The problem with 'commonsense' is that it's neither a sense, nor terribly common. It is, rather, that sweet spot where intuitions are useful. Which is to say,

The problem with 'commonsense' is that it's neither a sense, nor terribly common. It is, rather, that sweet spot where intuitions are useful. Which is to say, when you make a 'common sense' judgment, you are typically making a judgment based on gut/intuition. You are accessing your onboard neural network, which was trained against the corpus of your entire life, rather than some rule you don't feel in your bones. There are many problems with this, but for now, let's notice that intuitions honed at one level might not work on another. For example, intuitions about household debt are famously bad intuitions to apply to U.S. federal debt. So looking for e.g. commonsense solutions to the U.S. debt will probably not work -- you simply won't find one that feels right. This is because you (dear reader) are stochastically unlikely to have much experience in running a country. An analogous problem occurs in software. And what the intuition cannot usefully reason about, it must cede to systems and algorithms, all of which, by definition, feel 'un-intuitive', because they are literally substituting for your nonexistent and/or offtopic intuitions. Simply put, most peoples' intuition cannot scale, because the training corpus of most people does not include examples of sufficient scale, so they are stuck blindly applying rules and hoping for the best. There's just no substitute for experience, but scrum afficionados try anyway ;D