This repository contains source code for the SafeTab-P disclosure avoidance application. SafeTab-P was used by the Census Bureau for the protection of individual 2020 Census responses in the tabulation and publication of the Detailed Demographic and Housing Characteristics File A (Detailed DHC-A). A later source code release will contain the source code for SafeTab-H, the application used to protect the Detailed Demographic and Housing Characteristics File B (Detailed DHC-B).

Using the mathematical principles of differential privacy, SafeTab-P infused noise into 2020 Census results to create privacy-protected statistics which were used by Census Bureau subject-matter experts to tabulate the 2020 Detailed DHC-A product. SafeTab-P was built on Tumult Analytics, a platform for computing statistics using differential privacy. Both SafeTab-P and the underlying platform are implemented in Python. The latest version of Tumult Analytics can be found at https://tmlt.dev/.

In the interests of both transparency and scientific advancement, the Census Bureau committed to releasing any source code used in creation of products protected by formal privacy guarantees. In the case of the Detailed Demographic and Housing Characteristics publications, this includes code developed under contract by Tumult Labs (https://tmlt.io) and MITRE corporation. Tumult Analytics is an evolving platform and the code in this repository is from version 0.5.3 used in the production of DDHC-A.

The Census Bureau has already separately released the internally developed software for the TopDown Algorithm (TDA) used in production of the 2020 Redistricting and the 2020 Demographic and Housing Characteristics products.

This repository is divided into six subdirectories:

• configs contains the specific configuration files used for the production Detailed DHC-A runs, including privacy-loss budget (PLB) allocations and the rules for adaptive table generation. These configurations reflect decisions by the Census Bureau's Data Stewardship Executive Policy committee based on experiments conducted by Census Bureau staff.
• safetab_p contains the source code for the application itself as used to generate the protected statistics used in production.
• safetab_utils contains utilities common among the SafeTab products developed by Tumult Labs for the Census Bureau.
• mitre/cef_readers contains code by MITRE to read the 2020 Census input files used by the SafeTab applications.
• ctools contains Python utility libraries developed by the Census Bureau's DAS team and used by the MITRE CEF readers.
• tumult contains the Tumult Analytics platform. This is divided into common, analytics, and core directories. The core directory also includes a pre-packaged Python wheel for the core library.

WARNING: the pre-packaged Python wheel in this repository contains a binary version of pyarrow containing the pyarrow vulnerability identified in 2023 (https://nvd.nist.gov/vuln/detail/CVE-2023-47248). The SAFETAB-P code itself does not expose this vulnerability, but modifications should rebuild the wheel from sources beneath the tumult/core prefix or later sources from Tumult itself.