F8: Mark Zuckerberg presents his 10-year roadmap
On April 12 in San Francisco, Mark Zuckerberg played along the necessary exercise every leader of a major U.S. tech firm has to go through by giving the opening keynote at the F8 conference. Demonstrating the service’s global reach, his presentation was also broadcast live at 29 meetups around the world, from Paris to Nairobi, and Dacca to Bogota. However, the presentation (which is also online) was not broadcast over Facebook’s new platform, Live Video, which is no doubt still being tweaked ...
Mark Zuckerbook set out Facebook’s mission before an audience of journalists, developers and partners: “to give everyone in the world the power to share anything they want with anyone.” “Everyone” must have Internet access, which is the reason for the social network’s many initiatives (drones, more efficient relay antennas, free access to basic services, etc.) to reduce access costs, improve coverage, and persuade the three billion people who are not yet online to join the network.
“Share anything” means making it possible to share increasingly rich and immersive content, from photo to video to 360° capture. “Anyone” describes the possibility of sharing different things with different circles, from family and friends to fans and more distant relations, through various group applications (Messenger, Instagram, Whatsapp, etc.).
The mission is part of Facebook’s 10-year roadmap, which covers the expansion of the Facebook ecosystem, and that of its various products (like search and video), as well as technologies such as Internet for everyone, artificial intelligence and virtual or augmented reality.
In artificial intelligence (AI): open-source modules
Facebook’s concrete applications are proliferating, from Moments , an app that makes it possible to sort photos and tag people, to a translator that understands SMS language, its news feed, its ranking algorithm, and, lately, its tool for the visually impaired that automatically describesphotos. During the F8, the company announced its decision to make some of its AI modules (hardware and software) open source, particularly modules developed by its FAIR lab.
Deborah Liu, Director of Product Management, reviewed the Facebook platform’s many assets: 1.6 billion users, 9.5 billion dollars paid out to developers (70% of whom are located outside the United States), and more than 90 languages supported. To keep wooing users, Facebook is offering new software tools, like Account Kit, which streamlines the sign-up process by using an email address or phone number, and People Insights, which helps get to know an application’s audience based on demographic information (anonymous).
Website publishers are not forgotten, with the new Save [on Facebook] button which makes it possible to save content, and the Quote Plugin which lets people share a quote from an article with just one click. Lastly, the Instant Articles program is now available to anyone with a Facebook page. The program makes it possible to read content directly in the Facebook page without going through a browser or leaving the social network.
Messenger opens the door to bots, “smart” and otherwise
Another front Facebook is tackling: instant messaging. Messenger, the application developed internally, now has 900 million active users, compared with more than 1 billion for Whatsapp, the application purchased in 2014. More than 60 billion messages are now shared every day on the two applications, three times as many as reached by SMS.
Building on its success, Facebook has been attempting for the last year to turn Messenger into its own platform. The goal is to impose it as an alternate to mobile as a gateway for brands and the media – given that the traditional gateways are controlled by mobile operators (voice or SMS services) or Apple and Google (applications). Last year, the group introduced the Messenger for business service that lets businesses talk to their customers. This year, Facebook unveiled the Messenger platform for developing chatbots that can talk to users and propose new services without having to install an app, visit a website, or call a voice server.
Already, 30 companies, including CNN and the Wall Street Journal have started using the platform and are developing their own bots, available at a dedicated app store . Concretely, Facebook is giving developers an API to automatically send and receive messages (along with text, images and video) to and from Messenger users through an application hosted on a server. Users must still initiate the communication; the application then uses “conversational elements” to answer the request. They are visual lists of commands the bot can understand.
In this way, Facebook wants to avoid using command line key words, as is now being done in SMS (such as “Text 1 to vote for Loanna”), instead offering standard screens in the form of multiple choice “menus.” With CNN, users can read the list of news headlines and, for each possibility, request a summary in Messenger, or ask to be sent to a web page. Users can also enter a key word to get a selection of relevant news items. Some applications can offer “push” messages to follows the news item.
Messenger is not the first messaging application to offer this kind of integration; aside from SMS+ services, launched in the early 2000s, WeChat and Telegram have been offering their own bots for several years now. Microsoft also announced a Skype platform at the Build conference last month, and it would not be surprising for Google to do the same at the Google I/O conference to do the same.
These new platforms will promote the emergence of new players that are ready to provide turnkey “conversation solutions” to the brands and the media, triggering an onslaught of brand bots in chat services. The main risk is that most bots would not end up being that “smart,” and would be unable to understand the questions asked or create real value for users. In fact, there’s a risk that bots will simply steer users to existing tools, i.e. the website or app.
