Code Reuse as a Problem
I was thinking about this question on software delivery, and I kept coming back to the issue of repeatability and / or reproducibility. They matter, because if you don’t repeat a project then it becomes more difficult to improve the process you used to build the project. Engineering involves constantly improving the processes involved with design and construction in order to produce higher quality projects.
Software can rely heavily upon reuse due to its digital form. Instead of rewriting a module, we just call it again or copy it to the other system. Some examples are authentication / login or perhaps a logging function. There are many well known examples for those categories, and conventional wisdom is to reuse what exists instead of rolling your own.
Some Comparisons to Other Disciplines
In contrast, construction of physical systems (buildings, bridges) is nowhere near as reusable. It’s true that the blueprint of a house can be reused many times to build the same copy of the house, but the construction must be performed each time. Cut & paste doesn’t work like that in the analog world. Bridge blueprints are less reusable that houses because site conditions will vary.
Master builders are experts recognized for having designed and / or built tens, hundreds, or thousands of things in their area. For example, Frank Lloyd Wright, a world renowned architect and designer
designed more than 1,000 structures and completed 532 works. Contrast that with Anders Hejlsberg who has designed “just” five languages (Turbo Pascal; Delphi; J++; C#; Typescript). In many ways, it’s an unfair comparison because the domains are different. But at a broad level, the quantifiable production from two very intelligent people is vastly different.
Martial artists will say that mastery of a move comes only from thousands of repetitions. After a good portion of those repetitions have been put in, many martial artists are surprised at how a previously perceived to be complex kata or form has become simple. Instructors of those students will also notice how the motion becomes more fluid and purposeful as well as having an economy of motion. Likewise, experienced martial artists are able to pick up more complex katas more quickly than less experienced students. Experience from repetition has given them a framework or process that allows them to learn more quickly.
Woodworkers experience a similar transformation. Hobbyist woodworkers always refer back to their first project that required a lot of drawers. If they complete the project, they gain a new appreciation for the efficiencies that assembly lines produce. There are other benefits such as a better understanding of how to lay out the drawers parts on the sheet stock in order to maximize use of the wood. Compared to hobbyists, professional woodworkers are able to more quickly design, start, and construct items that they have made many times before. They also gain an ability to see inherent issues within someone else's design having made that mistake in their work.
So, does software reuse prevent software developers from becoming more proficient?
In many ways, software design and construction is always new. We don’t repeat past works, because if we can reuse a module, library, or system then we do. We’ll preferentially extend an existing system before rewriting the entire thing from scratch. But repetition is what allows us to find efficiency in the design and the construction. Anyone who has practiced a sport or physical activity will tell you that repetition is the key to becoming a good practitioner.
My question: Does software’s ability to be reused prevent the necessary process improvement and efficiency that comes from repeating a project?