[kragen]

> It's not at all clear to me that the end effect is as different as the proponents claim.

There have been studies (arXiv used to link to them, maybe still does) that show that open access substantially increases the number of citations a paper receives. That strongly suggests that open access works better for distributing research even within the academy, which is just common sense — open-access papers are easier to discuss, find links to on Reddit and Wikipedia, and often simply obtain. Outside the academy, where it's even harder to measure readership, we can expect the effect to be much more extreme, since most people outside the academy actually have to pay to read non-open-access papers.