If you're in the United States, you measure distance in miles and weight in pounds because that's what the people around you use. Similarly, if a community of scientists are using [X] and condition #1 is satisfied, we'll teach our fundamental ideas—task automation, modular programming, version control, and structured data—using [X], because those ideas are what really matter.
When someone says, "Teach [X]," it's almost always because they believe it's better. When asked for evidence, though, they almost always respond with personal anecdotes or the sort of a priori reasoning that led Aristotle to believe that heavy objects must fall faster than light ones. We can do better: we can measure the usability of programming languages through control experiments. And I think that means we have an obligation to: as I've said many times, if we want other people to pay attention to the evidence when it comes to things like climate change, we need to demand and respect evidence ourselves.
This post originally appeared in the Software Carpentry blog.