Software growth and improvements are made based on a shared mental model of its capabilities, operation, constraints, and edge cases.
Much like an organic system, the complex structure can develop unhealthy conditions that when untreated can cause complications, and is subject to misdiagnosis or mistreatment, further exacerbating its health and longevity.
Unlike an organic system, a synthetic system can be built with intention, designed for resiliency with intention of ongoing operation and future growth, and care can be taken to persist knowledge of that system indefinitely. This should favor convention and discoverability over configuration and documentation, where knowledge is easily found where and when it is needed.
The most formidable asset a long-term software product can have is institutional knowledge – the people who built it are intimately familiar with it when built, gaining experience and knowledge with its idiosyncrasies, and that knowledge fades wherever they lose contact, be it in operations, maintenance, or stale features.
When planning a new product, you should be committing a portion of those developers to its lifelong care, and you should treat them well. Some churn is natural, and the product care team can heal and overcome this provided there is sufficient retention, but if they are demoralized, lack psychological safety, lack the ability to guide and affect change, or experience significant churn then the product care team will not be able to recover institutional knowledge quickly enough. The product will bloat, experience bit rot, falter from configuration drift, accumulate vulnerabilities and bugs, wither and die.