CrossOver now also monitors Mastodon

Mastodon’s popularity has been rising since Twitter’s takeover by Elon Musk: the day after he announced it, 70,000 users joined Mastodon. According to Fediverse.party, it is today the largest decentralized social network part of the Fediverse, with a population of approx. 6 millions accounts, breaking the 2 million active users per day as of writing of this article. This figure in unprecedented since the creation of the software in 2016. 

As seen from the perspective of a user, Mastodon looks and feels a lot of Twitter. One can follow people and see their “toots’ (or micro-blogging posts), click on hashtags to view more toots using the same hashtag etc. There is however a major difference between Twitter and Mastodon as the latter is a federated social network and does not enforce an unique content policy, but each instance carries the responsibility of moderating toots on their own service.

Mastodon is decentralised, meaning that there is no central authority owning and running everything. The network is instead composed of  several independent entities, each running their own copy of the open source Mastodon software on their own server(s), called instances. These instances however are all using the same protocol which allows users from one instance to see and interact with messages of other instances (except if a server content moderation policy has caused the ban of another Mastodon servers, or individual users). Examples of instances are mastodon.social (i.e. the OG one) or mastodon.brussels , mastodon-belgium.be etc. Users join a particular instance and must therefore accept this instance’s rules.

As opposed to tweets on Twitter, Toots are presented chronologically to the Mastodon user. One only  sees toots from the people they follow from the most recent to the oldest, without any intervention of a ranking algorithm. At current time, no ads or promoted toots are presented to Mastodon users.

Trends on Mastodon 

Mastodon doesn’t use any content recommendation algorithm but rather relies on pure  statistics to list the most interacted upon content tags,  per instance. According to the Mastodon documentation, trends reflect “hashtags that are currently being used more frequently than usual”.

Trends on Mastodon are therefore totally different than Twitter trends, as there is no algorithm responsible for content display. Administrators and moderators of a server use an interface where they can see the hashtags that are most used, and it’s an administrator (or a moderator)’s decision to display or remove a trend. This means that administrators or moderators are responsible both for choosing to display trends, but also to remove them. If an administrator leaves a trend visible even if no one talks about it, it will still appear on the feed. This can lead to unusual examples, where you can have a trend associated with no toots.

How do we collect data from Mastodon

Crossover monitors the 30 biggest instances according to https://instances.social/, an open source site that analyses Mastodon instances.

Crossover collects data from Mastodon’s instance which has an Open Public API. Data collected are the trends as well as some information about the server like the number of users, total number of publications, total toots and rules of the server. Data is collected every hour.

The trends are displayed in the dashboard where one can choose among the 30 biggest instances, a time of crawling, and see the rank of a hashtag, the number of toots associated with it and the number of users that posted something using this hashtag.