Why We've Tried to Replace Developers Every Decade

The article follows a recurring promise from COBOL to AI: every decade brings a new tool that seems ready to let business users build software without developers. Each wave reduces some friction and helps people work faster, but none removes the need for judgment about complexity, security, integration, and maintenance.

It argues that software work is constrained less by typing or syntax than by the thinking required to handle edge cases and changing requirements. AI, low-code, and earlier tools can amplify developer productivity, but they do not replace the people who understand both the business problem and the technical landscape.

Reading notes

#

• The same promise returns in each decade: make software simple enough that fewer developers are needed.
• Apollo and Margaret Hamilton’s team show that software can be mission-critical, but that it also demands specialized knowledge and time.
• COBOL aimed to let business people write programs, but it still required training and produced a new class of specialized programmers.
• CASE tools promised code generation from diagrams, but they often failed because the diagrams still had to model the same logical complexity.
• Visual Basic and Delphi lowered the barrier to building applications, yet more complex systems still needed experienced developers.
• Web frameworks, low-code platforms, and no-code platforms continued the pattern of making some tasks faster without removing developer demand.
• The core difficulty is the reasoning needed to handle edge cases, concurrency, security, integrations, and maintenance.
• AI coding assistants can generate code, explain existing code, suggest improvements, and help debug, but they still depend on human judgment.
• Leaders are asked to judge tools by whether they help developers work on complex problems faster, not by whether they eliminate developers.
• The article presents the recurring dream as useful because it pushes the creation of tools that genuinely help, even if it never fully comes true.