Even though yeast is far more reliable than it used to be, many bakers still proof it before investing the time and materials to bake a loaf of bread. The extra few minutes waiting for it to bloom is cheap insurance to avoid a failed loaf a day later.

If you need to be sure there are no pits in your chopped dates, it makes sense to avoid mechanically de-pitted fruit. Every single date has exactly one pit, and if you find it yourself, you’ll know you found it.

We can’t do every task ourselves, and we can’t test every raw material, particularly if it’s a destructive test like whether or not this glass is tempered.

The math is simple, but easy to avoid: What are the chances that the component in question might fail, multiplied by the cost to the project if it does. Compare this to the cost of the test and you’ll know what to do.

In my experience, we focus on the easy tests, without thinking hard about the real costs. Three shortcuts to avoid: Tradition, proximity to failure and the vividness of the rare cataclysm.

Traditional tests might distract us from the checks we ought to be doing.

Proximity to failure puts our focus on things at the end of the process as opposed to thinking hard about the underlying components and system failure.

And vivid failures are failures that get our attention, but loud and urgent aren’t the same as important and useful.