Between 1996 and 2024, the U.S. Congress allocated approximately $464 million to a federal program designed to finance political subversion operations targeting the Cuban political system. An additional $986 million—since 1984—has funded Radio and TV Martí operations, whose content now circulates predominantly through Internet platforms and social media. For fiscal year 2025 alone, the State Department requested $20 million to promote the "free flow of information," with more than $5.6 million directed to support "independent press" initiatives. Congress appropriated $105 million in total for both programs in FY2025 and FY2026, even amid significant disruption in foreign assistance programs.
These figures are dispersed across multiple U.S. government databases—Foggy Bottom´s Congressional Budget Justifications, final appropriations acts, OMB´s apportionments data, federal audit reports available in the Federal Audit Clearinghouse, and expenditure tracking systems such as USASpending.gov and ForeignAssistance.gov. No coherent, consolidated dataset currently maps where this funding flows, which organizations receive it, or what projects they implement. The technical challenge is straightforward: consolidate fragmented public data into a single, queryable architecture.
I have proposed a technoscientific, theoretical-methodological conception organized in five operational phases, integrating technical rigor, systematic data analysis, and users´ input. The first phase develops automated scripts to extract information from each government database via APIs or bulk CSV downloads on regular schedules. The second phase normalizes raw data into a canonical format—parsing inconsistent date formats, standardizing organization names, validating data types, and handling missing fields with intelligent fallback logic.
With clean, standardized data in place, the third phase transitions from backend infrastructure to direct user-layer intervention. This involves developing, testing, and deploying a browser extension that alerts users whenever they visit a website with verified ties to U.S. federal funding—both current and historical. I have built a functional prototype and documented its architecture in a forthcoming scientific article. The extension reflects funding connections without making claims about content accuracy. Alongside it, detailed, substantiated profiles will be generated for every project and organization in the database, providing contextual information that powers the extension's alerts. Here, social scientists take the lead: tracing organizational relationships, identifying redacted names in public records where possible, and reconstructing funding chains that raw data alone cannot reveal.
Simultaneously, the strategy incorporates ordinary Cuban internet users through a citizen science pilot program. Users can flag potential disinformation operations and validate or challenge the extension's alerts—a mechanism leveraging what is known as "the wisdom of crowds." This distributes agency across the population rather than concentrating it in institutional actors. On the educational front, interactive guides, infographics, and gamification modules will be embedded into a website, allowing everyone to query the central database and reports in real time, building digital literacy and critical thinking skills.
The fourth phase focuses on consolidation and sustainability. The database and website are optimized for performance and user experience. A permanent working group is established with media representatives to help them leverage the platform's data and analyses. Innovation events and hackathons develop new tools on top of the existing infrastructure.
Most critically, the fifth phase establishes postgraduate training programs and elective courses in computational social science applied to cybersecurity for computer science and engineering students. This activates a continuous knowledge-action cycle: research generates inputs for prevention and intervention tools; the use and evaluation of those tools feed insights back into research. By anchoring this cycle in active user participation, the strategy transcends the limitations of information blocking—a passive, exhaustively demanding measure—toward something fundamentally different: proactive, evidence-based sociotechnical resilience that engages citizens as both researchers and subjects of their own defense.
Source for the cover image, generated with GPT technology.
