It is actually much, much easier for me to find content I like on AO3 than on websites that use algorithmic curation. And no, it’s not because I “didn’t feed the algorithm enough data.”
It’s because most algorithms have hardwired priorities that are fundamentally opposed to my personal tastes.
Algorithms designed to maximize “engagement” rarely differentiate between positive and negative engagement. A lot of content is popular not because it makes people happy, but because it makes people angry, and goads them into an argument.
This is the reason the term “doomscrolling” exists. The algorithm is not delivering “good” content. It is delivering manipulative content that enables people’s worst impulses.
Another issue I have with algorithmic feeds is they always prioritize newer content over older content.
I love reblogging 6 year old memes. I love rickrolling. I love reading 10 year old fanfic and watching 15 year old tv shows and reading 20 year old books. I follow a blog called @yesterdaysprint that posts old newspaper clippings.
For the past couple days, I’ve been getting a bunch of notes on a funny gifset I made in 2020. I love that something I made two years ago is making dozens of people laugh today. That would never happen on twitter, where memes die within a week.
Beyond having the wrong priorities, algorithms also have a lot of limitations.
I can tell an algorithm what things I like, but I can’t really explain to the algorithm *why* I like them, and this leads to a lot of BAD recommendations.
Take the Netflix algorithm for example.
I started watching a lot of Chinese television on Netflix, specifically because I am trying to learn the Mandarin Chinese language.
The Netflix algorithm doesn’t know the reason I’m watching those shows.
I’m pretty sure the Netflix algorithm just assumes I have an Asian fetish, because 90% of the shows it recommends me are Korean. No matter how any times I type “Chinese” into the search bar, Netflix keeps pushing Korean media recommendations at me.
If I dislike a show, the Netflix algorithm doesn’t know whether it was the cinematography I didn’t like, or a particular actor, or the fact that my favorite character was killed off, or all of the above.
The algorithm can only guess my reasons, and its guess is often wrong.
So even though I probably use Netflix a great deal more than the average Netflix user, the Netflix algorithm still does a pretty shitty job of recommending me content I actually want to see.
That’s why I still trust the media recommendations of my friends and the tumblr mutuals I follow more than the media recommendations of any algorithm.
I’m getting the sense, from your comments, that you have a uniquely difficult time finding content you like on AO3.
Most of my friends are my age, and none of us have ever have any problem finding content we like on AO3, and this is largely due to our technological upbringing.
Understanding how databases work, and how to get what you want from them, is a skill that takes practice.
Everyone in fandom who used the internet in the late 90s has spent a lot of time developing this skill. We had no choice! There were no algorithm-curated content feeds back then. If you wanted fandom content, you found a web archive that hosted content for that fandom, and you learned how to input search queries into that web archive’s database.
So we were all forced to learn basic database literacy.
If you haven’t spent any time developing this skillset, an algorithm will seem like an easier way to reliably find content you enjoy.
However, if you spend a long enough time building this skillset, your ability to find what content you like eventually SURPASSES an algorithm’s ability to deliver content you like.
So understand that when you say, “It’s impossible to find fic I like on AO3” what we hear is, “I never learned how to use a database.”
When you say, “Algorithms are way more efficient and reliable than looking stuff up on my own,” what we hear is, “Nobody ever taught me how to look stuff up on my own.”