There are unlimited ways of building software, from the overall architecture to the minutiae of the files structures, from the componentization to the preferred order of parameters on the APIs. Because of that, no two piece of software are the same, which is very much different from other engineering projects, like building a submarine where you will end up with two identical replicas even if they are built by different builders following the same specification.
Any software company (or software project) will define its DNA on its initial months. The tone of the application will be set by the initial developers, managers and designers, and it permeates through everything the application is and for the rest of the application life.
Changing this DNA is very hard. If you ever worked on version 5 or 6 of a product, you know how it is to maintain backward compatible code because of decisions made on version 1. Rewriting everything and dropping all “old ways” is like starting a new company and most of the time is not what investors, customers and partners would consider good business practice.
You can do facelifts on your application, and completely change the user interface and experience, like Microsoft Word 2007 did, but the core continues to be the same. You can replace components with faster versions re-written from scratch, but the DNA of the application will not be different, it will just mutate a bit. If, an application lives long enough, like AutoCAD, Microsoft Word or Unix, its current DNA mutated so much that it doesn’t resemble the original one anymore, but such change probably takes 15 years or more.