[bluGill]

I have long written the pyramid with the vertical axis of runtime. At the bottom are tests that run so fast you would run them every build even if there is no possible way your code changes could break those tests. Then tests that you run if they test what you just changed. Next tests you run on all CI builds [locally only when they fail]. Then tests you run regularly but not ever CI builds (every night or every week). Then tests that are so expensive you run them rarely - these generally are manual tests where a human is running the test in some real world condition that is hard to setup.

There is value in every level.