WikiCred Lab Notebook
To substantially improve verifiability on Wikipedia, we need to improve the capacity and quality of existing processes, practices, tools, and initiatives with a sense of urgency, but also with a view towards purpose and efficiency.
This is why we have begun the WikiCred Lab Notebook: an effort to marry the reflection and rigor around tracking and validating these efforts, while also recognizing the distributed and creative ways that Wikipedians work.
Overview
[edit]When readers land on an article, it needs to be a reliable source of information. Reliability across Wikimedia is under pressure in several ways, which is what the WikiCred community and others have sought to consider and address, including:
- Backlogs of low-quality content: Fabricated citations, AI-written text or COI/UPE content slipping past review, and fewer readers reaching the encyclopedia directly are some of the challenges posed by generative AI. Meanwhile, government sources once thought to be reliable are disappearing and degrading.
- Coordinated manipulation: Disinformation actors and state-linked campaigns treating Wikipedia as a target, which requires sustained contact with Foundation legal and trust & safety teams.
- Uncoordinated learnings and extended volunteers: In contrast, many efforts exist to try and support Wikipedia’s reliability, but these efforts are dependent upon the goodwill and focus of those distributed around the world and across time zones, with both volunteers and staff repeating and re-learning insights and advances. All the while more poor content comes in every day, and core volunteers are overwhelmed.
What WikiCred Lab Notebook hopes to help do is to make sure that Wikipedia - our free, beautiful, deeply human encyclopedia – remains worth the visit.