of internationalization. Then there is the removing the language barrier side, which is localization. This is about making our product accessible in languages other than English.
Then there is removing the cultural barrier, which means making our product aligned with your cultural context. This sometimes gets confusing for people, so I want to give a quick example of what I mean when removing the cultural barrier because nobody wants to change the product. Facebook is there for you to connect with friends and connect with family and so everybody in the world has this desire to connect with friends and family. It is inherent to us as humans and we are not changing that. If you, for example, live in Indonesia, one-third of the population there only has one name. There is no first name and last name. There is just one name, so in our original registration process, we ask for a first name and last name because we are a real name platform, so somebody from Indonesia with one name comes in and says, how do I fill this out? They fill in their first name and we say, we also need your last name now and they are very confused. Either they work around it by putting their name in twice and imagine you show up on Facebook as Florian Florian, that would not feel too good, or they just drop out of the process. This is bridging the cultural barrier. When you sign up in Indonesia you only get one name bar to fill out at the registration time and that is it. You can put in one name and it is fine. Those are the three things: global enablement, removing the language barrier, and removing the cultural barrier.
You asked about language and how it fits into Meta’s overall mission. I am focusing on the language barrier now and I divide this into three different categories. There are major languages. When I say major languages, it is any language that is spoken by millions of speakers fluently, so I would be a speaker of English in this definition, although I am a native speaker of German. First, there are major languages that are represented well on the internet, I call them major internet languages. Everybody is very familiar in this industry with those languages. It used to be the big eight and then it extended to 10 and now we are at about 30 where most companies localize into major internet languages, into what I call the top 30. This is Japanese, Chinese, German, French, and so on. There is also a lot of content on the internet in those languages in general. Then the second category is major languages that are not well-represented on the internet and the third category is all the other languages and that includes endangered languages. There is a huge list maintained by UNESCO of endangered languages. Of course, we serve the major internet languages, but our focus is on the major languages that are not well represented on the internet today. Think about Amharic, which is spoken in Ethiopia. Think about Fula from West Africa. Think about Khmer from Cambodia. Think about Sinhala from Sri Lanka. Those are languages that are spoken by millions, but they are not well-represented on the internet today and this is part of our strategy in terms of removing the language barrier to work through, not just the major internet languages, but the next languages that have the potential to become big online and it is part of our strategy. We have to remove the language barriers for those types of languages because otherwise, we cannot bridge that gap. We cannot bring the world closer together and currently Facebook is available in over a hundred languages. We added a few last year, we are adding 10 plus this year so we are going at a pace where we are trying to bring these languages online to fulfill our mission.
Esther: What do you think about the scale of internationalization operations at Meta? It must be immense, but are there ways that you think about the resources, internal, external, the volumes and the projects, and everything that you are working on?
Iris: First of all, it is my job to break things down and make it easy because that is how you communicate, not just externally, but internally too, to other executives because they need to understand what your department is doing and how you bring value to the company. In terms of scaling, we are available in more than 100 languages and we translate millions of words every week and so the scale is mind-boggling. When you think about millions of words every week, it is a very optimized operation end-to-end and when you think about what Facebook did back when it started, it was an app for college students. It was started in a dorm at Harvard and then it became popular very quickly and fanned out to different colleges. Very soon people around the world in colleges wanted this and people around the world outside of colleges wanted this, to connect with family and friends. I have reconnected with many of my friends from college. I wish I would have had this in college because some of them I cannot find and at the time it was a product with a very young audience and it was a concept that was very new. Social media as a concept was very new and what Facebook did at the time, it pioneered community translation at scale because this was the way to get the right terminology, to roll this out extremely quickly into many languages. Over the course of time, community translators have helped make Facebook available in major internet languages and in endangered languages. This is changing. This has gone from hundreds of thousands of community translators contributing to under 10,000 community translators contributing. We are seeing a shift here, especially with the focus on major languages that are not well-represented on the internet.
We are about to announce that we are ending the community translation program at Facebook and it will not be available anymore. We are moving to a supplier source process. We, of course, have had a supplier source process for working with MLVs and SLVs for some time. It was mixed, but we are moving fully to a supplier source process and I am excited about this because we have thought deeply about this. This is exactly the thing that I am looping back to with your question in terms of how do you think about this? How do you make this work? One of the things that we found too when you are looking at major languages that are not well-represented on the internet, they are also not well-represented in the MLV catalog and so this is a problem. Of course, I do not want to discredit the great work that MLVs have done here. They have started Hindi which 10+ years ago did not use to be in the MLV catalog. Now, it is a major language so they are reacting to this. They are moving very quickly in a way but they are not moving fast enough for us and so we have also built over the last few years a direct from SLV sourced model. We have put the tech in place that we can directly source from SLVs and we are working with SLVs in many countries, including African and Southeast Asian countries to help build translation infrastructure for these languages and get us to the next stage where 100 languages are standard. This is where we have changed from sourcing also through the community to supplier sourcing completely.
Florian: We are talking about the Facebook Translate App, where you can log in and there is a big drop down with a lot of languages. You are moving away from this to a Meta managed SLV-focused approach here. You are probably also calling on some of these super-specialized in-country SLVs to support that process?
