Attention Overflow

Written By James Thorburn
Please Like and Subscribe

What will happen when we have no more likes to give

When we are trying to influence public opinion, or gain support for an initiative or to draw people into our art or our business, we are presented with a basic challenge. How do we get people's attention? Where do we get their attention?

With the combination of AI and social media algorithms, this task has gotten tricky.

AI does nothing better than optimize to a rewards system, so it's most likely that it will win any war for attention on social media. I think that it will be a pyrrhic victory, though, because human attention receptors will be continuously saturated to the point of exhaustion.

We will have no more likes to give.

The venues in which attention has primacy, such as algorithmic feeds on social media sites, will be overwhelmed with nearly infinite AI-assisted content, and the fractional hit of dopamine that occurs with each new bit of content will trend to zero.

Where supply is infinite, value is zero.

I believe the "newness" of algorithmic social media feed content will come to feel old, and the new new will be distinctive individuals, with coherent perspectives, expressed in novel containers that are uniquely theirs. But the new new won't be the form, rather it will be those specific individuals.

For example, my family chat is just an iMessage chat. It is 10x more compelling than anything an algorithm is going to serve me. There are the five distinctive voices of my three older sisters, our mother, and me, and the complex, persistent engagement of a family. No random influencer chimes in with his hot take on the Wordle score my mother posted at 2am for unknown reasons - that's our job!

"Members-only" chats are becoming more popular along these same lines - coherent perspectives of people with whom we want to have persistent meaningful engagement.

The degree to which distinctive engagement will occur through traditional social media mechanisms is secondary to the fact that there will be other venues, more suited to the task. The burgeoning newsletter sector is primarily using your email In Box as its platform.

Newsletter authors use social media to supplement their engagement. The likes on their posts are irrelevant, the comments are more likely to be part of an ongoing conversation, whose primary home is, for example, their Substack page.

"Subscribe" is superior to "Like" in every way.

The problem for social media companies is that they need to keep users on the platform for as long as possible, so they need infinite like-able content to feed the scroll, which AI is ready and able to provide.

This is not a problem for humans, because we don't actually need to spend any time at all on social media platforms. The finite nature of the family chat (days will go by with no activity) is not a problem anyone needs to solve.

Podcasts provide another attractive venue, and their rise in popularity has oddly mirrored that of social media, despite sitting on the opposite side of the attention:engagement spectrum.

Marc Maron, one of the pioneers of the format, started his podcast (“WTF”) in 2009 as a failed comic, talking to his (usually more successful) comedian friends. This evolved into an hour long interview podcast, in which Maron would attempt to find genuine connection with his guests (typically actors, musicians, directors, writers and comedians).

He did not conduct interviews in a neutral voice like a journalist would (or like an AI persona might). These conversations were steeped in the six characteristics that AI Lacks:

  • Backstory

  • Friends and Colleagues

  • Humility and Self-doubt

  • Accountability and Shame

  • Humour

  • Point of View

This style of interview was adopted by many others (Howard Stern, Joe Rogan, etc.), some of whom went even longer than an hour.

In a media environment that assumes vanishing attention spans, somehow this format has been flourishing.

That is to say, there is evidence that sustained personal engagement is up to the challenge of competing with attention-optimized content, which is good news for humans.

Social media feeds will no doubt continue in a zombified state, potentially forever, with any number of users continuing to find micro-entertainments sprinkled through hours of mindless scrolling. These feeds are perched precariously on the 90-9-1 ratio of lurkers, occasional posters and super-posters.

But what happens when the share of super-posters goes up, thanks to AI? Won't the 90% of lurkers recede into the bushes, Homer Simpson-style, and won't the 9% of 'natural posters' give up because their hand-crafted content is lost in a flood of AI slop.

What does a feed of 90% super-posters look like? I don't think anyone will ever find out, because it will very quickly go to 100%.

But those feeds will not engender engagement, which is what we seek when we are trying to influence public opinion, or gain support for an initiative or to draw people into our art or our business.

Anyone who has had a post go viral knows how minuscule the pot of gold is at the end of the rainbow, how little investment each of the million likers of your weird cat video has in their relationship with you, and how completely detached their tapping of an icon on a screen is from the relationship with you.

The viral post is allegedly the holy grail, and yet it is effectively person-less.

Story-telling thrives on recurring characters, interwoven themes, longer time horizons. Story telling invokes a resonance between teller and reader.

Attention-based systems are incoherent and random. There is no narrative. Every hit is an island.

To develop persistent, authentic engagement over time, we will need to match how people consume information now, which is some equivalent of scrolling through posts - sequences of image and text diptychs, and with longer form podcast-style conversations.

Distinctive personalities, making persistent connections with individuals, will defeat AI in the competition for mindshare. Our task is to find natural ways to express our humanity in these tools, without getting lost in the likes.

Refs:

James Thorburn
Previous

There is no I in AI
Next

Stack Underflow