Fred Brooks presented “No Silver Bullet - Essence and Accident in Software Engineering” at the IFIP Congress in 1986, and it circulated as University of North Carolina technical report TR86-020, dated September 1986. Its central claim is blunt: there is no single development, in technology or management, that by itself promises even a tenfold improvement in software productivity within a decade.
Brooks draws a distinction between two kinds of difficulty. Accidental difficulties are the ones we happen to create with our tools and methods, and these we have steadily reduced. Essential difficulties are inherent in the nature of software itself, including its complexity, the need to conform to many external systems, its constant changeability, and its invisibility.
Because the essential difficulties dominate, Brooks argued, polishing tools and languages yields diminishing returns. The remaining hard work is conceptual, and there is no magic technology that makes specifying and designing a complex system easy.
The paper has been debated ever since, with later writers arguing about whether any advance, such as the rise of reusable components or modern frameworks, ever amounted to a silver bullet. Brooks himself revisited the argument years later and largely stood by it.