In many situations, such as limited test time or resources, it is important to test software in a way that maximizes the number of faults discovered early in testing. Traditionally, test prioritization has been used to attempt to place the tests most likely to reveal faults at the beginning of a test suite. However, what if the time allowed for testing after a change to the code is restricted to even less than the time required to run a single test? I will present test factoring, a method that uses existing slow tests and information about a program change, and attempts to create fast tests that are likely to reveal faults introduced by that change, but may rarely produce incorrect results. I will indicate the challenges presented by developing such a method, and an experimental framework that can be used to validate it. I will also discuss continuous testing, a development environment feature that can be enabled by test factoring.