How Nairobi's Digital Archives Became Cluttered With Ghost Images — and Why It Now Costs Real Money

Government agencies and public institutions across Nairobi are sitting on terabytes of duplicate image files — the same photographs, ID scans, and infrastructure records stored two, three, sometimes a dozen times across different servers — and the bill for that disorder is climbing. The problem, known in IT circles as duplicate image accumulation, did not arrive overnight. It was built, slowly, through years of fragmented digitisation drives, incompatible procurement cycles, and a bureaucratic culture in which deleting a file felt riskier than keeping it.

The issue matters right now because Kenya is mid-way through an IMF austerity programme that has forced line ministries to justify every shilling of operational expenditure. Cloud storage and server maintenance contracts — previously renewed quietly — are now line items that Treasury officials are scrutinising. Redundant data is redundant cost, and in a fiscal environment where the Ruto administration has already faced a Gen Z-fuelled tax revolt over the Finance Bill, the politics of wasteful digital spending are unusually sharp.

How the Duplication Happened

The roots go back to at least 2018, when several parallel e-government initiatives launched in close succession without a shared data architecture. The Huduma Namba national identity programme, run out of the then-Ministry of Interior offices along Harambee Avenue, collected biometric photographs that were stored locally at registration centres, uploaded to a central server, and — in many cases — re-scanned when applicants returned to correct errors. Each iteration created a new image file. By the time the Supreme Court issued rulings that complicated the programme's rollout, hundreds of thousands of near-identical portrait images were already sitting in multiple repositories.

Separately, county governments operating under the Nairobi City County structure began their own digitisation efforts. Land records from Makadara Sub-County offices, business permit photographs from River Road licensing bureaus, and infrastructure inspection images from the Nairobi Metropolitan Services were each stored in siloed systems. When agencies wanted to share files, the path of least resistance was to duplicate rather than link. A single pothole photograph on Jogoo Road might exist in the Nairobi City County records system, in a Kenya Urban Roads Authority folder, and in an e-mail attachment archive — each a separate file, none referencing the others.

The Silicon Savannah startup ecosystem, centred on Westlands and the Nairobi Garage co-working hub on Lenana Road, has long offered tools to address this. Several Kenyan tech firms, including those that have gone through the Chandaria Business Innovation and Incubation Centre at the University of Nairobi, have built deduplication software tailored to Swahili-language metadata and Kenyan ID formats. The adoption rate inside government, however, has been slow. Procurement rules requiring open tenders, combined with a preference among senior IT officers for established international vendors, have repeatedly sidelined locally built solutions.

The Scale of the Problem

Precise government-wide figures are not publicly available, but the Kenya ICT Authority published a data governance audit framework in 2023 that acknowledged duplicate records as a category-one data quality risk across national systems. Storage costs for enterprise-grade servers in Kenya — whether co-located at the government data centre on Ngong Road or hosted through regional cloud providers with nodes in Nairobi — run broadly between Ksh 8,000 and Ksh 25,000 per terabyte per month depending on redundancy tier, according to publicly available pricing from local managed-service providers. For agencies holding dozens of duplicate terabytes, that is a recurring expense with no operational return.

The informal settlement upgrading programme operating in Mathare and Mukuru kwa Njenga has added a newer dimension. As community land titling and household mapping surveys produce fresh batches of aerial and ground-level images, field workers uploading from mobile devices over patchy 4G connections frequently trigger automatic retry uploads, creating duplicates at the point of capture before files even reach a central server.

Administrators who want to begin correcting the problem have a practical starting point: the Kenya ICT Authority's data governance guidelines already require new procurement contracts to include deduplication protocols. Agencies that can document storage rationalisation efforts before the next Treasury budget cycle — submissions are typically due by September — are better positioned to defend their IT budgets. The technical fix, running automated hash-matching scripts to identify identical files, is not complicated. What has been missing, so far, is the institutional pressure to make someone responsible for pressing the button.