When the term algorithm is used, it is often equated with declining reach in the social media world. This applies most strongly to the
Confusion? Or do algorithms perhaps only bring order to social media feeds? If you look at the factors for the Instagram and Twitter algorithms, for example, they are more than manageable. Any content, whether Facebook post, Instagram photo, or tweet, is rated by the algorithm. This evaluation, usually referred to here as Relevance Score or Relevance Model.
The factors are often identical. For example, a recent article on the Twitter algorithm was published in the Twitter Engineering Blog. The way Twitter rates content is very similar to Facebook and Instagram.
Factors Twitter Timeline Algorithm
Three characteristics are analyzed by the Twitter algorithm:
's The Tweet: Tweet timeliness, use of Twitter cards (photos, videos, link preview) and interactions
The sender of the tweet: Past interactions with the sender, the connection to the sender and the origin of the connection
's Recipients: What content and topics are relevant and how heavily Twitter is used
and many more we don't know in detail
If we compare that now with the Instagram algorithm, there are a lot of parallels. It is based on connections, interactions and personal relationships. Purely in terms of factors, algorithms in social networks are not witchcraft, but are based on the same approaches.
Social media algorithms in the sign of relevance score
Since the algorithms are programmed and not maintained manually, topics such as deep learning and machine learning are nothing special, but the rule. A large number of companies rely on these technologies and we encounter them almost every day. Deep and machine learning mean that content can not only be evaluated faster (relevance model).
The algorithms learn and pursue a goal. In the case of Twitter, for example, there is more interaction with tweets and an increase in dwell time. Again, the parallel to the Instagram algorithm. Since its launch, Instagram has managed to increase impressions, dwell time and the number of daily active users. Twitter, too, is reporting an increase in interactions.
An algorithm is not critical to the success of a social network or company. At least not alone. Companies that use algorithms cleverly and effectively, however, usually have an advantage.
Facebook algorithm pioneers
On the F8 Facebook also has a short overview of algorithms and the Auf der F8 hat Facebook auch nochmals eine kurze Übersicht zu Algorithmen und der How the News Feed algorithm works is given. Again, many parallels to Twitter and Instagram, which have certainly been inspired by Facebook. Facebook has a higher focus on fake news and click-baiting. But the explanation reads almost identical to Twitter's.
I also liked the statement of Mir hat auch die Aussage von Gunter Dueck at the re: publica very well. As a mathematician, I'm irritated that people are afraid of algorithms. I can even write one that I wouldn't be so terribly afraid of.
Algorithms in social networks are often based on the same factors. The relevance score decides on the individual feeds and the adaptation and optimisation is carried out by deep and machine learning.