Planning for failure isn’t always about avoiding it

In the lexicon of most developers, testers, and product owners “failure” is something to be studiously avoided.

As IT professionals we have developed entire systems and philosophies to ensure that failure does not result from the use of our work products. Test-Driven Design. Six Sigma. Continuous Integration. Automated Testing. And on and on.

This is certainly a good thing. None of us want what we work so assiduously to deliver to fail to perform to expectations. Our continued ability to delight clients depends on it. So, in some cases, do the lives of members of the public who fly on the planes, wear the pacemakers, and drive the vehicles that contain the end-results of our work.

But, what if failure were something to be designed into a product, to be celebrated as part of a properly functioning system? Can this ever be so?

My first reaction when this thought occurred to me was to respond with “Of course not! Failure is not an option!”

(I’m an Apollo fanatic, as you can tell.)

But, thinking further and more creatively, another thought came to me: “Yes, it could be. There are situations where failure is planned for and not failure is itself a failure mode.”

Before you look further, can you imagine what those might be? Keeping in mind that “failure” can be defined as “when a component behaves in a manner which would usually be considered a fault”.

(Pause for effect.)

If I were an aeronautical engineer working on a new aircraft wing design, the last thing I would want to happen is for the wing to come apart while in flight because I specified too thin a piece of aluminum for the flap hinge. That is a failure (the separation of one working service from another in a catastrophic manner) that is not to be desired.

But, in keeping with the idea of the failure of a metal surface, and thinking about why that might be a good thing, does anything come to mind?

Go on, take a break for a moment to think about this–pop open a soda or beer if you need to.

Anything?

Well, when you lifted that tab on your drink can, didn’t the metal covering the opening fail in the exactly right way to permit you to pour the liquid out while keeping metal fragments from ending up in it?

There you go–a planned failure, in other words. The metal was scored in exactly the right way, to exactly the proper depth, to permit the cover to failure before the tab reached its failure point. Time and again, hundreds of thousands of times a day, this failure mechanism just works.

As do the rip-strips along the top of resealable plastic food packages.

As do the tops of bottles of auto oil, the film over containers of lunch meat, the seams along plastic molded parts.

All designed to fail in a specific, controlled, repeatable manner.

How cool is that?

Makes you think, eh?

Richard Fall

About

I am currently the National Solution Architect, Digital Platforms and IoT for Sogeti, working from the Des Moines, Iowa office. My interests lie in the areas of micro-services, SaaS, and IoT systems.

More on Richard Fall.

Related Posts

Your email address will not be published. Required fields are marked *

5 + 6 =