I'm assuming this is a half-facetious joke from the creators, but its core intent is sound: valid feedback doesn't always get to the desk on which the buck stops. Not necessarily out of a pursuit of "see no evil, hear no evil..." but just because there's too much noise in the overall dataset of complaints/opinions, and not enough tools or efforts to filter and sift for possible leads. The recent GM recall debacle comes to mind, in which 260 complaints purportedly related to ignition problems were submitted to the NHTSA yet apparently not acted on [1]. The NHTSA's defense is that 260 complaint records is not statistically significant...but...by their criteria, it seems that a lot of complaint trends would fall in that category. The bigger issue is that not enough information is gathered about the complaints to make an informed decision about recalls. Sure, but can the regulators (or carmakers) honestly say that the people who would prioritize such fact-finding are fully aware of what the complaints data (all which can be downloaded by anyone from the NHTSA website [2]) contains at any given point? It's not about taking action when the truth is encapsulated perfectly in a database, but making the effort to sift that database efficiently for leads.

tldr: there's a lot of room for improvement and innovation in filtering feedback for further investigation and action, not just in prisons but much of the public sector.

(for simplicity's sake, I'm ignoring the obvious factor of companies/regulators deliberately and maliciously not giving a shit/covering-things up...which is of course, always something to be aware of)

[1] http://www.nytimes.com/2014/03/09/business/auto-regulators-d...

[2] http://www-odi.nhtsa.dot.gov/downloads/