[jdbernard]

Yeah. As I've gotten older I've cared less and less for "my code." If the new effort is demonstrably better than the existing system, then I'm happy. Who cares who wrote the old stuff, this is better.

The problem, of course, is that often the new system is not better, it's just newer. Sure, it may be simpler and easier to understand, but is that because it is an inherently better solution, or is that because it doesn't deal with all of the the real-world conditions that the productions system has grown to support over time?

Even then re-writing is often a good thing. You can re-approach the problem domain with all the insight you've gained building the old system.

[jey]

Ideally there would be good enough tests of the old system that if the new system passes the same tests, then there's high confidence that it will perform correctly in production too. That way we get the benefit of being able to refactor or reimplement entire subsystems while still having almost the same assurance as the old system.