Live data from Hacker News

Twitter's Recommendation Algorithm

blog.twitter.com

481–490 of 1001 posts

Re: Twitter's Recommendation Algorithm

#481
post #291

Earlier quoted context omitted.

NY Times, WaPo, LA Times and other major accounts too https://www.thewrap.com/ny-times-la-times-not-pay-for-twitte...

Seems dumb of them. Cost is trivial and their competition that isn’t so politically motivated will have a much further reach. The smart move would be silent on the policy change, pay, and support rival platforms as they can. Instead they will eventually pay and look like they lost.

It's smart because Musk already blinked

The top 10,000 are getting exemptions and won't have to pay

https://variety.com/2023/digital/news/twitter-business-month...

Re: Twitter's Recommendation Algorithm

#482

Context: I teach at Princeton and study social media and recommendation systems. From a very quick skim of the repositories, this appears to be quite limited transparency. The documentation gives a decent high-level overview of how Tweet recommendation works—no surprises—and the code tracks that roadmap. Those are meaningful positive steps. But the underlying policies and models are almost entirely missing (there are…

Can i audit your classs for free?

Re: Twitter's Recommendation Algorithm

#483

Earlier quoted context omitted.

That would be great (unweighting bluechecks) but they actually plan to go in the other direction: Starting April 15th non-bluechecks won't show up in the "For you" section (the algorithm timeline) at all. Unpaid users are being written completely out of the algo. https://twitter.com/elonmusk/status/1640502698549075972

Inaccurate. Musk stated that people you follow will continue to show up. I only use the 'for you' feed when I'm bored and want stupid dopamine hits, I leave it on Following almost all the time. But that's on desktop, my understanding it that it keeps resetting itself for mobile users (of whom I am not one).

He only said that later; he seems to be working on a "say we'll do feature and then change it all after people yell at him for weeks" process.

Re: Twitter's Recommendation Algorithm

#484

Earlier quoted context omitted.

I thought it was interesting how it explicitly doesn't boost independents. So much of the two-party system is self-reinforcing.

Also how about.... everyone else in the world who is not an American voter?

Only anecdotally.

But ever since Musk took over the amount of US political content has significantly increased in particular from the right despite me not living in the US.

It's hard to tell whether previously political content was weighted less and Musk has removed those controls or whether they are now weighted higher.

Re: Twitter's Recommendation Algorithm

#485
post #290

Earlier quoted context omitted.

Yep, having had created a few throwaway accounts I definitely got a sense of how the algorithm compensated for the majority of users who aren't super active. And it makes sense -- most new users aren't going to want to spend account creation picking 50 accounts to follow. But if someone has hit the follow button 1,000+ times, it's reasonable to have some faith that they've seen a lot of tweets and know what they want…

Is the "following" tab an option for you? Or what would you like the "for you" tab to do? Filter out heavy tweeters? Prioritize more popular tweets?

The Following tab limits to my followers, but in reverse chrono order. This is also not desirable b/c it limits to whoever’s posting whenever I’ve happened to check the feed.

I’d like to see interesting tweets from a few hours ago, and not just Australian tweets when I’m up late at night.

Re: Twitter's Recommendation Algorithm

#486
post #379

Earlier quoted context omitted.

Aside from the spam PRs, there is actually one PR that fixes a bug: https://github.com/twitter/the-algorithm/pull/242/files

Modern Java actually allows `10_000` for that very reason, as does Scala ( https://scala-lang.org/files/archive/spec/2.13/01-lexical-sy... )

also javascript - https://github.com/tc39/proposal-numeric-separator

Re: Twitter's Recommendation Algorithm

#487

Great pull request here which improves the algorithm: https://github.com/twitter/the-algorithm/pull/17

Me feed has lately been full of accounts that have blocked me. Like, I see a tweet from someone unknown, click their profile and it says I'm blocked.

So wonder if some value is wrong in one of those constants. Anyways, the blocking feature is broken..

Re: Twitter's Recommendation Algorithm

#488

Earlier quoted context omitted.

I thought it was interesting how it explicitly doesn't boost independents. So much of the two-party system is self-reinforcing.

Also how about.... everyone else in the world who is not an American voter?

This is what it means to let another country own your social media. Their ideas and memes unconsciously get preferential treatment. This is maybe not a good thing, but it is what it is.

Re: Twitter's Recommendation Algorithm

#489

Earlier quoted context omitted.

Also how about.... everyone else in the world who is not an American voter?

Only anecdotally. But ever since Musk took over the amount of US political content has significantly increased in particular from the right despite me not living in the US. It's hard to tell whether previously political content was weighted less and Musk has removed those controls or whether they are now weighted higher.

Some of the power users Musk reportedly had boosted are specifically right wing political posters, like catturd2. But then he's also boosted some high profile left leaning politicians, so it's not exclusive. It does mean you're more likely to see right wing American political content either way, which has to be annoying for people outside of the US.

https://gizmodo.com/twitter-algorithm-aoc-ben-shapiro-cattur...

Re: Twitter's Recommendation Algorithm

#490
post #421
post #11

From https://github.com/twitter/the-algorithm/blob/7f90d0ca342b92... ( "author_is_elon", candidate => candidate .getOrElse(AuthorIdFeature, None).contains(candidate.getOrElse(DDGStatsElonFeature, 0L))), ( "author_is_power_user", candidate => candidate .getOrElse(AuthorIdFeature, None) .exists(candidate.getOrElse(DDGStatsVitsFeature, Set.empty[Long]).contains)), ( "author_is_democrat", candidate => candidate .getOrEls…

what is vits? private val DarkRequestAnnotation = "clnt/has_dark_request" private val Democrats = "democrats" private val Republicans = "republicans" private val Elon = "elon" private val Vits = "vits"

Very Important Tweeters
Post reply on HN