Who’s afraid of algorithmic capture?

People often break out online because they do something unexpected. Over time, they tend to become seemingly interchangeable with similar creators or unrecognisable from what originally made them famous. Is it the influence of newfound wealth and status or something about the act of posting online?

Audience capture is the theory that content creators are beholden to the whims of their audience and are pushed to more extreme expressions of what they believe the audience is looking for.  A topical example, outside of the social media realm, is the escalation in the extent of the "makeovers" performed on America’s Next Top Model. Seeing the reaction from audiences and ratings gained from performing drastic (and often unflattering) hair transformations, the team escalated to permanent cosmetic dental work. The goal of the procedures, and the overall show, was never to create models – it was to keep audiences coming back to watch people suffer in their attempt to become them.

Recommendation algorithms mean ‘following’ a creator no longer guarantees you will be shown their content. This introduces another element shaping the content we see and the people that make it. Algorithmic capture describes the opaque process behind what so many audiences feel but struggle to properly place blame for. The same style of content, the same ‘tricks’, the same look and sound and feel are favoured by recommendation algorithms and dominate what the audience ultimately can see.

The two have an unholy symbiotic relationship – it is a common belief amongst creators that giving the algorithm more of what it has already shown the most people is the best way to ensure your content is seen. Human audiences, on the other hand, can only be shown the same thing by the same people a finite number of times before they start craving novelty. To make something completely original is to risk the algorithm not showing it to anyone; to continue to make the same thing is to risk your audience turning on you. Only one of those has an immediate impact on your analytics.

Most attention is good attention when you’re getting paid per interaction regardless of whether the comment is effusive praise or a character assassination.

A niche example of this was explained well by Jamaal Burkmar in this video, “I think I’m done with the skit people”. Jamaal explains how what is often passed off as parody in serialised short form content focusing on ‘types’ of people (Letterboxd bros, film-character girlfriends, creepy coworkers) seems more like simple imitation. It fails as real satire as there is no second layer, and there is no catharsis in the pure reproduction of a type. The impression you're left with is the same as if the person were genuinely saying or doing those things. I would go further to say most of this content I have seen is aimed at women, either mocking them or the behaviour they’re subjected to by men, and seems vaguely or explicitly misogynistic. While a few people manage to be genuinely funny doing it, there are only so many times you can laugh at someone pretending to hate their girlfriend before you start to wonder who the joke is for.