I put it down to the poor state of our mathematical education. The naive understanding of economic systems libertarians display is actually fairly similar to what you see in introductory dynamical-systems courses. You can, with great care, construct some "well-behaved" systems where you present a set of rules, and the expected thing happens: you push lever X and Y happens. Raise the minimum wage and unemployment rises!

But of course, most systems are not so well-behaved, so that is only a lead-in to the real examples, where you change the setup slightly and suddenly analysis grows more complex. Now what happens when you push lever X depends on what region of the state-space you're in! And sometimes pushing lever X can change the region of the state-space you're in, too! There is feedback and internal dynamics. I see very little libertarian economics that displays even a 101-level understanding of any of that.