The reality is that there is very little actual innovation happening in programming. Most of the ideas in use today have been discovered decades ago. The big reason for churn is that people don't bother learning about what's been done before, and keep reinventing the wheel.

You don't see in other disciplines such as physics or chemistry where people spend years learning about existing research before actually starting to contribute.

With programming the barrier to starting to write code is much lower than to actually learning the background research. People don't bother looking at what's been done already, and just start "inventing" things.

More often than not this result in half baked ideas because the authors of projects don't really think about the full scope of the problem. Then once the project starts hitting limits in practice, people start kludging things on top of it, and eventually it becomes unwieldy to use. Then somebody comes by and does the same thing simplistically again, and the cycle repeats. Nothing new is learned in this process, and you get churn for the sake of churn.