Blogs
National data strategy must move forward
Registries must be corrected, systems connected, safeguards tested, local-language datasets developed and computing capacity expanded. Government can begin with priority datasets, independent quality audits, named data stewards and interoperability pilots. Practical testing will expose problems faster than prolonged institutional argument, writes Godfrey Mutabazi
Godfrey Mutabazi. (File)
By: Admin ., Journalist @New Vision
____________
OPINION
By Godfrey Mutabazi
According to the recently published National Data Strategy, only 36% of the indicators in Uganda’s Fourth National Development Plan have supporting data. The strategy also identifies fragmented databases, inconsistent standards, weak data sharing, inadequate infrastructure and low public trust. Addressing these weaknesses will determine whether Uganda can harness artificial intelligence effectively.
On October 21, UBOS executive director Chris Mukiza challenged the strategy as unlawful and duplicative. His objection deserves legal clarification but should not halt a programme extending beyond official statistics. The question is whether the strategy is a statistical programme or a wider framework for national data and AI readiness.
Most of the data the strategy deals with is not produced mainly for statistical reporting. It is operational data generated by identity systems, telecommunications, banking, payments, health, taxation, land administration and public services. For AI, that data must be accurate, current, interoperable, secure and machine-readable. Official statistics can be drawn from it, but its uses are much broader.
UBOS does not need to write the technical requirements for AI; they already exist. ISO and the International Electrotechnical Commission publish standards for AI management systems (ISO/IEC 42001), data quality for machine learning (the ISO/IEC 5259 series), risk management and system lifecycles. The International Telecommunication Union sets standards for communications and digital infrastructure, and the IEEE develops engineering standards for intelligent systems. UNESCO and the OECD provide principles for responsible AI, while the US National Institute of Standards and Technology’s AI Risk Management Framework is widely used.
Manufacturers and software developers build these standards into products sold worldwide. Most operational software in Uganda was designed, acquired or deployed without UBOS data guidelines. Uganda’s task is to adopt suitable international standards, adapt them to national circumstances and enforce them through its technical and regulatory institutions. UBOS may prescribe classifications and quality rules for official statistics, but it has no need to rewrite standards for software, data centres or AI systems.
To date, many government ministries, departments and agencies still lack sufficiently accurate, complete, current and interoperable data. This raises a fundamental question: who enforces the standards? Each institution remains accountable for the information it collects. UBOS oversees official-statistics standards; NITA-U enforces technical and interoperability requirements; UCC regulates data communications; the Personal Data Protection Office enforces lawful processing; the Bank of Uganda oversees financial and payment systems; and NIRA operates the national identification register. The central problem in practice is weak coordination, auditing and enforcement across government.
Nothing in the strategy prevents UBOS from improving official statistics or coordinating statistical standards. The more urgent gaps lie elsewhere. NIRA’s own reports have disclosed rejected applications, records not linked to other systems and millions of printed identity cards still unissued. Correcting the register is a job for NIRA and its technical partners. It is no reason to transfer control of national digital infrastructure to UBOS.
UBOS’s concern about duplication merits attention. The strategy assigns responsibility for standards and data quality to both UBOS and the ministry, and that overlap should be resolved. But overlap is a reason to harmonise, not to discard two years of consultation. Cabinet can define the boundaries without stopping implementation.
Registries must be corrected, systems connected, safeguards tested, local-language datasets developed and computing capacity expanded. Government can begin with priority datasets, independent quality audits, named data stewards and interoperability pilots. Practical testing will expose problems faster than prolonged institutional argument.
Manufacturers and software developers build these standards into products sold worldwide. Most operational software in Uganda was designed, acquired or deployed without UBOS data guidelines. Uganda’s task is to adopt suitable international standards, adapt them to national circumstances and enforce them through its technical and regulatory institutions. UBOS may prescribe classifications and quality rules for official statistics, but it has no need to rewrite standards for software, data centres or AI systems.
