We work in an environment where hearsay and taste drive change instead of studies and models.
While VCs and influencers encourage us to jump on the emotional UX and viral social-network train to make our ideas succeed, we tend to not consult our logs first. After all, tapping in the dark is not science, but that’s sort of another topic.
Richard Minerich writes about the shift in programming languages away from proven ones to scripting languages. I tend to agree that dynamic languages should mainly be used as glue. Building a house out of porous cardboard could work if you’re an experienced professional, but it’ll probably fail for most of us. That said most of my code to date is dynamically typed because it’s just way to comfortable.
However there’s no test that’ll cover every single error case on the other side there always will be static and correct formulas.