Facebook understands this risk well and is making its smartbot platform available to developers. That platform, wit.ai, was purchased from the startup of that name in 2015. The platform allows users to talk with a bot in natural language and the bot can predict what to do. This platform is akin to the bot framework announced by Microsoft last month. In the second presentation, Facebook also demonstrated the possibilities of voice conversation with a smartbot through a combination of voice recognition, image recognition, and natural language analysis.
Facebook Live Video, one more step toward Facebook TV
Live video is another focus for Facebook. In January, the social network expanded its live broadcast functionality to all users. Users can now use their mobile devices to shoot video and stream it to friends and fans. Realizing that such videos attract ten times the comments of video on demand, and that some draw bigger audiences than TV shows, in February, Mark Zuckerberg decided to shift into top gear and expand the platform team tenfold.
The surge has already resulted in the introduction of new video readers that allow viewers to express their emotions, as with Periscope, as well as better visibility for live broadcasts on the site and in Facebook applications. However, a mobile app will still be needed to broadcast video. At the F8 conference, the company therefore announced its decision to open up its Live Video platform to allow live video to be streamed from a portable camera, a drone or a video portal, all through a new API.
For now, only certified publishers and certain developers have access. One of the first partners, Buzzfeed (which drew more than 800,000 viewers to watch a watermelon blow up) is preparing a televised game concept to be broadcast from its sites around the world. For its part, CNN is looking into the option of communicating with its audience during broadcasts and including responses in the video stream.
Facebook is not overlooking video on demand: it announced a new rights manager that allows publishers to manage use of their content and prevent freebooting, i.e. the illegal release of video downloaded from other services. It was this practice that helped video take off on social networks.
Among other things, video is benefiting from advances made in artificial intelligence.Facebook will soon offer the option of automatically adding sub-titles to a video using voice recognition, tagging people who appear in a clip and when using facial recognition, and identifying all live broadcasts of a single event using a multicam function.
Facebook is making enormous investments to draw video professionals to its platform, even paying some of them. Can it make headway against established players like YouTube and Twitch? That will depend on its monetization strategy, which is yet to be spelled out.
Virtual and augmented reality, the last phase?
The last though not least of the topics addressed at the F8 conference: virtual and augmented reality. A lot has happened since the Oculus buy two years ago, including the marketing of a new low-cost version of the Gear VR headset in 2015. Users have already viewed more than 2 million hours of 360° video so far.
Since it was launched commercially, the first consumer Oculus Rift headset has been offering more than 50 virtual reality “experiences.” This headset is more immersive thanks to a system to track head movements. New Oculus Touch handsets will go on the market within a few months, enabling better interaction with the virtual environment. Mark Zuckerberg has said this will only be the first version. Others will follow, and incorporate advances in miniaturization. The final goal (within 10 years?) is to offer a virtual and augmented reality headset the size of a pair of sunglasses. As of then, some devices will be on the road to extinction: Facebook’s CEO claims, for example, that a TV set (which allows several users to watch a video at the same time) will be no more than a $1 app.
Aside from headsets, Facebook is investing in the other end of the chain to improve immersive content production tools, particularly for video capture. The company has therefore introduced Facebook Surround 360, a camera prototype that shoots in three dimensions and 360 degrees, paired with a stitching software. The camera produces 8K of content per eye, which can then be broadcast by a specific streaming protocol (and, it probably goes without saying, over a very fast Internet connection) to a VR headset. The camera itself should not be commercialized; Facebook plans to publish the specifications and source code this year so that others can use it by purchasing the required components (at a cost of $30,000). The approach the social network is taking is akin to that used by Google and its Jump camera, announced at the last Google I/O.
However, one of the main goals of Facebook’s research into virtual reality, which connects it to the social network and its artificial intelligence and video streaming initiatives, is to perfect “virtual social presence”: having the feeling of being with other people in virtual reality, when they’re not physically present. Much of Wednesday night’s presentation focused on the various experiments with “social VR”. A demo showed two people in two different places, each wearing an Oculus Rift headset, talking in a virtual space. Each person was represented by an avatar and a pair of hands. They could not only collaborate, but also “teleport” in 360° photos and even take selfies of the landscape and publish them on Facebook.
One of the people in charge of research and development for Oculus then laid out the challenges of virtual social presence. It must be possible to capture participants’ emotions, not only from their faces (hard, when a person is wearing a headset), but also from their bodies, to understand their body language. It must then be possible to broadcast these emotions live and reproduce them for other participants, so they can grasp the nuances. Lastly, it must be possible to implement “social prediction,” i.e. the possibility for the system to anticipate a movement or intention to mitigate the lags in transmitting data over the Internet.
Users of the Facebook VR of the future will then be able to feel like they’re with loved ones or strangers on the other side of the world. Maybe they’ll never come out again?