Five Decades of Government Software, None Compatible

Federal, state, and local government agencies operate IT systems built across five decades using incompatible data formats, schemas, and identifiers, preventing cross-agency data exchange even when legally mandated. The US federal government spends over $100 billion annually on IT, with approximately 80% devoted to maintaining legacy systems — some running COBOL on mainframes deployed in the 1970s. These systems use agency-specific data schemas for citizens, addresses, transactions, and case records that cannot be mapped to each other without custom point-to-point integrations. A single individual may have incompatible records across Social Security, IRS, VA, Medicare, state DMV, county courts, and municipal services — with no shared identifier, no shared address format, and no shared data model.

Cross-agency data exchange failures directly harm citizens. Veterans eligible for multiple federal programs must re-enter identical information at each agency because VA, DoD, SSA, and HHS cannot share records. Disaster response is delayed because FEMA cannot automatically cross-reference damage assessments with SBA loan applications, IRS income data, and HUD housing records. Fraud detection suffers because siloed data prevents agencies from detecting individuals receiving conflicting benefits across programs. The European Commission estimated that poor interoperability costs EU member states €24 billion annually in duplicated data collection, manual reconciliation, and delayed service delivery.

Enterprise data standards (NIEM — National Information Exchange Model) define common data vocabularies but adoption is voluntary and incomplete. The US Federal Data Strategy (2020) mandated data governance practices but provided no enforcement mechanism or technical infrastructure. API-based integration layers allow modern systems to communicate but cannot extract data from legacy mainframe systems without expensive custom adapters. Cloud migration programs (FedRAMP) modernize hosting but don't address data schema incompatibility — agencies move incompatible systems to the cloud, where they remain incompatible. Estonia's X-Road provides a successful model of government data interoperability but was built on a small population with unified digital identity — conditions not replicable in larger, federated systems.

Lightweight data exchange standards that translate between agency-specific schemas at the boundary — analogous to how HL7 FHIR enables healthcare data exchange without requiring hospitals to restructure internal systems. Canonical data models for the most common government data objects (person, address, case, payment) that agencies map to without replacing internal systems. Shared identity resolution services that probabilistically link records across agencies without requiring a universal identifier. Incremental modernization strategies that wrap legacy systems with API layers rather than requiring full replacement.

A team could select a specific cross-agency data exchange scenario (e.g., veteran benefits enrollment across VA, DoD, and SSA) and map the data schemas used by each agency for overlapping data elements, documenting where schemas diverge and what minimum translation layer would enable basic data exchange. A data engineering team could prototype an identity resolution service that probabilistically matches records across two simulated agency databases using name, address, and date-of-birth variations. Relevant disciplines: data engineering, public administration, systems engineering, information science.