I have a theory that software drives people insane

The internal cost of changing software is obscured, which encourages continuous, costly, and often unnecessary modifications.

The article "Software Drives People Insane" by Graybeard argues that the unique conditions surrounding software development lead otherwise rational individuals to lose their sense of proportion. These conditions include speed, money, complexity, abstraction, and freedom to change. Unlike physical construction, where the cost of changes is evident, software modifications often have hidden, accumulating expenses related to context switching, regression risk, and architectural erosion. This opacity makes it easy to pretend changes are free, leading to frequent, often uncritical, adoption of "cool ideas." The potential for immense financial upside further inflates the perceived importance of minor decisions, turning mundane tasks into high-stakes gambles. Software's inherent mutability, combined with industry pressure for constant velocity and innovation, creates an environment where organizations and their components are treated as perpetually unfinished and in need of continuous modification, fostering a culture hostile to patience and stability.

The author posits that software's inherent nature—speed, money, complexity, abstraction, and malleability—creates a "perfect storm" for organizational neurosis. Unlike physical projects with visible costs of alteration, software changes incur silent expenses in context switching, regression risk, and architectural degradation, making them seem deceptively free. This illusion encourages a culture of constant iteration and change, driven by the potential for massive financial gain and the industry's obsession with velocity, often leading to unnecessary complexity and a lack of patience for systems to mature.

This dynamic explains why software organizations can become self-reinforcing cycles of activity, where complexity is pursued for its own sake, providing work and status. The industry's lack of natural endpoints for 'done' means that software, and by extension the companies that build it, are perpetually treated as unfinished projects requiring constant 'debugging,' optimization, or pivoting. This relentless mutability, fueled by venture capital and a fear of stagnation, ultimately makes it difficult for good work to simply "sit there and work" without continuous, often detrimental, intervention.