I think the answer for Tableau goes back to the first sentence you quoted. A dashboard that supports exploration and self-reliance will be different from one designed to communicate something like "how are we doing today?" I totally agree that the former probably won't look good in a PDF, however the latter might be perfect.
Jonathan, I don't see how including dashboard controls, and sliders aid communications, or clarity. It just ends up looking like a hack job -- a lazy viz. As if you don't care about the viz you're producing; don't care about your audience, and/or don't care about your reputation as a data viz guy.
Don't get me wrong, I've produced my fair share of bad looking vizzes, but I am currently trying to up my game, and swearing of PDF exports is my first step. After all, how much longer does it take to take a few screen shots (without the filters/parameters/sliders), arrange them nicely in a word doc, add some better looking labels, and them export to PDF? Sure it does take longer, but it looks so much more professional, and that is never not worth it.
I think we're talking about the same thing here...I agree that a dashboard designed for exploration with the necessary controls, etc. will just be a hack screenshot as a PDF. My point was that it's also possible to design a dashboard that is focused on answering some relatively simple questions (without much if anything for interactivity) that can look good both on screen and in PDF (or in my case, pasted WMF into PowerPoint). Whether it's worth it to go to that amount of effort in building the dashboard vs. screenshots and manual annotations is a related but separate question.