How does OA benefit my research?

Jan Jensen has written an interesting post describing how his decision to publish only on Open Access outlets has influenced the way he tackles research questions.  One of the benefits he points out is that choosing to publish in a journal which performs a "scientific soundness-only peer-review" instead of a "sexyness/interest and scientific soundness peer review" allows him to focus on "truly challenging and long-term research questions without worrying whether or where I will be able to publish".  I think that option already existed before OA and the advent of the mega-journals: we simply had to decide to be satisfied with publishing on IJQC or Theochem whenever the Editors of JPC, JCP, JACS, Angewandte et al.  pronounced our research "too specialized and not of enough interest to our broad readership", and to accept the derision of peers who look down on papers published on those and other low-impact journals. (I admit I am often guilty of this).
To me, the true advantage does not lie on OA itself, but on the open review model (used e.g. by PeerJ), which allows authors to publish the reviews at the same time as the paper. I feel this functions as a much stronger "validation" of the quality of the work, as readers immediately have access to a truly independent measure of the strengths and weaknesses of the manuscript.
How does OA benefit my research? I am not sure it benefits my research methodology and/or choice of research questions since, as one of only two computational chemists at a small teaching-driven University, I  have long decided to research whatever obscure subtopics catch my fancy due to obvious lack of resources to compete against larger/well-funded groups working in sexier topics/enzymes. My decision to embrace an open science model, in contrast (e.g. figshare) has benefitted me more directly by forcing me to archive my results in a more transparent way, with proper "understandable" filenames instead of idiossyncratic names chosen on the fly... That is something I should have done anyway even without the open science model, but that was the nudge which brought me to the "Light" side.

When the description of methods in a scientific paper becomes optional.

I have just read a paper describing some very interesting tailoring of enzyme specificity on a P450 enzyme. I was, however, surprised to find that no description of the experimental methods was present in the paper itself, but was only available as Supporting Information. Upon examination of the instructions for authors in the journal I learned that, although being online only (and therefore lacking any space constraints), this publication enforces a 40-thousand character limit on the published papers and specifically states that the experimental section is optional. Traditionally, Supporting Information includes accessory data which would be cumbersome to include in the paper.  In this journal, it functions instead as a cumbersome way to access a vital part of information which should be part of the paper. I cannot even begin to understand why any reputable publisher would, in the absence of any printing costs, force their authors to split their manuscripts and "demote" the potentially most useful portion of the paper to the Supporting Information.
That's ACS: proudly claiming to "[publish] the most compelling, important primary reports on research in chemistry and in allied fields" while making it difficult for readers to have access to that same information.