Why am I seeing this post?
Algorithms have had a tough few years. Every time there’s an update to any of the ways that the social newsfeeds surface content to us, the internet shakes its collective fist at the enigma that we call “the algorithm.” Facebook has announced that they’re going to be giving us a set of tools that shed a lot more light on how they work, which they hope will dispel some myths surrounding their mysterious invention.
To use the new feature, just tap the ellipses at the top right of any post in your newsfeed. As with all Facebook/Instagram features, it takes a while to roll out to all users, but reading between the lines, it looks like we’ll all have this one pretty quickly.
Who needs robots in our newsfeeds?
The reason we need algorithms in the first place is that we simply create and follow too much stuff to possibly take it all in, so the equation is set up to surface the content that it believes we’ll be most likely to engage with. For all of their hard work, the platforms have been getting very little love from their users:
I wish more people interacted with my IG. Algorithm been working against me for ages tho
— Jake Thompson (@__27K) March 31, 2019
INSTAGRAM ALGORITHM IS KILLING ME pic.twitter.com/2D2IccRL4l
— 🌹IG: dxm.beauty (@dxmwright) March 31, 2019
My final story for @slate is about the mystery of a seemingly random, 119-word local crime brief that somehow became Facebook’s most-shared article of 2019 so far. https://t.co/xVhBUhLnBs
— Will Oremus (@WillOremus) March 29, 2019
If the algorithms are set up to show us what we want to see, then why all of the hate?
The problem is that the question of what we want to, or need to, see is a very subjective one, and machines aren’t (yet) very good at subjectivity. They are great at following rules, but hard and fast rules don’t allow for the fact that we may Like a cat video more often than a thoughtfully written piece of editorial content, but that doesn’t mean we all want newsfeeds full of cats.
So how does it work?
1. It takes inventory
The moment you log into Facebook, the machines start their work. The first thing it does is collect an inventory of all of the content that it could potentially serve up to you. That includes all of your friends’ recent posts, and all of the posts from the pages that you follow.
2. It looks for signals
This is where the AI kicks in: the algorithm scans all of the content for signals about what it is and what it’s about, and then it creates a fingerprint of that post. The fingerprint is the source of much of the controversy in content distribution because it’s what runs the risk of categorizing only the content that agrees with our viewpoints as relevant to each of us. But, up to this point, all that the algorithms have done is create a well organized library – the magic happens in the next step.
3. It makes predictions
Here’s where things get really interesting. A machine is actively working to predict what humans will and won’t do based on trillions of data points. Based on what we know about data and human behaviour, you’d think they would be so smart that they could absolutely nail down what we’ll like every time. You may find it somewhat of a relief the next time you log in and see content that’s not a great match because it means that the robots have not yet figured us out.
Those predictions are based on many factors, including Affinity, Weight, and Decay. These three are left over from the original EdgeRank system, but they’re thought to still make up an important cornerstone of the predictions system.
Simply put: the content that you’re predicted to like most is from a source that you regularly interact with (Affinity), the post has lots of Likes & Comments (Weight) and it’s been posted recently (Decay).
4. It creates a score
Based on the first three factors, Facebook/Instagram bundles them together to create a score that uniquely ranks that single piece of content with your single user profile. Higher score = higher ranking in your newsfeed.
All of the above information has been cobbled together through a combination of Facebook’s semi-cryptic releases and through years of experimentation, and the addition of “Why Am I Seeing This?” information is going to go a long way to help us all track exactly how the algorithms are working.
It means that our Followers are going to have more information about why our content is showing up for them, and it also means that we’re going to be able to infer a lot more about how our posts are getting distributed.
Every time a development like this happens, the brands that benefit most are the ones that take it seriously, that pay close attention, and that make real shifts in their strategy based on the lessons they learn.
The newsfeed “Why Am I Seeing This?” button joins a feature that ads have had for a while. While the organic post feature may take a little while to appear in your version of Facebook, the Ads feature was rolled out a while ago and is available to everyone.