The Waterfall model for system development has been a de facto standard since the 1970s. But did you know that Dr. Winston Royce, the man who is attributed with being the first computer scientist to formally describe the model, said in his famous 1970 paper that it doesn’t work? If you’re an advocate of the Waterfall model, keep reading. You’re in for an ‘AHA’ moment.
Dr. Royce never used the term “Waterfall” in his article; nor did he invent the term in reference to the model. In some texts he is accredited with inventing the model which he did not. Both linear and iterative and incremental methods can be traced as far back as the 1930s to the work of Walter Shewhart (pronounced “shoe-heart”), a quality expert at Bell Labs who devised a sequence of “Plan-Do-Study-Act” cycles for quality improvement. Shewhart’s work was expounded upon in the 1940s by William Edwards Deming, the father of the post-war Japanese industrial renaissance. Surprisingly, Royce never advocated for the use of Waterfall as a viable methodology. He called the model “grandiose” and argued that it doesn’t work because requirements change over time. He presented it as an example of a model that is flawed and non-practicable. It is a process that simply does not compute. Royce further proposed that the Waterfall model needed five modifications before it could possibly work as a system development model.
So if the position of the man whom many wrongly believe created the Waterfall model is that it is a busted process, how then did it become ubiquitous? If you scrutinize some of the scientific articles written about systems and software engineering, particularly software engineering, many writers say something like “Waterfall is a proven method (Royce, 1970).” They cite the paper mentioned above which makes no such claim at all.
One reason unsubstantiated claims become quoted as “facts” is that an author cites a reference because some other author did so in his/her publication. Then another author picks up on it, and then another and another, multiplying the effect when not one of them ever read the document they are citing. They each trust the other person corroborated the specifics and got it right. There is no single source of “citation checks” for authors to verify the facts of their references. They either do the verification leg work themselves, depend on a research assistant or risk damaging their credibility by placing their confidence in someone else’s work even if that person is considered a trusted source.
The ubiquity of Waterfall then is due to the laziness of writers and the US Department of Defense. (You need to get the book to understand that last statement.) If you want to learn more and uncover the truth about the the most prolific system development method in history, buy The Ultimate Guide to the SDLC. It’s all laid out for you in Chapter 2.
Author’s note: I’ve noticed that this site has been visited by many people searching for Dr. Royce’s paper. The document is available from the IEEE organization at a nominal cost.
Andy Langton says
“But did you know that Dr. Winston Royce, the man who is attributed with being the first computer scientist to formally describe the model, said in his famous 1970 paper that it doesn’t work?” Where does he state this? If you read the paper he introduces various views of a development process starting at the simplest, where he comments throughout on the potential risks of failure, but his final suggestion is both iterative and, in his opinion, capable of successfully developing large complex systems at a low cost. The ‘waterfall’ labeling from my understanding came about through the Agile political movement when developing a demon to bash to enable them to promote their less cost effective development approach.