So .... IF your ARB has looked at the protocol and says "this is purely observational, or records, etc. and is untraceable to the source, so therefore meets the conditions of the exempt categories for observational research" THEN one's pretty much in the clear without permission from subjects IF that's what the board agreed to.
We've had precisely *no* trouble getting blanket exceptions from our IRB for public weblog data. Either for citation purposes, or for using data-as-data. They seem to agree with us in thinking that public blogging should be covered by the same standards as any other public utterance or publication. Most social science or comm-centric projects should just get IRB permission out of habit. It is a good practice to follow. I think maybe you are tilting at windmills, a little, too... --e