Declarative modernity

In thermodynamics, entropy describes the tendency of any closed system toward disorder. But there is a subtler phenomenon than physical entropy: narrative entropy, the progressive degradation between what is declared and what is executed until the two stop referring to the same object. The Dominican Republic has been at it for al…

In thermodynamics, entropy describes the tendency of every closed system toward disorder. But there is a phenomenon subtler than physical entropy: narrative entropy, the progressive degradation between what is declared and what is executed until the two no longer refer to the same object. The Dominican Republic has been operating under that invisible law for at least a decade. The February 2026 announcements are its most recent expression.

President Abinader announced an agreement with NVIDIA to create a Center of Excellence in Artificial Intelligence. ITLA was presented as a strategic axis of national technological development, with 12 campuses, more than 24,000 young people reached, and a recent graduation of more than a thousand professionals in cybersecurity and telecommunications. In parallel, Minister Santos Badía spoke of an educational revolution, of structurally merging MINERD and MESCYT, and of training skilled labor for tourism in Pedernales. The GSMA launched Open Telco AI to accelerate sector-specific artificial intelligence models. International experts urged regulating AI to protect democracy. All of it in a single week.

Read in isolation, each headline projects strategic direction. Read in historical context, it confirms a pattern I have documented since 2018: symbolic industrialization. A model where the institutional gesture precedes —and often replaces— installed capacity. Where the framework agreement is communicated as a result, enrollment as transformation, and visible infrastructure as evidence of innovation.

The NVIDIA case is illustrative. An agreement with one of the most valuable corporations on the planet generates immediate symbolic capital. But the semiconductor industry, which the government ties to this narrative, is not built with educational cooperation agreements. It requires R&D ecosystems with sustained investment over decades, proprietary intellectual property, structured tax incentives, a critical mass of engineering PhDs, and regionally integrated supply chains. The National Strategy for the Promotion of the Semiconductor Industry (ENFIS) exists as an instrument of intent. What is missing are the verifiable productive assets that give it substance. A diplomatic-technological instrument has value. Mistaking it for consolidated industrial policy has a cost.

ITLA deserves recognition for its territorial expansion and for widening access to technical training in historically excluded areas. That is real. However, access and productive absorption are different variables that operate on different time scales. If graduates of technology programs do not find formal placement in knowledge-intensive sectors within the Dominican productive system, the country is financing the training of talent that other markets monetize. This is the phenomenon of The invisibles: local professionals who live in the country but work remotely for foreign economies, boosting the GDP of other jurisdictions without fully paying taxes in their own. An educational revolution that does not come with productive transformation generates individual mobility, rarely structural transformation.

The proposal to merge the ministries of education and higher education introduces an additional variable. The emphasis on training labor for tourism in the south of the country shows sectoral coherence with an economy that depends on services. But it raises a long-term tension: an education policy designed for immediate labor absorption in traditional sectors is unlikely to produce, at the same time, the capabilities that a knowledge-based economy demands. César Hidalgo has documented it with data: a country's economic complexity depends on the diversity of combinable productive capabilities, not on the volume of workers trained in a single sector. Tourism training is necessary. Training in artificial intelligence, model design, data engineering, and advanced computer science is strategic. The question is whether the system can do both at once or whether sectoral urgency displaces investment in complexity.

The hard data do not support the narrative of progress. The sustained decline in the Global Innovation Index and the Latin American Artificial Intelligence Index is a documented fact that requires no repetition, but a response. The new editions of both indices are published this year and offer a concrete opportunity: to show that the 2024 and 2025 announcements translated into measurable capacity. If the results again show a relative drop while the narrative keeps projecting leadership, the gap between discourse and structure will stop being an analytical diagnosis and become a political fact. The ratio between consumption of imported solutions and investment in local development continues to describe a dependent digital economy, inserted in the periphery of the global value chain. Structures are not changed by headlines.

Langdon Winner wrote that artifacts have politics. Technologies are not neutral: they embed power relations and define who produces and who consumes, who designs and who implements. An AI Center of Excellence that transfers knowledge into the country has potential value. But if the transfer model does not go beyond the demonstration layer, what is produced is sophisticated consumption of foreign technology with the aesthetics of regional leadership. Consuming technology can be a stage; depending on it permanently is the problem. The difference between real and symbolic transfer is not one of degree; it is one of nature. It is shown with verifiable assets: patents, utility models, sovereign datasets, and code contributed to reference projects.

