We get a lot of positive feedback about the articles that appear on reallyread.it, especially the ones that get selected for Article of the Day. Over the course of the last few years, we have tended to focus on promoting other aspects of the platform, things like our tracking technology and our commenting community, and yet many people keep saying that the best part of reallyread.it is that we have the best articles.
Along those lines, some people have incorrectly assumed that we, the reallyread.it team, also serve as community moderators. We definitely don’t. We abide by the same rules that our community must abide by. For example, we can’t edit or delete our comments and we can’t comment on things we haven’t really read. So when it comes to the curation of articles on our home page, we don’t have any special privileges; we put all of the power in the hands of our algorithm!
So, how does stuff turn up on there in the first place? Every article and story that ends up on reallyread.it is there because people in our community have read it. Since we’re tracking reading behavior on an individual level, we have a huge dataset - hundreds of thousands of article interactions, week over week. This dataset provides a granular view on (1) what’s being read (2) how it’s being read, and (3) where people are skimming and scanning, etc. If you’re reading this blog post, you probably won’t be surprised to hear that only a tiny fraction of interactions are full reads. A vast majority of the time everyone’s just bouncing around on the shallow surface of the web. (So cheers to you, reader! Keep digging deep.)
Here’s another way to think about it: the lists on the homepage are basically just complete lists of every single article we’ve got. You don’t see hundreds of thousands of articles for the same reason you don’t see millions of Google results. You just see the best ones. This is why “curation” and “sort” are actually different ways of describing the exact same thing.
For now, articles on the home page can be sorted in two ways: Hot and Top. The Hot list takes three factors into account: number of reads, number of comments, article length. Additionally, it considers those factors in the context of time, prioritizing recent reads and comments. The Top list is exactly like the Hot list except it doesn’t prioritize recent activity, so it is essentially a ranked list of articles with the most engagement of all time on the platform.
The Article of the Day is automatically selected at midnight EST. The honor is bestowed upon the highest ranked article on the Hot list that hasn’t already been Article of the Day. So, nothing can ever win twice. And no two articles can win on the same day.
In some ways, the algorithm is the soul of the community. Everyone contributes and the output keeps getting better as the community continues to grow. More data means better results. And there are no echo-chambers within reallyread.it since everyone sees the same articles. Personalization can be a good thing, but it can also lead to a situation where you end up seeing some small piece of the world instead of the whole (sometimes messy) thing.
We’re in the business of getting people to read, so we can’t have sub-par content hanging around. Reading is magic, or at least it can be, and we aim to give our users the best chance of experiencing that magic every time they open our app.
We’d love to know what you think. How do you like the articles and stories featured on reallyread.it? What do you think about the way we curate articles? Any other thoughts or ideas? Leave a comment!