Should This Test Be Automated?

For the decision to automate a test to be a good one, the value of re-running the test must exceed the cost to automate it. People often measure value and cost in the wrong terms. In this double-track presentation, Brian Marick contends that value is not ease of execution, but the likelihood of finding future bugs. And cost is not time spent automating, but rather the manual tests that will never be run.