Please let me know because my non-dev friends (some are devs from trash companies) are telling me we can finally figure out why *insert opposite side* is being upranked and our side is unfairly being downranked. Looking forward to forking it and submitting pull requests that improve the Twitter algo and build trust for everyone in a fair, decentralized way.
😂😂😂 you think open sourcing the algo will fix human nature?
at least it’ll make it more clear to the world why I’m being paid 20x what these dumbasses are being paid
Open sourcing whatever does not go both ways. The fact they release their algo inner workings doesn’t mean you can send your changes in. Android is open sourced, yet you cannot submit changes to the android OS that google licenses to all phone makers including Pixels. You can use the open sourced algo in your own social media platform if you have one though. Good luck 👍
Maybe we end up discovering it was 500k lines of if..else all along
When did blind get so bad at recognizing sarcasm? Literally all of u idiots so far not realizing it in this post..
When did you get so bad at distinguishing sarcasm from plain dumb people?
Read it again. And again if u need to.
There are biases everywhere. This post is biased with incomprehensibility of ML and limited definition of open source. ML and relevance is biased towards many simple factors and coefficients that can be simplified and shared in public domain.
All jokes aside, I would like to understand this more, seeing as Jack Dorsey wants to make twitter function at the protocol instead of company level. If anyone works on ranking teams, what are your opinions on whether open sourcing your codebase is a good idea? What are the pros and cons? And assuming you have to do it, what would that exactly entail, where do you draw the line on what gets open sourced? For example, are you going to have a readme of how you log and collect data, how and where you store this data, how your company-specific infra for building and evaluating models works? Or do you somehow try to expose all this too? Also my understanding of the purpose of open sourcing something is to let others use it + let them develop on top of it and improve it. Is that even feasible or is the metrics being optimized too Twitter-specific and useless to others? I also don’t understand how anyone who doesn’t have access to Twitter’s infra can even develop on top of it.
Do you know how machine learning works, Meta?
I’m sure I can just tweak the coefficients, submit a pr, and hook it up to your infra and start running experiments on users
Think it’s a formula like (.7*liberal +.3*conservative)*comments/100