What does focusing on engagement mean for the platforms?

Platforms love engagement, but they have to pretend not to like many of the things users actually engage with.

by Rob Corbidge

Published: 15:37, 27 July 2023

Rob Corbidge is Head of Content Intelligence at Glide Publishing Platform, applying the latest knowledge about advances and ideas in the publishing industry to our own product and helping clients get the most from their content.

What advantage can publishers gain from a fuller understanding of the way the dominant social and search platforms surface content? 

Given that the current relationship between publishers and the big internet platforms is generally one of supping with the devil using a very long spoon, what weaknesses do the platforms exhibit that give us hope for the longer term?

An interesting piece of research by Tom Cunningham looks at the ways various aspects of ranking by content engagement are dealt with by the platforms. Cunningham is a data scientist who has worked at both Facebook and the Outlet Formerly Known As Twitter (OFKAT).

Cunningham describes the tricky corner that the major platforms frequently find themselves in: users favour sensitive content, yet platforms don’t want sensitive content but don’t want to be seen to be removing it.

Sensitive content also drives engagement, and engagement is a function of retention, the ultimate aim.

"Engagement necessarily measures the immediate reaction of a user to a piece of content, and thus ranking by predicted engagement will surface content that appears to be good," as Cunningham puts it.

Of course "sensitive content" is hard to define. This research characterises "nudity, bad language, abuse, hate speech, hyper-partisan politics, etc". There's a lot in that "etc". One person's sensitive content is another's splendid joke.

So the tech overlords, in attempting to create their online "town squares" have found that the people shouting crazy stuff at the edges of the square can be quite compelling for quite a lot of people. Difference attracts attention. TikTok, for the moment, seem mostly to be hosting dance competitions on their bit of the square.

Cunningham also says that the cost of running misleading headlines, or trashy content, is much higher for "traditional" media than for the platforms. 

This is true. One look at Facebook's latest transparency report for US user data indicates little has changed in the type of content that is popular, and of the types of account posting it. It's not awful, it's not great, but it's a reflection of the tastes of its consumers. Proper publishers dominate the Top 20 viewed links, but FB dismisses that with "The top 20 links seen here collectively accounted for 0.01% of all Feed content views in the US during Q1 2023." Thanks. Why do we bother?

When Cunningham writes "hard-headed engineers often argue that a user’s preferences are revealed in their engagement and that evaluating quality is paternalistic" I find myself in the hard-headed engineer camp, even though I'm not an engineer. You can't code for people. Too many variables.

Engagement for the platforms also brings another benefit, when a user interacts with the content by "liking", "retweeting" or commenting, a small increase in the value of that content is made. This means that the platforms have a vested interest in those who create controversy or a bit of a stir, as they drive engagement. 

And before anyone frames this in terms only of polarising figures or subjects of instant notoriety, how about simply a chainsaw fan group on Facebook, where someone posts a video of an equipment failure and declares "This company makes junk!". It might get just nine comments, but it's worth nine comments. Apply scale and you have a lot of engagement.

An interesting piece of the engagement puzzle is identified by Cunningham. As he states:

1. Measures of content quality have zero or negative correlation with engagement
2. Downranking low-quality content (equivalently, upranking high-quality content) increases retention. This is somewhat surprising because engagement and retention have a positive correlation, meaning the three correlations are not transitive

So, just because someone engages with something, it doesn't mean they'll form a positive opinion of it in the fullness of consideration, even if that consideration is unconscious. 

Therein is the simplest advantage publishers have. We can't hold our content at arm's length as the platforms still do at the moment. Although the arm is getting legislatively shorter, therefore we must hold it close and make the difference the relationship with the user. 

Easier said than done, but it can be done.