The Twitter algorithm is now officially public.

[King Possum]

It can be found here: https://github.com/twitter/the-algorithm

The only programming language I know is G&M code so this is all meaningless to me, but I look forward to seeing some analysis.

 

[Breadbassket]

I can't really understand this but I look forward people to dissecting this and trying to game the algorithm as a result.

 

[Make Anime Illegal]

I don't think we can know too much about the "trust and safety" shit without knowing what data it was trained on unfortunately.

 

[King Possum]

This was found rather quickly.

Claimed to just be for metrics, and Elon says it's getting removed ASAP in a space.

 

[Sneed or Death]

Besides the established userbase and the pre-existing relationship with advertisers, isn't the algorithm the entire business model? I'm not that tech literate but this seems like a bad business move, no?

 

[Finger Pistols]

I'll wait paitiently for the kiwis with more technicall knowhow to do their analisis and either laugh with them on how incompetently it is writen or be interested in it's autistic machinations.

 

[Argrabius]

Besides the established userbase and the pre-existing relationship with advertisers, isn't the algorithm the entire business model? I'm not that tech literate but this seems like a bad business move, no?

I mean your putting aside their entire idea of brand, no one can make a copy of twitter and reach any amount of success close to them. releasing this could be great for PR with transparency

 

[Octavia]

Based by Elon tbh, release a fake algorithm to fool the masses !!!

 

[Maldavius Figtree]

Besides the established userbase and the pre-existing relationship with advertisers, isn't the algorithm the entire business model? I'm not that tech literate but this seems like a bad business move, no?

Writing a simple Twitter clone takes a single intermediate programmer and a weekend at most. The value is not in the code.

 

[Looney Troons]

Neat. Cannot wait to see this get fully dissected.

 

[AskeDead]

very interesting. following the thread to see what comes of it

 

[DiggieSmalls]

I'm not learning scala to understand this. Glad its public though.

 

[Slav Power]

Neat. Cannot wait to see this get fully dissected.

I can't wait for people to cry that now the alt-right nazis will abuse it to manipulate the algo to serve nazi propaganda and that Twitter has to be shut down permanently.

 

[Manul Otocolobus]

Unfortunately the majority is written in a language I am not familiar with...

 

[moocow]

Scala, Java and Rust. Why am I not surprised?

 

[wrangled tard]

Scala, Java and Rust. Why am I not surprised?

It was an intentional move to make sure only people with the stripiest of socks will understand. Common tactic among programmer "women".

 

[mush mouth bong bitch]

If you look through the Trust and Safety training models and called models, trust and safety models only appear to have been trained (and run) on English tweets and emojis.

 

[moocow]

If you look through the Trust and Safety training models and called models, trust and safety models only appear to have been trained (and run) on English tweets and emojis.

That's not even remotely surprising. If the kind of people who fritter about trying to police harmless speech and constantly scream about "toxicity" ever knew what's said about them in the safety of other languages, they'd cry themselves to death.

 

[AgendaPoster]

Meh

 

[mush mouth bong bitch]

That's not even remotely surprising. If the kind of people who fritter about trying to police harmless speech and constantly scream about "toxicity" ever knew what's said about them in the safety of other languages, they'd cry themselves to death.

Right, just interesting to see it spelled out in the code. It also indicates that they did the training themselves on internal Twitter data since multilingual toxicity models for bert exist