The Twitter :marseybluecheck: ranking score

Taken from the source code

85
Jump in the discussion.

No email address required.

Reported by:

lol reporting tweets de-rank them

:#marseychudnotes:

Jump in the discussion.

No email address required.

presumably just for you though? the variable name implies that its whether youve clicked report or not

Jump in the discussion.

No email address required.

foiled :marseyitsover:

Jump in the discussion.

No email address required.

Shadowbanning twitter users by reportmaxxing :marseylaugh:

Jump in the discussion.

No email address required.

Did they just pull those weighting coefficients out of their butt?

Jump in the discussion.

No email address required.

I always forget how these "algorithms" are all just arbitrary shit someone decided on by tweaking shit until it "felt" right. The way they describe them is so scientific or researched when in reality they're just some dude throwing ideas at the wall and adding on random functions until it's some kronenberg monster of cobbled together parts

Jump in the discussion.

No email address required.

Obviously, what else can you do

Jump in the discussion.

No email address required.

Train a NN or something? Idk nigg

Jump in the discussion.

No email address required.

that's just making the machine pull it out of it's butt instead of a home grown human butt.


Give me your money and I'll annoy people with it :space: https://i.rdrama.net/images/16965516366194396.webp

Jump in the discussion.

No email address required.

Even if you did you'd probably want to distill the results to a simple formula rather than do a full model inference every time the ranking score is calculated.

Jump in the discussion.

No email address required.

sounds like a lot of effort

Jump in the discussion.

No email address required.

With a reward function that's based on the same arbitrary gut feeling about what the results ought to be? That's the same thing but more work, and you can't even make a predictive model

Jump in the discussion.

No email address required.

er

Jump in the discussion.

No email address required.

er? er what? speak up, don't stutter.

Jump in the discussion.

No email address required.

N I G G E R

Jump in the discussion.

No email address required.

:#marseyill:

Jump in the discussion.

No email address required.

Still more reliable than most current research :marseysmirk:

Jump in the discussion.

No email address required.

They're probably relational to each other and have been slightly tweaked if anything didn't work right. Same as bideo game balance.

The exact values probably don't matter very much.

Jump in the discussion.

No email address required.

What determines whether something “is negative feedback”? How can rDrama harness this power?


https://i.rdrama.net/images/17235685217415228.webp

Jump in the discussion.

No email address required.

It's when you click the "show less often" button id guess.

Jump in the discussion.

No email address required.

Replay with "Ratio"

Jump in the discussion.

No email address required.

Ask chat gpt

Jump in the discussion.

No email address required.

What a shit unreadable lang

Jump in the discussion.

No email address required.

The engagement coefficient probably implies that twitter is vulnerable to an attack where you promote someone else's tweet to entirely the wrong audience, thereby driving down engagement and drive up "negative feedback". This kills the tweet in the algorithm and sends its poster to the shadow realm.

Bad faith actors take note!

Jump in the discussion.

No email address required.

:#marseydeux:

Jump in the discussion.

No email address required.

$44 billion algorithm

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.