Please show me more tweets from the people I follow 🙏
@karpathy @kuroke01 @gallabytes We need a complete overhaul of the algorithm
Gauntlet AI founder Austen Allred stepped into a fresh round of complaints after Elon Musk called for a full X algorithm overhaul, specifically asking that content from accounts people actually follow gets lifted above the current recommendation-heavy mix that often buries it.
Please show me more tweets from the people I follow 🙏
@karpathy @kuroke01 @gallabytes We need a complete overhaul of the algorithm
Users such as creator sophie recalled earlier versions of the platform delivering hyper-personalized feeds full of relevant, high-quality exchanges that sparked more posting and genuine connection, a contrast to today's algorithmic flow.
Executive Nikita Bier noted that older X versions actively surfaced content liked by followed accounts, while the present system appears to do this only implicitly and at lower volume, often overwhelmed by other ranking factors.
Many users praise X's custom timelines for surfacing desired community content, while others criticize the algorithm for favoring outrage and ruining personalized feeds.
No Digg Deeper questions have been answered for this story yet.
In the old algorithm, we used to have an explicit recommendation source for this: content your friends like.
The current algo is smart enough to serve that content implicitly—but I think overall frequency of friend-of-friend content is lower (or gets drowned out during video slop waterfalls).
Sharing this feedback with the algo team
what i miss about older versions of X was how much the timeline felt like a hyper personalized community. i was excited to log on because id see the people i interacted with and the amount of cool thoughts per session was orders of magnitude higher, which was inspiration to post
@netcapgirl Yes, we need to get better at showing posts that contribute to network durability and not just optimize for consumption.
Great content will engage you, but emotional bonds is what keeps you.
amazing. i shared in another reply that maybe it’s overindexed on the “what” vs the “who” which to me has always been the magic of the site. it used to really know what accounts i loved and would be more likely to show me their post with 6 likes than one i wasn’t interested in with 60k. this experience was crucial to how i started posting - i was consistently inspired by what i saw on the timeline and genuinely wanted to be a part of it

We've prototyped a million versions of sliders, prompts, etc. (Credit to @benjitaylor for this one)
However, some issues we haven't reconciled:
A. It's unclear that algo preferences can be expressed through toggles
B. Even if we solved (A)—i.e., the user could perfectly express their preferences to us—could X ingest that to give them what they asked for? Timelines are complex
C. Is there any context where we override a user's preferences? For example, if they opted out of news or politics but there's time-sensitive information about an earthquake or terrorist attack.
Having said all of this, we launched Custom Timelines and Snooze Topics and they've been great features for power users.

@nikitabier @netcapgirl I have 12,000 angry bees on my "For You" page demanding i go outside.

@nikitabier @netcapgirl アルゴリズムに囚われる前提を疑え 一度アルゴリズムを捨ててXがどうなるか試して それがあなたの世界を変えるきっかけになるかもしれない

@nikitabier @netcapgirl I dont feel like my followers see my content. Even when I get a bunch of likes, it seems to come from strangers. The net effect is that my posting gets optimized for the enjoyment of the masses instead of the people who know me or my content. More of a generic experience

@nikitabier @netcapgirl I thought you owned the “algo part”. You’re just all talks - and horrible at execution.
take rest, let someone qualified handle the algo part.

@netcapgirl If you like one thing that’s off topic you it ruins the whole algo and get spammed with everything
I had to make a separate burner just to look at sports just bc my whole feed here became sports related

@MickeySteamboat @nikitabier @netcapgirl it does add some friction, ideally you want it just to work right off the bat.
it's a hard problem of how to serve people what they want in the perfect proportions.

@Austen Also, once I interact with a post, they don't have to fill the entire timeline with those category

@netcapgirl e/acc turned out to be grifters or crypto scammers or both.
look at this slop crash out

@Austen there is an entire feed dedicated to that purpose.🤔

@Austen There is literally a tab called "following" and no one is using it.

@nikitabier @netcapgirl Oh but you don’t work on it and it’s it your fault and they don’t report to you . @elonmusk you hired the internets biggest scammer - see comments 🧵

amazing. i shared in another reply that maybe it’s overindexed on the “what” vs the “who” which to me has always been the magic of the site. it used to really know what accounts i loved and would be more likely to show me their post with 6 likes than one i wasn’t interested in with 60k. this experience was crucial to how i started posting - i was consistently inspired by what i saw on the timeline and genuinely wanted to be a part of it

@Austen Instead of for you go to following. Hope this helps.

@nikitabier @netcapgirl maybe it's time to offer different algo flavors, a little bit of customization, for you and following is weak (following for me is flooded with reposts)
people love customization sliders & diff presets

@shacrw_ thanks sepoy for clarifying on his behalf.

X used to be a place where real relationships were built. People who became friends here sometimes met in real life, and became friends there.
Now it feels lost. We don’t see the posts from people we chose to follow and interact with, instead we see a flood of posts we are not really interested in.
This happens just because we liked a post on certain topic, or because the algorithm thinks it knows our interests and preferences better than we do.
By following and interacting with certain people, we made a conscious decision about what we want to see and engage with.
Great to hear Nikita that you listen to the feedback. Something got really lost, would be fantastic to get it back.

@nikitabier @netcapgirl extremely appreciate this, it's probably the most important change you could make imo