Open Data Project Encodes US Individual Income Tax Forms Line by Line

Overview

A software project on GitHub named `emmilco/us-tax-model` encodes the Internal Revenue Service's individual income tax forms and instructions as open data. Every encoded line includes citations linking back to archived IRS pages. The repository implements tax forms and logic up to the 2024 tax year, while the underlying calculation engine remains active for other IRS tools.

Structure and Verification

The project models tax calculations through a stack of layers designed to verify computations and trace data back to original sources:

• Fact Graph: A declarative knowledge graph written in Scala that tracks known and unknown variables.
• Operations: Calculations rely on a closed list of approximately twenty discrete operations, such as addition, subtraction, rounding, and table lookups, implemented without floating-point arithmetic.
• Proof and Testing: The engine uses two independent evaluators and formal properties proved in Lean 4 to check mathematical consistency.
• Independent Evidence: To address the challenge of reading accuracy, the project uses cross-validation through independent transcriptions, model reviews, and comparisons against third-party tax engines.

Limitations and Practical Implications

The repository operates strictly as an independent analysis and verification tool rather than an official filing service. It does not file tax returns; instead, it generates a values sheet and explicitly refuses to process returns that fall outside its supported scope. Users and developers can inspect the open-source repository on GitHub to review how specific instructions and formulas have been transcribed and tested.