For large-scale actions in the technology sector, little effort can be more daunting than running a security update on a software platform with more than 2.6 billion users.
But that’s exactly what Facebook does every time it deploys an update. Of course, this mitigates the potential for terror by making batch modifications and performing an incredible amount of internal testing. But at the end of the day, you never know exactly how a given change could upset the delicate balance of users that keeps Facebook on people’s screens.
Fortunately for Facebook, the company’s artificial intelligence team recently developed a pretty smart way to ensure that these software updates and changes don’t spoil its platform and don’t cost users: they’ve built a fake Facebook-wide social media network full of robots. to test things out.
By a corporate blog post:
TO IMPROVE SOFTWARE TESTING FOR THESE COMPLEX ENVIRONMENTS- PARTICULARLY IN THE AREAS OF SECURITY, SECURITY AND PRIVACY PRODUCTS – FACEBOOK RESEARCHERS HAVE DEVELOPED WEB SIMULATION (WES). WES IS A NEW METHOD FOR CONSTRUCTING THE FIRST LARGE-SCALE, VERY REALISTIC SIMULATIONS OF COMPLEX SOCIAL NETWORKS.
Basically, it’s a copy of Facebook filled with bots. Robots are driven by AI models representing the human capabilities of social media. In essence, robots can add friends, like posts, and generally do everything a person can do on a given social media platform.
These robots don’t look like the ones you’re used to seeing on Twitter (scream @crème_infinie on Twitter) that exist just to respond when a text trigger occurs. They are intended to simulate the experience of using a social networking site.
According to Facebook:
WE USED WES TO CREATE WW, A SIMULATED FACEBOOK ENVIRONMENT USING THE PLATFORM’S ACTUAL PRODUCTION CODE BASE. WITH WW (THE NAME IS INTENDED TO SHOW THAT THIS IS A SMALLER VERSION OF THE WORLD WIDE WEB, OR WWW), WE CAN, FOR EXAMPLE, CREATE REALISTIC AI ROBOTS THAT SEEK TO BUY ITEMS THAT ARE NOT ALLOWED ON OUR PLATFORM, SUCH AS FIREARMS OR DRUGS. BECAUSE THE BOT ACTS IN THE ACTUAL PRODUCTION VERSION OF FACEBOOK, IT CAN SEARCH, VISIT PAGES, SEND MESSAGES AND TAKE OTHER ACTIONS JUST LIKE A REAL PERSON. HOWEVER, ROBOTS CANNOT INTERACT WITH ACTUAL FACEBOOK USERS AND THEIR BEHAVIOR MAY NOT AFFECT THE ACTUAL USER EXPERIENCE ON THE PLATFORM.
QuickTake: This seems like a very smart way to determine whether a security feature or new user functionality is working properly without risking an interrupted user experience in the human version of the production code. I hope these simulations will become the status quo for social media networks.
But realistically, the simulation itself solves Facebook’s biggest problem: human users. Which I wouldn’t give for an invitation to the robot version.