I would like to believe that the quality of our system and documentation are so high that there has been no need for Stack Overflow questions, but that's probably not the reason. :)
I think it's true that Hackystat's primary user base is academia. I believe this is because there are a lot of software process/product tools that are much simpler to install and use if you are only interested in relatively simple attributes of your process/product. If you just want to know basic trends in coverage or whatever, then there's probably a Hudson or Maven plugin that will do the job just fine with less configuration and installation overhead than Hackystat.
Where Hackystat starts to look good is when you want to do something more complicated, such as inferring whether or not your developers are doing Test Driven Design by comparing their development behaviors to a rule-based operational definition of TDD. This kind of "complex" question is much better addressed with Hackystat than with a Hudson or Maven plugin, but goes beyond the typical analysis needs of most companies.
Another factor is that while Hackystat's architecture is language-independent, the vast majority of sensors are designed to collect data only for Java-based systems. This creates a barrier to entry for systems written in other languages: to make use of the analyses, one would have to first implement sensors for the new language.