I agree that science cannot truly be considered open while the results are essentially embargoed in highly expensive journals. The typical going rate for a single article is $35, which made sense when they photocopied and snail mailed to you, but is about 2 orders of magnitude overpriced for internet access. I find that when I do drag myself off to a physical library, I typically have to scan through 10 to 20 papers to find 1 that is worth reading in detail. The abstract is typically useless for determining the quality of the paper. Now do the sums, to find a couple of usable papers you will need to look through 20 to 40 papers, which at $35 a pop is going to cost you $700 to $1400. This meets no definition of open that I am aware of, and is a significant barrier to information access for a non university scientist.
Even when you go to a physical library these days, you'l find they typically have only a small number of the journals you find yourself needing. This is because the majority of professionals already do all their journal research online.