Data Counterfactuals What changes when the training data changes?
Overview Grid Memos Related Areas and Papers
WIP Explorables
Graph Subset-lattice companion that treats dataset edits as graph moves API Request/response style explorer for querying grid and graph state directly 3D Viewer Heavier prototype with the more complex 3D visualization
GitHub Repo
Menu
Explore the site Overview What data counterfactuals are and why they matter Grid Interactive toy grid for exploring train/eval counterfactuals Memos Written notes: the launch post, formalisms, and more Related Areas and Papers Curated shelf of neighboring literatures and representative papers
WIP Explorables Graph Subset-lattice companion that treats dataset edits as graph moves API Request/response style explorer for querying grid and graph state directly 3D Viewer Heavier prototype with the more complex 3D visualization
Project links GitHub Repo ↗
Memo archive

Memos

3 memos
  • March 13, 2026 10 minute read 2244 words

    1. Introducing Data Counterfactuals

    A shared launch post and site memo arguing that many questions about value, data dividends, selection, privacy, poisoning, and data leverage are really questions about what changes when training data change.

  • March 11, 2026 16 minute read 3545 words

    2. Formalisms

    A draft memo lining up several tasks that fit the data counterfactuals frame: influence, valuation, selection, distillation, unlearning, privacy, poisoning, and other neighboring cases.

  • March 11, 2026 6 minute read 1418 words

    3. Glossary

    Working definitions for the key terms used across the memo, explorer, and related areas and papers page.

Explore the site

Overview Grid Memos Related Areas and Papers

WIP Explorables

Graph API 3D Viewer

Project links

GitHub Repo ↗