Facebook wants to use a bot-populated simulation to fend off scams in the real world
We now have a closer look at Facebook researchers' "WW" project, a parallel-world version of the social network populated entirely by bots, via The Verge's James Vincent.
Why it matters: The simulation of the entirety of the social network is designed to help the company anticipate and forestall new forms of online mischief and scams using a copy of Facebook's actual codebase.
- The simulation app is "headless" — that is, there's no front-end user interface, no pages to read, just data interacting with other data to produce more data.
- Facebook's goal is to run experiments to learn how to discourage bad behavior in the "real" Facebook — like city planners simulating traffic flow so they can know where best to place speed bumps.
- The researchers say they haven't yet produced findings that have been used to improve the "real" Facebook's defenses, but they expect that to happen by the end of the year.
The big picture: Simulation is one of the most fundamental and oldest tricks a computer system can perform.
- Facebook itself is a single world-sized system that users interact with.
- Now the company has a second version of the system to help it find new ways to get users in the first one to do what it wants them to do.
Our thought bubble: You could look at this as an ingenious and well-intentioned effort on the part of Facebook's engineers to deter bad behavior on the scale of the world's largest digital platform.
- But you could also conclude that Facebook is a colossal Skinner box, and we are 2.6 billion rats.