<-- home

On the ability to reproduce problems

Probably one of the most notorious features of the scientific method is the way all it’s apparatus is enrolled around the ability to reproduce procedures, experiments, results.

With software development it’s the same. When something is hard or impossible to reproduce one can only imagine what’s going on. No real progress come out of it. The engineer is forced to float around a mesh of ideas and just wonder which one is more probable with his limited knowledge.

My point is: you’ll never be able to solve anything without reproducing it first. And reproduction has to be simple in order to test it properly. If you have to take 10 steps to reproduce a bug, probably you will not test if enough. The ideal number of steps to reproduce something is 1. One command.

So improve your tooling and architecture in order that things are easy to reproduce. In web development I think this relates to learning to learn how to use CURL. It relates to knowing shell and being able to run composed commands hacked by yourself.