Practices or methods being followed in software manufacturing are mostly like an assembly line of a manufacturing company and this will not lead to new product development, according to Craig Larman, chief scientist, Valtech. |
Addressing a seminar on "Bringing Agility into Offshore", Larman said that software practices being taught in companies are based on a restricted paradigm and can be classified under "Waterfall Method". |
The source of Waterfall Method was Dr Winston W Royce, who, in 1970, said that software practices must begin with analysing the requirements, designing, programming and then testing these applications. He had also called for repeating this process several times to achieve efficiency. |
By narrating examples, Larman highlighted that a new product cannot be developed with the Waterfall Method. Again, with various examples Larman also pointed out that Agile, Iterative and Evolutionary Methodology is the way to develop new products. It is a need based process and not a method, combining various processes and methodologies. |
"There is no way to estimate the true cost, effort and size of the project even before it takes off. Several studies have proved that reliable estimates given during the beginning of the project could increase of decrease by four times in size, cost and effort when it ends. Software development is not predictable manufacturing," Larman explained. |
Giving examples on how well-known organisations suffered as a result of following the Waterfall Method, Larman said, "Until, 1987, the US defence department followed the Standard 2167, which was basically the standards of software. The Standard 2167 was based on the Waterfall Method. US even shipped this to its allies." |
However, when failure rates increased, US government deputed Frederick P Brooks Jr, chairman of Defence Science Board, which was a Taskforce on Military Software. Brooks called for stopping the Waterfall Method and following Agile methodologies. This came into effect from 1987." |