I think this "poor writing", as in hard to understand writing with long backreferences is due to the same reasons as why many academics are terrible at explaining things, as well as why source code is hard to read.
I should know, others usually like pointing out how terrible I am at explaining things. And I think I know why.
When I make a naive attempt at describing a complex, interconnected concept, such as a piece of math, or a program that does a thing I start with a mental picture - a graph of interconnected concepts. Then I try to describe the picture - the same way one would describe a painting. I describe each visible part separately, often in great detail, then I try to describe the links and interconnections between parts.
This sort of 'accurate' descriptive style naturally leads to huge backreferences, and people usually have a ton of trouble following along.
However I think there's a way of getting your point across - one basically needs to reproduce the thought process of how one would come up with the given model given the real world constraints, illustrating the issue and its proposed solution with examples. It is often said that people are much better at deriving the rules from a few examples than the other way around.
However this sort of descriptive style takes both a lot of effort, and usually is less precise, which often is not acceptable, such as in the case of contracts or computer code.