A new timestamp or a slightly longer heading can create a visible difference without being a defect. The tool needs to help someone separate those changes from actual problems.
What would you test before relying on the report? Would you rather tune the checks narrowly or review more differences manually?
.
