281–290 of 1001 posts
> the main neural network part of @Twitter recsys algo is based on 2021 work of #SinaWeibo - Chinese clone of Twitter
Earlier quoted context omitted.
\* \* These author ID lists are used purely for metrics collection. We track how often we are \* serving Tweets from these authors and how often their tweets are being impressed by users. \* This helps us validate in our A/B experimentation platform that we do not ship changes \* that negatively impacts one group over others. \* From: https://github.com/twitter/the-algorithm/blob/7f90d0ca342b92...
So now engineers working on the algo can ensure their launches won't lower Elon's tweet visibility. Looks like those remaining at Twitter have a knack for corporate survival.
So as expected, there is exactly nothing that favors posters from one side of the political spectrum. I don't expect that this article will do anything to calm down those who are convinced otherwise though. Well written article, from an engineer's perspective.
Well, it does say this: Ranking is achieved with a ~48M parameter neural network that is continuously trained on Tweet interactions to optimize for positive engagement (e.g. Likes, Retweets, and Replies). This ranking mechanism takes into account thousands of features and outputs ten labels to give each Tweet a score, where each label represents the probability of an engagement. We rank the Tweets from these scores.…
"Today, the For You timeline consists of 50% In-Network Tweets and 50% Out-of-Network Tweets on average, though this may vary from user to user." I have spent significant effort creating a network and there you go choosing to ignore my efforts by putting in 50% of crap-I-don't-want-to-see. That is why I despise your algorithm.
I'm confused, then why not just use your 'followed' feed instead of 'for you'?
Great! But nothing is going to change until people realize that the problem is the feedback loop . It's not the recommendation engine itself, it's the fact that there's no way "out" of the feed that the engine produces. It recommends you stuff, you have little choice but to engage with it, and then it trains on that information . This is the problem with most of social media today. It is a very well known problem in…
Earlier quoted context omitted.
I believe LeBron James said recently he isn't going to waste his money on a blue checkmark, so it should be interesting to see what stays and what goes.
LeBron doesn’t get $84 of value from Twitter? Definitely not a political statement going on there.
I wonder what the "author_is_elon", "author_is_power_user", "author_is_democrat", and "author_is_republican" labels are for [1]. [1]: https://github.com/twitter/the-algorithm/blob/main/home-mixe...
Elon is addressing this in the Twitter Space right now. "It definitely shouldn't be dividing people into Republican and Democrats; that makes no sense[...] you've identified something we should be getting rid of right away."
// we only keep unfollows in the past 90 days due to the huge size of this dataset,
// and to prevent permanent "shadow-banning" in the event of accidental unfollows.
// we treat unfollows as less critical than above 4 negative signals, since it deals more with
// interest than health typically, which might change over time.
val unfollows: SCollection[InteractionGraphRawInput] =
GraphUtil
.getSocialGraphFeatures(
readSnapshot(SocialgraphUnfollowsScalaDataset, sc),
FeatureName.NumUnfollows,
endTs)
.filter(_.age > Twitter has several Candidate Sources that we use to retrieve recent and relevant Tweets for a user. For each request, we attempt to extract the best 1500 Tweets from a pool of hundreds of millions through these sources. We find candidates from people you follow (In-Network) and from people you don’t follow (Out-of-Network). > Today, the For You timeline consists of 50% In-Network Tweets and 50% Out-of-Network Twee…
Perhaps if you did follow so many people they got drowned out, but with substantially fewer following, those recommended tweets were a big part of what I saw. Especially in the last year or so before Musk took over: Twitter went a lot more aggressive and didn't just show tweets which people you follow "liked", but also other tweets, which the algorithm somehow determined you might like, which was often wrong, and, mo…
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. Showing a few out-of-network tweets seems reasonable (I got enough as it is through followings' retweets). But 50% of a feed that already can't fit tweets from thousands of followings just feels like shit.
The worst part is that the share of in-network tweets seems to be highly concentrated to the last 10 or so people I most recently interacted with, e.g. seeing the same user over and over just because I liked one of their tweets the other day. Which makes sense to save on computation costs, but it's pushed me into a much tighter bubble than I ever had when the timeline wasn't so out-of-network focused.