Our tracking feeds are very powerful: not only they let you subscribe to keywords and boolean expressions, but they also let you subscribe by filtering on meta-data.
As you know, each feed processed by superfeedr has a popularity rank, combining several signals. Using this, our tracking feeds can be used to subscribe to the most peopular feeds. It’s fairly simple:
- Build a track feed with this query:
- Test it by using our search API.
- Subscribe to the track feed.
- After this, your endpoint will receive each entry from all the feed which match your query.
Here’s lthe stream of stories published by sources with a popularity rank greater than 5, in english only (
query=popularity>6). We use Server Sent Events to stream the new items to this page as they arrive.
Distribution of content published by popular feeds
When we introduced our popularity ranking, people asked what was the ranks distribution for the feeds. The distribution is exponential: there are orders of magnitude less feeds with popularity P than with popularity P-1. The graph below shows the distribution (percentage of total feeds) for feeds with a popularity greater than 1. (we removed anything below because it acconts for 85% of feeds!)
Each of these feeds obviously have a different velocity, which means that the distribution of published items by popularity of their feed is slightly different. Here, stories published by feeds with a popularity smaller than 1 account got 97.1% of stories, so we decided to skip it in the graph below as well.
As you can see, below a popularity of 5, the order of magnitude of stories published is similar. This means that feeds with a popularity of 1, 3, 4 and 5 are each more a lot more verbose than feeds with lower velocities. This is not surprising as this represents feeds from forums, verbose blogs (with multiple authors) or 2nd-tier news sites which tend to publish a lot more than less popular individual blogs for example.
After 5, though, the number of stories drops sharply as there are so few very popular feeds.