The Call for Simplicity in the Era of Digital Complexity

In the current technological landscape, dominated by increasingly complex systems and distributed architectures, a counter-movement is emerging that advocates for a return to software simplicity and compactness. The "Fits on a Floppy" manifesto, despite its name evoking a bygone technological era, does not refer to the physical media of the 1980s, but rather to a development philosophy that prioritizes lightness and comprehensibility. The central idea is that by imposing an artificial limit on a software project's size, one is forced to keep it intrinsically simple.

This perspective resonates with the challenges many CTOs, DevOps leads, and infrastructure architects face today, particularly when evaluating alternatives between on-premise deployment and cloud solutions. Growing complexity can translate into high operational costs, maintenance difficulties, and less transparency regarding data and operations. The "Fits on a Floppy" manifesto positions itself as a beacon for those seeking to navigate this scenario, offering an approach that values control and efficiency.

The Philosophy of Lean Software and Wirth's Law

The concept of lean software is not new. As early as 1995, Niklaus Wirth, a pioneer in computer science, published "A Plea for Lean Software," an essay that inspired Belgian consultant Bert Hubert to write "A 2024 Plea for Lean Software" in 2024. Hubert, like Wirth, laments that "software gets slower more rapidly than hardware gets faster," a principle known as Wirth's Law. This law highlights how software inefficiency tends to erode hardware performance gains, a particularly relevant issue for intensive workloads such as those of LLMs.

To demonstrate the validity of his argument, Hubert developed a web image-sharing tool, implemented in under 2000 lines of code and distributed as a compressed Docker file of only 1.7 MB. This concrete example shows how it is possible to create useful and functional applications while maintaining minimal sizes. Other modern projects follow this path: the Dillo web browser, for instance, is still distributed on floppy disks, and programming languages like Hare and Janet (the latter with a download of just over 2 MB) aim for compactness. The goal is not just to save space, but to improve the readability and long-term maintainability of the code, crucial aspects for the sustainability of any IT infrastructure.

Data Sovereignty and Cloud Alternatives

The debate on software simplicity is closely intertwined with that of data sovereignty and deployment strategies. The "Fits on a Floppy" idea and the call for lean software serve as a warning against excessive dependence on complex and often proprietary ecosystems. Many experts, including Bert Hubert himself, criticize the approach of those who propose a "sovereign cloud" as a solution to public cloud problems, arguing that it would merely be another form of lock-in.

The true way out, according to this philosophy, is to discard failing models and return to a simpler approach, where organizations own and manage their data on their own servers. This alignment with the KISS Principle (Keep It Simple, Stupid) suggests that the best solution is always the simplest possible. For companies evaluating AI/LLM workloads, this means seriously considering on-premise deployments, which offer greater control over security, compliance, and long-term costs, avoiding the complexities and uncertainties of the cloud. AI-RADAR, for example, offers analytical frameworks on /llm-onpremise to evaluate the trade-offs between these different strategies.

A Path Towards Radical Simplicity

Adopting an approach that prioritizes compact software and lean infrastructure is not the only path, but it represents a valid alternative for those seeking greater control and transparency. The idea of "Just Use One Big Server" and relying on experienced technicians for its management is an example of how infrastructure can be radically simplified. This contrasts with the tendency to layer unnecessary complexity, often driven more by marketing than by actual technical needs.

The history of computing itself offers examples of how powerful and revolutionary systems emerged from limited resources: consider software like Lotus 1-2-3, which redefined business management by operating on 360 kB floppy disks. Today, with incomparably superior hardware resources, the opportunity to create efficient and controllable software is even greater. For CTOs and architects, embracing radical simplicity means not only optimizing TCO but also reasserting sovereignty over their data and infrastructure, an increasingly critical value in the age of artificial intelligence.