Here’s an idea that I think can be interesting in terms of how you view testing (as an activity) and tests (as an artifact).

I should note that the basis for my current incarnation of this idea is due to having just binge-watched three seasons of The Leftovers. While doing so, I was reminded of an idea I had about handling a tool apocalypse. That idea was basically imagining that every tool we use to support testing — at least the automation and management parts of it — were destroyed. Could we still function?

I would often present this idea to teams just to learn how much we were relying on artifact crutches in the form of tooling. This wasn’t necessarily to suggest that we should not be using those tools. It was simply to make it clear how much we do, in fact, rely on them.

The Leftovers allowed me to refine that idea somewhat. No real spoilers here, but for whomever has not watched the show it’s essentially predicated upon the idea that in one second, 140 million people all over the world disappeared. So that’s between 2.3 and 2.4 percent of the human population. They all just simply “poofed.”

So now let’s imagine our test repository with its population of tests. And let’s say some (random, as far as we know) percentage of those tests underwent a “Sudden Departure” event.

The powerful question here is: How effective and efficient would our testing still be?

This actually can be an interesting experiment to try out. Particularly if you try out the scenario multiple times, having random tests be “disappeared” in each scenario. And maybe it’s not so random. Maybe you introduce something that selects tests to remove based on some criteria, such as tests that are “too short” or tagged a certain way or whatever else comes to mind.

This kind of “what if” scenario can also give you discussion points around the level of detail in tests. For example, if you have a series of workflow tests that may duplicate some aspects but tend to cover wide areas of end-to-end functionality, does this mean your test effort will “survive” better even if more of those kinds of tests are removed? Or is the opposite true? You can also consider this scenario across the scope of testing, such as unit, or integration, or whatever other delimiters you use.

I recommend you challenge yourself and your teams with this kind of idea. Beyond being instructive, it can be a lot of fun.

Well, as fun as an apocalypse can be.