Today, we are introducing the public beta of the Fur You feed on Furrylist. After our popularity-ranked Hot feeds, it is our first true algorithmic feed. It seeks to prioritize relevant posts from your network, scoped to accounts on Furrylist.
(Did you know? The Furrylist feeds service and our moderation tooling, is fully open source, including the Fur You feed's code.)
How to try it
To try the Fur You feed, open it in your Bluesky app of choice, and start scrolling. If you want it visible on your Home screen, tap the pin icon (📍).
If you like the feed, you can show that by using the heart icon. If you have feedback or suggestions, please join our Discord server and share it in the #fur-you channel.
Despite some testing by friends of Furrylist, the feed is still very much experimental. If the feed breaks or you run into anything you don't like or that looks unexpected, please let us know too! We want to make a feed you like.
How Fur You works
Fur You works very similar to the For You feed, with a few minor but important differences. Most importantly, it only shows posts from furry and furry-adjacent (i.e. brony, pup, therian, poke-/digimon, ...) accounts that asked to be added to our list.
We start with your most recent 500 likes and find users, who have at least an overlap with two of those posts, called your similar users. We now find all posts that your similar users have liked that were posted in the last three days by accounts followed by @furryli.st (aka. vetted accounts).
We rank posts by the most recent like by one of your similar users. We boost some posts from the accounts you follow, that you have liked before, that were liked by someone you follow, or that have less than five likes by people followed by vetted accounts.
We add a small bit of randomness to the post ranking, so it feels less mechanical and static. We also prevent posts from the same account showing directly after each other because that was a big usability problem in the early design phase.
The vibe we're going for is starting with a majority of posts from people or accounts you've seen (or even follow!) to slowly expanding to accounts in your network. You can stop at any time or keep exploring if you're curious.
How we got here
At time of writing, Furrylist features 32 custom feeds. Almost all of these feeds show posts in reverse chronological order, meaning that the newest posts are shown first. This format is great for following live events like conventions or for quieter feeds like our furry literature feed.
However, feeds like the regular art feed, which shows thousands of posts a day, can become very overwhelming and show increasingly irrelevant posts. That is why we created the Hot feeds. They are based on a regular feed and rank posts by popularity.
The ranking algorithm for our Hot feeds is inspired by Reddit's same-named Hot ranking. It ranks posts by popularity over time. Instead of showing a list of the most liked posts in a day or week, it looks at posts that were posted and frequently liked recently.
The desire for more relevance
The Hot feeds are great, and our most used feeds. But popularity isn't necessarily a measure of good or relevant posts. Feeds with a niche focus like the aforementioned furry literature feed can surface posts you care about but they are bound to a specific topic.
Now, we could throw machine learning (or “AI”) at this task and create a non-deterministic social media algorithm like the big players (Instagram, TikTok, ...). Perhaps it would actually be good! Or, in the worst case, it would create an environment of perverse incentives and massively degrade the vibes of the Furrylist feeds.
Therefore, we spent the last few months trying to come up with a good deterministic algorithm that we can explain transparently and stand behind the vibes it adds to our feeds on Bluesky.
One of the feeds we've found very interesting in our research is spacecowboy's For You feed. While it also works on likes, it “finds people who liked the same posts as you, and shows you what else they’ve liked recently.” It checked all our requirement boxes: it is explainable, shows relevant posts to each user, and sometimes surfaces more quiet posters.
Architectural challenges
We first started designing a new “algorithmic” feed in January of 2026. One of the biggest challenges in making a good algorithmic feed besides the algorithmic is performance. We want the feeds to load in a reasonable amount of time for the thousands of users, who use our feeds every week.
After discovering the For You feed, we realized that our current architecture and database design wouldn't support a more complex algorithmic feed that is unique for every user.
We upgraded our database version for the second time since creating Furrylist, stopped storing hundreds of millions of unnecessary likes (which caused an outage), and improved our database indexes, so we can generate the Fur You feed in at most a few seconds.
One challenge remains: if you have recently joined Furrylist or interact with people who did, not all likes and follows are tracked in our system, so some boosting may not work well for everyone. We will address this during Fur You's public beta by backfilling this data.
We are also currently not respecting the so-called content visibility declaration that allows users from opting out from being shown in algorithmic feeds, like Fur You. We also plan to address this by only showing their posts in non-algorithmic feeds (like Hot or reverse chronological feeds). For the moment, we still lack this data too though.
Dealing with NSFW posts
Unlike the large social media companies, we believe in erotica and pornography as a valuable avenue for self-expression. These types of posts are very prevalent and well-liked in the furry community and our feeds. Therefore, we don't want to limit engagement on NSFW posts for people interested in NSFW posts.
We ended up designing a mechanism that only deboosts NSFW posts for people who already don't engage much with that post category. It only applies to people whose likes are composed of less than 10% of NSFW posts.
The deboost is linearly dynamic. To give you a sense of its impact, if only six percent of your liked posts are NSFW posts, posts by people you follow will still be boosted more than posts by people you've never interacted with.
We've love to improve this mechanism with your feedback too, so we can show the most relevant posts to everyone without having to broadly censor types of posts.