I wonder why nobody has tried to conduct an experiment to create a completely disconnected subnet on Facebook with robots that mimic communication and browsing friend feeds, with detailed logging of everything for subsequent analytics? Start with around 500-1000 “people”. Use real posts (which can be taken from another social network, first checking that it hasn’t yet appeared on FB) and perform full scanning of post outputs and recommendations for each of the 500-1000 participants. This would yield big data that could be made publicly available for processing by anyone interested. Technically, it’s not complicated or costly, right? There are farms for clicking posts, but they are not disconnected from the external world, and properly conducting experiments is not feasible.
Hi, I'm Rauf Aliev. 