As this O'Reilly article, Toward the next generation of programming tools, points out; software developers are used to bad tools. Being able to use bad tools seems like some sort of hazing or right of passage necessary to become a computer programmer.
That is going to change.
Why does software for using XBRL have to be ugly, hard to use, and not produce a quality result that matches current quality levels of a financial report created using current processes? Sure, the processes need to change in order to automate a significant portion of the currently far too manual processes. But if the quality level is not maintained or improved, there is ZERO probability that modern approaches to accounting are going to be adopted. ZERO.
The point of accounting is to get things right. That is why two ledgers are used (i.e. double entry accounting) when one ledger could work but the probability of errors would increase.
Wired's article The End of Code helps you understand some of the possibilities. The way I see it is that business professionals will configure logical rules and logic engines will do the work.