Iris: You are absolutely right. We have given this a lot of thought and we have made a very big effort to figure out how to do this. We think that moving to a fully supplier sourced model that is very inclusive of sourcing directly from SLV, especially for the major languages that are not well-represented on the internet, is the next thing to do. We are still working with some MLVs. This is part of our portfolio. We have a very diverse portfolio of suppliers overall at our scale. I mentioned millions of words translated per week. This is a scale that demands a diverse portfolio of suppliers, but the supplier strategy of sourcing directly from SLV and doing it with a large focus on majority languages that are not well-represented on the internet to help build translation infrastructure is new. When I say build translation infrastructure, I want the audience to understand that in many of those languages there is no Alliance Française that tells us what is the correct way to spell things, what are the correct words? Often there are many dialects within that language. We have to work with a diverse set of people from the language to make sure we are not hitting a tone that might be one dialect versus another. For example, Swahili is a major language spoken in East Africa and Swahili in Tanzania is different from Swahili in Kenya, so we have to find ways to create a version of Swahili that speaks to everyone. This is very tedious, very difficult work, and I am super proud of our team there and of the people on the ground, and the SLV providers that work with us on this and make this work. It is hard work.
Esther: In terms of this SLV-led strategy, would you say that was mainly in response to the fact that the MLVs do not have adequate language coverage across all of these underserved languages? Or do you think the benefits of having an SLV-led strategy would have always convinced you to land on the SLV approach anyway?
Iris: It is a portfolio of responses. One of the reasons is that it is just as hard for MLVs to build this translation infrastructure as for anyone. We thought we could do this with the tech that enables making the translation process as easy as possible. Another thing is the quality element. When you think about translation, all of us are thinking about cost, but as a consumer product, I think deeply about quality every day all the time. I want the best quality. How do I get the best quality? I get the best quality the closer my relationship is to the actual person who translates the content or who translates the words for us. Pre-pandemic we used to have a translation summit every year at Facebook, now Meta, where we would invite all the key translators of all of our major languages into one big room in Dublin or in Menlo Park and we would have a big summit and the exchange was incredible. Now we cannot invite all translators, but we can invite a couple of lead translators or so, and they go back with all this knowledge about the product, about the context, about how we want to do things. This also encouraged us to go directly to SLV because the more you can minimize the process from the actual product, where the fewer people are in between that line, the better your quality is going to be and that is important. It is cost-effective, but it is quality significant. We have seen the results of this. We have seen the engagement, so this is something that was quality-driven, cost-driven, and circumstantial in terms of, let us figure out how to source these languages-driven.
Florian: How does internationalization look in the future metaverse? What shift are you expecting to happen? Speech-to-speech, avatars, customization, the whole gamut.
Iris: Facebook, the product, changed how people connected. Instagram is part of that, Facebook Messenger is part of that, WhatsApp is part of that, as well as part of the Meta portfolio today. In that sense, the metaverse is an evolution on that continuum. It is going to allow people to connect in different ways in the metaverse. When you think about the metaverse and how it is a very immersive experience that we are trying to build, my first thought on this is that everything will not just be about the written word that we translate today. Things are going to be about the spoken word. I want to be James Kirk, tap on my communicator and I want to talk to anybody in the world in English or German and they could speak their language and I want this just to be as seamless as Klingon and English in Star Trek. This is the dream. As an industry, I think machine translation has to become part of it. There always will be translation by humans. Of course, we need it, but the need to even have translation everywhere ubiquitously is going to force us to have it also by machines. It is going to force us to think about how we contribute to this shift to the spoken word and get that translated seamlessly and not just the written word. That is a major thing that we have to think about. How do we supply know-how and what is needed to make these models work and all of those things?
Secondly, there will be a shift to further personalization. I call it local at a global scale. At Meta, we are consumer products by and large. We never want to do localization by finding the least common denominator in this context. We want you to be in your local context, but then at the same time, we have to build anything that we do at scale. Think about Yahoo Finance using red for a good day, where the stocks are up in China, and using red for a bad day where the stocks are down in the US. Think about the sticker packs that we have on messenger, by and large people in different parts of the world respond to different stickers. That is just a fact of life. That is how we interact. We are in our own cultural context and so those are little things that we do today. When you think of this immersive experience in the metaverse, you need to think about the space that you create. What is your office going to look like? What is your house going to look like? What is your beach going to look like? You need to think about the artwork on the walls and the color schemes and all of those things that are highly personal. With your avatars, how do you want to be represented in the metaverse? How do we do a good job, first of all, of having an extensive catalog? This is where localization becomes very important. Providing that extensive catalog of things to enable this local at a global scale personalization, to enable this local experience within your cultural context and then the scale is how do you rank it. That is obvious because if I am from Germany, I might want to see the lederhosen but if I am from Indonesia, I might not even recognize that. How do you make sure every user gets what they like? What speaks to them? Those are the two shifts. The written word will still be there, so do not misunderstand me when I say shift but it is the addition of us thinking about the spoken word more and more and it is the shift to further personalization to have a local experience. I am lucky to get to work on this. I am excited about removing the language barrier, removing the cultural barrier to truly bring the world closer together so this is what gets me out of bed in the morning right now. There is an opportunity for the whole industry to rally behind this and enable this immersive experience for everyone, regardless of where you are on the planet, what language you speak and what your cultural context is. Half the languages in the world spoken today do not even have writing systems and so this could be an advantage for these languages and potentially enable these languages to become part of this digital world and contribute to their vitality in a way. That is our dream.