The News Feed is perhaps the most important feature of Facebook, and there is no doubt that the company spends a lot on researching algorithms that improve its content. However, it turns out that the algorithms are not enough – sometimes asking real people direct questions yields better results.
According to Backchannel, since the inception of its News Feed in 2006, Facebook has realized that simply observing the reactions to a user's timeline and counting likes and shares cannot reliably discover the content that people like.
A Project Manager of the News Feed gave an example comparing posts to donuts. If someone brings Donuts you will definitely eat them because they are tasty. But that doesn't mean you won't eat something else besides Donuts.
To improve the News Feed, Facebook is already testing a new feature on a random sample of its members in the United States. Using a special version of the social network, it provides the ability to rate content. Raters are asked to interact normally with the content, as they always have, then answer questions related to their experience, and also write a paragraph describing their feelings about what they went through.
This particular research started just in August 2014, and has already provided some interesting ideas, such as this confirmation of Facebook's content.
Facebook currently has about 600 reviewers for the project, but intends to continue adding new members as the tests will continue. With an ongoing trend towards automation, this research conducted with the interaction of real people is a welcome change.
