I'm an open source developer of testing related tools including Selenium and Watir, and a passionate advocate for using automation to improve digital confidence.Between implementations I've created and worked with, I've seen many different approaches to test automation. My talks and presentations are often based on things I've screwed up in the past, or patterns I've found to be unsuccessful and the approaches I've found that work better.I'm a champion for good testing practices and processes. Automation provides an opportunity to revolutionize functional DOM to Database testing. The methodologies and approach should be fundamentally different from manual testing. Tests need to be atomic, autonomous and short, leveraging parallelization for scale and REST APIs for setting state and validating results. The biggest cost of automated testing is not the creation or execution of tests, but the maintenance of tests. If you are spending the majority of time validating that test failures are not actual problems and tweaking intermittently failing tests to get them to pass, you aren't doing it Right. The way to ensure a worthwhile testing effort is using a well crafted framework that provides consistent and trusted results.