Analysis

Open source or proprietary? How governments should choose software for public infrastructure

Analysis · Published 2026-10-16

Close-up of a monitor showing lines of colour-highlighted source code
Photo: Ilya Pavlov, CC0, via Wikimedia Commons.

Few debates in government technology generate as much heat as open source versus proprietary software. Advocates of open source point to sovereignty, transparency and reuse; vendors point to support, accountability and maturity. For digital public infrastructure, the right answer is rarely ideological. It depends on what is being bought, for how long, and what the state can manage itself.

First, separate three questions

Discussions often blur three distinct choices:

  1. Standards: are the interfaces and data formats open? This is usually the most important question for interoperability and exit, and it applies whatever the software licence. See interoperability and open standards.
  2. Licence: is the code published under an open-source licence that allows use, study, modification and redistribution?
  3. Delivery: who builds, hosts, operates and supports the system: in-house teams, a system integrator, or a software vendor?

Open-source software is often delivered by commercial integrators, and proprietary software can implement open standards. The worst combination for public infrastructure is a proprietary product with proprietary interfaces and data formats, operated by the vendor with little in-house understanding.

What public policies say

Several governments have formalised a preference for openness, with different degrees of force:

Civil society campaigns such as the Free Software Foundation Europe's "Public Money? Public Code!" (launched in 2017) argue that publicly funded software should be published under free licences by default.

The case for open source in public infrastructure

The case for proprietary or commercial products

Cost: compare totals, not licences

"Open source is free" is the weakest argument for it. The cost of a public system is dominated by design, integration, hosting, operations, security, training and support over ten years or more. Open source removes licence fees and can reduce long-term costs through reuse and competition between suppliers; it does not remove the need to fund maintenance. The Log4Shell vulnerability in December 2021, in a widely used open-source logging library, showed how much critical infrastructure depends on under-resourced projects. Germany's Sovereign Tech Fund, created in 2022 to invest in open-source maintenance, is one response.

Security: neither model is automatically safer

Open code allows more people to find vulnerabilities, but only if someone looks. Closed code hides vulnerabilities from attackers and defenders alike. Security depends on practices: secure development, independent audits, rapid patching, dependency management and incident response. Contracts and procurement rules should require these, whatever the licence.

A decision framework

ComponentDefaultWhy
Core public rails (identity, exchange, registries)Open standards, state-owned or open-source code, rights to data and documentationLong life, high lock-in risk, public accountability.
Specialised components (biometric matching, HSMs)Best available product behind standard interfacesPerformance matters; interfaces preserve exit.
Commodity IT (office, email, cloud)Competitive procurement, open formats, portability clausesMature markets; avoid format and data lock-in.
Custom applications built for the statePublish code unless there is a specific security reason not toPaid for by the public; reuse by others.

Whatever the choice, three contract clauses do more than any licence debate: the state's right to its data in open formats, rights to the source code and documentation it needs to change supplier, and a funded exit plan.

Sources