Eric the Gray

Testing When It Really Matters

This is an EPI-Pen. They come in pairs and this one’s twin saved my life yesterday. I’m allergic to the stings of bees, hornets, and yellow jackets. Yesterday I was about to see a play at an outdoor theater. It was a perfect October day here in Wisconsin, sunny and in the mid-seventies. I was with my family and we were all excited to see a play together. But, we were outdoors in the fall when the yellow jackets are very aggressive and I got stung just 5 minutes before curtain time.

An allergic reaction to a bee sting can kill some people in a few minutes. The EPI-Pen can temporarily stop the reaction and allow you to get to a hospital. Typically, you get about 30 minutes before the reaction starts again. The pen gives you a shot of epinephrine when you press it against your leg. It fires a half-inch needle and then injects you, all automatically, while you are starting to gasp for breath. Pretty scary.

The interesting thing here to me, aside from still being alive, is that this device had to work. There are no second chances, it has to just work, right now. How do they make a product like this? Once it is used you have to throw it away. Therefore, the individual device that I used was not tested. To me this would be an engineering nightmare.

This is pure speculation on my part, but I suspect that actually coming up with a design for the product was quite simple. The really hard part was to design a way to predict that it would work for me yesterday. I can imagine that a lot of statistical analysis enters here too. They are probably made in batches and a percentage of each is tested. Once in a while an entire batch is destroyed by testing all of them. The production of this life-saving product is mostly about testing, testing, and more testing.

I’m a software developer. The testing of software seems to be in its infancy compared to this. There are now some good testing frameworks and I know lots of people who are really trying to change our culture. They want us to think about testing first and not as a secondary thing that we do a bit when we are done with the “real” work of the development. I’m not sure that software “engineering” will ever get to the point that hard engineering is already. Where the real work IS the testing and the work that users interact with is just the result of our good testing. I’m not even sure that the kind of person who is attracted to software development is capable of proper testing.

This experience will make me look harder at testing now. I have been really trying to find the right way to articulate how I was feeling about testing. I think I might have found it.

By the way, I am just fine now. Take care.