I wonder if closed-access publishing is part of why academia seems so insluated from the "real world". People write for their audience, and if the general public can't read academic papers, then academics are going to write as if only other academics are reading. Likewise, if research output is difficult to access, the feedback loop between ideas and implementation is broken; folks outside academia can't easily comme…
We're writing that way because we are the explicit intended audience anyway - the purpose of journals and conferences is not to write about our research and see who wants to read it, but entirely the opposite, to make up a venue that a particular research community wants to read and then ask for submissions that would be interesting to other researchers.
Public dissemination is not part of the research-publish feedback loop which drives the actual research, it's a (useful) output out of that loop but not really part of it.
The feedback loop is an exchange of novel research, not finding out "what others think of your work". If you've got a better method, that's interesting; if you've got a novel use case of a method that I know, that's interesting; if you've got experimental evidence that contradicts mine, that's interesting; but that's all novel research that would be publishable even with all the existing filters, and pretty much requires the person to be familiar with the field (by which I mean having spent at least hundreds of hours reading relevant research beforehand).
Comments, especially uninformed ones, have too low signal-to-noise to be worth reading - any active field of research produces more than a person can read anyway, so if anything the researchers want stricter filtering that reduces noise. That's a major purpose for the more selective venues; I want someone else to read and reject most of academic publishing so that I don't have to read it just to mutter the same objections that the reviewers would've had, just with more obscenities.