Surface Data Commons

This is a set of Blog Notes for the first blog post regarding the Surface Data Commons.

Key Themes:

• Building a machine learning data commons.
• Roadmap to list Data Commons Governance strategy.
• Which cooperative pattern works best.

Relevant writings:

• Whose Language Counts as High Quality? Measuring Language Ideologies in Text Data Selection: Showcases the data imbalances regarding linguistic and cultural diversity in datasets used for major popular models. This prompts better data curration centered around user needs and desires.
• Decentralised Autonomous Organisations (DAOs) as Data Trusts: A general-purpose data governance framework for decentralised data ownership, storage, and utilisation: How DAOs can give pools of users more control and ownership over the data they create.
• A Speculative Sketch of a DAO, with Open Collective: One way a DAO and a Cooperative can structure itself for effective governance.
• ML and NLP Research Highlights of 2021: Just a sample of NLP improvements over 2021.
• How Many Data Points is a Prompt Worth?: Good motivation for building prompt based datasets.