by Frederick P. Brooks, Jr.
In Chapter 16 of The Mythical Man-Month, Brooks Jr. talks about the "silver bullet." Computer hardware continues to grow by leaps and bounds, but software is unable to grow at the same speed. Software projects continually get behind schedule, are not as efficient, and never make the deadline. Brooks argues that there is no silver bullet; "there is no single development, in either technology or management technique, which by itself promises even one order of magnitude improvement in productivity, in reliability, in simplicity." There are two types of difficulties in software development that plague projects: essential difficulties (those inherent to the problem itself, such as algorithms or development techniques) and accidental difficulties (those that concern the development process, such as development machines, machine time, and programming language). Major essential difficulties include the complexity of software as compared to hardware, conformity (that software products might be controlled by other parties), changeability, and invisibility (that the product is not easily visualizable). Many of the accidental difficulties have actually been resolved through developments in technology, through high-level programming languages, time sharing of machines, object-oriented programming, and artificial intelligence. While many advances have helped with the essential difficulties, such as rapid prototyping and incremental development, there is yet to be a silver bullet that allows for a large growth in software development efficiency.
In Chapter 17, Brooks refines his original statement about the silver bullet with an update almost ten years later. After critics have issued rebuttals to his theory, he has decided that there actually is a silver bullet. First of all, visualization techniques and diagramming heavily help the process. But, the most important thing is to focus on quality of software instead of just productivity in the software development process. When you focus on quality, you learn how to produce great software and find techniques and shortcuts that end up helping the productivity of the team. This whole system in called the "Vanilla Framework." In all, although a lot of things have been developed to help software development be more efficient, but we shouldn't sit around waiting for a major breakthrough. Instead, we should focus on the incremental upgrades that can be made and continue to look for small ways to be more efficient.
I think for the most part that I agree with Brooks about software development and efficiency. It's easy to continue to make advances in the hardware for a system, but hard to have a large leap in efficiency in a software product. Because of the complexity of the system and how software is always changing (to keep up with the new hardware), it's hard to keep up. I definitely agree more with Brooks' revision of "No Silver Bullet," in that instead of focusing on large-scale order of magnitude developments, we should focus on the small improvements that can be made. This includes higher-level languages, new debugging techniques, version control, more useful and complex IDEs, web development, and new forms of wireless and touch-based computing. If we continue to incrementally upgrade our software development tools and methodologies, we can still see great improvements over the development processes of the past.
No comments:
Post a Comment