> So you can improve it with either more empathy or more automated-form design. I work on a CI system that supports an app with more than 150000 tests. At that

> So you can improve it with either more empathy or more automated-form design. I work on a CI system that supports an app with more than 150000 tests. At that size we have to remove tests to manage intermittent test failures. In the beginning we were removing those tests manually and developers would argue with us a lot. Especially about the methodology that we used to determine that a test is intermittently failing. Since we switched to a bot the arguments stopped. I guess it seems less arbitrary if an automated system disables tests rather than an human.