I like this metaphor as well. It happens in code and it happens in systems. And often we call technical debt "legacy" because changing it would require too much

I like this metaphor as well. It happens in code and it happens in systems. And often we call technical debt "legacy" because changing it would require too much re-organization. I used to use the metaphor of coral beds for systems design. All the new development you're doing on a system is the vibrant ecosystem in the coral, and then you ship the system, and now when parts of that ecosystem becomes irrelevant you are still stuck with it for "legacy reasons" and now its bones are calcified into the reef underneath the system. Think of it like the VGA register layout for video cards on x86 PCs. Fully mapped frame buffers make so much more sense for a 64 bit system but so many things are constrained by the legacy of code that assumes the VGA register set that it isn't worth changing. As I've become more experienced, I have come to better appreciate the cost of technical debt which has the side effect of slowing me down because I find I don't want to commit to code until I fully understand the system, even if it currently passes all tests.