Antifragility is a concept developed by N. N. Taleb in his book, “Antifragile: Things That Gain from Disorder”. How could it be applied to automated tests?
Usually tests are run against a predefined set of input parameters, so they do not validate all possible options. Not to mention that validating everything is practically impossible.
How could we improve our tests so each run increases trust in our code? For free?
There is a simple way to ensure this – instead of static input parameters use randomized input. Each run uses another set of parameters, checking another point from the input parameters space. The more runs the greater the chance of exposing a hidden bug, should one be present. And more confidence, in case of successful runs. The usual: proper range of inputs is the key.
In other words our tests will gain from chaos. So they are antifragile by definition.
