Online journals cannot be seriously interested in getting the public to purchase their content. Consider the case of the paper Error and the Will, which is typical (perhaps even on the low end) of what I've seen in terms of cost.

I came across this paper using Google Scholar. I can't find a digital copy anywhere, even using my alma mater's library system. I find myself asking these questions:

• Without knowing the content, should I spend the $30USD for it? That's a risky endeavour just to read it, not knowing if it contains something useful.
• I can "request permissions" to get a copy for personal use (although, oddly enough, "read it" is not one of the options when you do that, although "reuse in a thesis/dissertation" is). It takes up to five business days to see if that is acceptable and how much it will cost.
• If I'm trying to to write a paper and I want references -- say, 10-15 or so -- that will be well upwards of $300 just for the references I actually use, notwithstanding the ones I don't use. That's a huge risk with essentially no chance of a payoff.

Even if you are going to charge me to read something, at least learn a lesson from iTunes and Steam: make it easy and I won't bother with less "respectable" methods.