In a significant move towards transparency, X, formerly known as Twitter, is making its core ranking and filtering algorithms open source. This unprecedented step aims to demystify how content is presented to users, particularly within the crucial 'For You' timeline, which appears when you first open the app. The company is releasing the source code for this default feed, along with its central ranking engine, onto GitHub under the permissive Apache v2 license.
This expansion of X's open-source effo rts is substantial, increasing the codebase size by roughly 10 to 15 times. It now includes detailed model configurations, filter mechanisms, and the intricacies of the core ranking system. This means parameters used to weigh various signals, which dictate which posts get displayed, are now publicly accessible. X's VP of Product, Keith Coleman, highlighted that users will be able to see the code responsible for pulling and ranking posts for individual users, as well as systems that filter out content violating their rules. Notably, some of these systems, like the ranker and scoring mechanisms, can even be run independently by the public outside of X's infrastructure.
Beyond just releasing the code, X is also introducing a new transparency tool. This feature, accessible via an 'Under the Hood' section in the app's settings, will allow users who have posted at least 10 times in the past month to download their aggregate statistics as a JSON file. This file will detail if their a ccount or specific posts have been subject to any labels or impacts from X's ranking systems over the last month. The company is even open to considering pull requests from the community, envisioning a future where the X algorithm is not only visible but also potentially improved by the public, a concept Coleman described as 'amazing' and 'cool'.
Fuente Original: https://slashdot.org/story/26/08/16/0223258/x-open-sources-its-ranking-and-filtering-algorithms?utm_source=rss1.0mainlinkanon&utm_medium=feed
Artículos relacionados de LaRebelión:
- Capital Ones AI Strategy Open Models Custom Data Power
- Nvidia Embraces Open-Source AI to Boost Spending
- Nvidia Launches Its First Open-Source AI Model
- Claude Mythos 5 Puerta trasera en proyecto open source
- Valve Funds Open-Source RADV Driver for Windows
Artículo generado mediante LaRebelionBOT
No hay comentarios:
Publicar un comentario