To date, many government ministries, departments and agencies still lack sufficiently accurate, complete, current and interoperable data. This raises a fundamental question: who enforces the standards? Each institution remains accountable for the information it collects. UBOS oversees official-statistics standards; NITA-U enforces technical and interoperability requirements; UCC regulates data communications; the Personal Data Protection Office enforces lawful processing; the Bank of Uganda oversees financial and payment systems; and NIRA operates the national identification register. The central problem in practice is weak coordination, auditing and enforcement across government.
Nothing in the strategy prevents UBOS from improving official statistics or coordinating statistical standards. The more urgent gaps lie elsewhere. NIRA’s own reports have disclosed rejected applications, records not linked to other systems and millions of printed identity cards still unissued. Correcting the register is a job for NIRA and its technical partners. It is no reason to transfer control of national digital infrastructure to UBOS.
UBOS’s concern about duplication merits attention. The strategy assigns responsibility for standards and data quality to both UBOS and the ministry, and that overlap should be resolved. But overlap is a reason to harmonise, not to discard two years of consultation. Cabinet can define the boundaries without stopping implementation.
Registries must be corrected, systems connected, safeguards tested, local-language datasets developed and computing capacity expanded. Government can begin with priority datasets, independent quality audits, named data stewards and interoperability pilots. Practical testing will expose problems faster than prolonged institutional argument.
The strategy itself still needs corrections. Its cover gives the period as 2026/27 to 2030/31, while the scope section assumes 2025/26 to 2029/30. Many baselines remain “to be determined”. One target appears inverted, calling for fewer than 90% of primary datasets to be error-free when the intention is plainly the reverse. These defects can be corrected administratively.
Financing is the larger danger. The strategy is costed at about sh1.66 trillion over five years, figures that remain assumptions until Parliament converts them into annual appropriations. About a fifth is expected from data services that largely do not yet exist. Government should explain what may be commercialised, how that revenue was estimated and how public-interest and personal data will be protected.
Security funding should be consolidated. The objective covering data protection and security tools shows no dedicated funding, while the infrastructure programme separately allocates sh1b a year to security tools. Both provisions belong under named institutions with measurable outcomes. UBOS cannot take on AI security or personal-data protection when those functions are already assigned elsewhere.
The National Data Office will need real capacity. It is expected to coordinate a programme worth more than sh1.6 trillion without additional staff. Reassigning officers may create an office on paper without the specialists to deliver. An independently verified dashboard should disclose money released, systems connected, datasets cleaned, and security incidents resolved.
A strategy does not become illegal because it extends beyond official statistics, nor does it remove UBOS’s mandate. UBOS remains essential to statistical integrity, while the ministry and technical institutions lead digital and AI readiness. Technology is advancing exponentially while governments’ laws change at a slower pace.
Progress must prevail over institutional territorialism.
The writer is the former executive director of Uganda Communications Commission
Financing is the larger danger. The strategy is costed at about sh1.66 trillion over five years, figures that remain assumptions until Parliament converts them into annual appropriations. About a fifth is expected from data services that largely do not yet exist. Government should explain what may be commercialised, how that revenue was estimated and how public-interest and personal data will be protected.
Security funding should be consolidated. The objective covering data protection and security tools shows no dedicated funding, while the infrastructure programme separately allocates sh1b a year to security tools. Both provisions belong under named institutions with measurable outcomes. UBOS cannot take on AI security or personal-data protection when those functions are already assigned elsewhere.
The National Data Office will need real capacity. It is expected to coordinate a programme worth more than sh1.6 trillion without additional staff. Reassigning officers may create an office on paper without the specialists to deliver. An independently verified dashboard should disclose money released, systems connected, datasets cleaned, and security incidents resolved.
A strategy does not become illegal because it extends beyond official statistics, nor does it remove UBOS’s mandate. UBOS remains essential to statistical integrity, while the ministry and technical institutions lead digital and AI readiness. Technology is advancing exponentially while governments’ laws change at a slower pace.
Progress must prevail over institutional territorialism.
The writer is the former executive director of Uganda Communications Commission