There are three levels of technology transfer. The first produces aesthetics: certifications, demonstration labs, institutional visits. The second produces operation: applied projects with local data deployed in production. The third produces sovereignty: registered intellectual property, contribution to international standards, proprietary computing capacity, and talent retained in the national productive system. The history of technological development teaches that Centers of Excellence are validated by their movement toward the third level, not by their inauguration.

If after three years the Center of Excellence cannot show those metrics, it will have served as a public relations platform, not as an engine of development.

The low digital engagement of the news stories tied to these announcements is a secondary indicator, but a relevant one. The communication pieces generated between zero and five interactions on social media. For a narrative that seeks to project confidence to international investors, the density of domestic public conversation is part of the signal. When the institutional story generates no organic conversation, the effect is more declarative than mobilizing.

The GSMA's launch of Open Telco AI adds an important global nuance. The telecommunications sector faces structural stagnation: five years of growth below that of other technology verticals. AI is its lever for reinvention. But the GSMA itself acknowledges that current models do not meet the specific needs of networks. This forces the development of sector-specific capabilities. For the Dominican Republic, where internet penetration reaches 85% but the usage gap exceeds 40%, adopting AI in telecommunications can improve operational efficiency without transforming the productive fabric. Connectivity without capacity for social monetization is installed infrastructure with no systemic return. Without robust data centers, without energy policies aligned with high-density computing loads, without public APIs accessible to SMEs, the country will keep operating on foreign infrastructure. Dependence ceases to be merely technological and becomes strategic.

There is a dimension rarely discussed in the local debate: AI requires stable energy, clear regulatory frameworks, and sovereign computing capacity. The projection that AI will represent 2.2% of global GDP by 2030 is not an automatic guarantee of value capture. A country can consume AI without producing it, host data without designing models, or specialize in specific verticals where it has a comparative advantage. Capture depends on positioning in the value chain, and positioning is built with cumulative investment, not with periodic announcements.

What connects this week's headlines with the analysis of the past four years is a constant: the gap between technological discourse and systemic capacity. I have called this pattern symbolic industrialization because it describes a concrete mechanism, not a metaphor: chip factories are announced without local design capacity, centers of excellence are presented without sovereign computing infrastructure, an educational revolution is declared without metrics of job placement in high-value sectors.

Symbolic industrialization operates as a low-cost political circuit: the announcement produces coverage, coverage produces institutional legitimacy, and legitimacy captures agenda and budget. If that cycle is not anchored to verifiable series, execution becomes unnecessary because the political return has already been extracted. Narrative entropy is not an absence of information; it is a loss of comparability over time. When everything is a milestone, there is no trajectory. Without time series there is no public policy; there are events.

The cycle reproduces itself because incentives sustain it: electoral logic rewards what is visible over what is cumulative, international cooperation and large vendors reward branded projects over technological sovereignty, and institutional fragmentation lets each entity optimize for its own headline without being accountable for the system.

Declarative modernity works as a geopolitical signal. It lowers perceived risk and projects strategic direction. But when the signal is not accompanied by public, periodic, and auditable indicators, it loses signaling power with the actors who really matter.

The opportunity is to institutionalize a monitoring system that links every major technology announcement to a public dashboard: committed versus executed investment, talent trained versus talent placed in jobs, digital services exports, registered intellectual property, installed computing infrastructure capacity, associated energy consumption. A National Technological Capacity Index with verifiable dimensions: productive capacity, strategic infrastructure, retained human capital, a coherent regulatory framework, effective inclusion. Quarterly publication, open data, auditable by academia and the private sector. Without that minimum governance, the index becomes just another announcement.

Our diagnosis can be dismantled with facts. It would be enough to execute a multi-year computing infrastructure budget and publish its quarterly execution. It would be enough to measure local job placement of technology graduates with salary traceability. It would be enough to produce an official series of cybercrimes with classification and trend. If those data series appear, the country would stop communicating modernity: it would be accumulating it.

The Dominican Republic shows strategic intent. Intent is not execution, just as enrollment is not absorption, an agreement is not a result, and an announcement is not cumulative policy. The challenge for 2026 is not to communicate more. It is to measure better. Without rigorous measurement, the technological revolution remains a recurring promise. With consistent metrics, it can become State policy with verifiable impact on human development. Narrative entropy is fought with verifiable information, and verifiable information is built with institutions that are accountable, not with headlines that stand in for them.