A Plea for Curation

A Plea for Curation
Photo by BoliviaInteligente / Unsplash

This just might do nobody any good. I've had this post drafted for while, and it might seem a little out-of-date given the current focus on AI. Nevertheless, there's still a point to it.

Back in 2023, mere weeks before Twitter rebranded as X, Meta launched their newest product: Threads. People flooded to the site and it became one of the fastest to hit 1,000,000 users. There was just one complaint: it forced you to use a feed that was dictated by an algorithm, not just a timeline of who you follow. The app was soon updated to have a “Following” tab that shows posts from who you follow in chronological order, but the default feed is still dictated by the algorithm, and there’s no way to have the Following tab open by default when you launch the app.

This is why I prefer Bluesky over Threads, and gets to the core thing I’m looking for in any social media platform - I’ve already curated a list of who I follow, so I don’t want the algorithm to try to do it for me, especially if it’s going to insert content from people I don’t follow.

It’s been extensively documented that the purpose of any social media algorithm is to keep you on the site. Analogies of slot machines and the chemical rewards our brains give us are well founded and understood. Like many people, I try to pry myself away from this without disconnecting entirely. “Doomscrolling” on my phone in the evening just leads to guilt about wasted time, but not doing so leads to a fear of missing out.

As a compromise I recently looked into something that I haven’t used for ages: RSS feeds. There’s a nostalgia I have for the internet of 20 years ago, when content was published on blogs as long-form articles. RSS was a useful way to curate your own collection of sites and keep up with them. There was a finite amount of content, and your attention was never demanded - you only paid it as much attention as you wanted and to the sites you had picked.

There are plenty of RSS readers still out there. In an effort to see if I could go back to those habits, I asked ChatGPT for some recommended sites to follow. Of the ten it recommended, two were behind paywalls and three no longer had working feeds. After explaining the issue to it a few times, it then spent nearly three minutes thinking, verifying which feeds might still work, and gave me a list of sites which broadly worked, but didn't quite match what I was looking for.

I do wonder if a lot of this started with the fall of Google Reader. At the time it was scrapped it was an unexpected announcement that upset a lot of people - it was one of the most popular ways of following sites via RSS. Google is known for pulling products, even if users are quite vocal about how much they like the product in question. Reader was removed in favour of a focus on Google+ – Google's effort at competing with Facebook. That project was also eventually cancelled for low usage, but it was a focus on the social aspect rather than the content.

The typical way of finding content now is through social media – Facebook, TikTok and Twitter have algorithms designed to push engaging content to us to keep us using their apps.

But the algorithms are flawed.

They are self-reinforcing, so if the algorithm recommends something to me and I click on it, it thinks I must like it and pushes more. If you're wondering about the radicalisation of people via social media, this is the entrance to the rabbit hole. It bears a striking similarity to the challenge facing AI of model collapse, in which a machine learning algorithm trained on data it produces tends to produce worse results. Unfortunately the result in this instance isn't comedic or even frustrating inaccuracies. It's people being fed more and more extreme, dangerous and false narratives, the polarisation of our culture, and the rise of racist, sexist, and discriminatory content.

I'm not for a second suggesting that the fall of Google Reader is a cause of this, but I do think it is a symptom.

It's entirely understandable that an instinctive reaction to the amount of dangerous material available on the internet is to more tightly control how people receive it, but I do wonder what would happen if we let people curate it themselves instead. Rather than taking control away from users, what would happen if companies ceded that control to the users instead? We would certainly lose some convenience and the instant gratification of opening an app to find a list of ready-to-go content, but it would force us to put much more thought into what media we wanted to consume.

Realistically, I highly doubt social media companies would ever build products to allow that level of curation by default. I'm utterly convinced that Facebook's move from a feed of everything posted by your friends to a list of suggested content is quite deliberate. But if we were all given the option, I can't help but feel the internet would be just a little bit better.

@robl@hachyderm.io | @robl.bsky.app | github.com/rob-lowcock