Thanks for your insight, it's really great hearing from a librarian. One sad, but interesting note: while it seems ridiculous, our costs to publish a paper have typically increased rather than decreased. The margins on traditional journals are actually shrinking, not growing. This is mostly because we simply can't easily drop print entirely for legacy journals. New ones we can launch digital only (although there are lots of reasons that launching new journals is increasingly difficult too). But even though it sounds ridiculous, our customers (you, the librarian, or at least your peers) often demand print in addition to the electronic. So now we have to pay for the electronic distribution as well as the print.
This is purely a problem of legacy. If you launch a new OA journal there isn't the same print expectation. And granted, it's a problem that will go away eventually. But I continue to be surprised that we can't simply stop doing all printing of journals.
And yet the amount of money libraries spend on getting access to published scholarly research has gone up astronomically too. [1] [2] (even if you normalize it for inflation, or per university FTE, or as a percentage of library or university budgets, whatever. Astronomicaly.)
So where's the money going?
When Elsevier reported 36% profit margins in 2010 according to Wikipedia[3], I have trouble believing that publishers are being squeezed too much. At least not Elsevier and other mega-publishers. I can't easily find googlable historical Elsevier reported profit margins, but I would be shocked if they have actually gone down rather than up over the past 10-15 years.
For comparison, it looks like traditional major book publishers have profit margins of around 10%. [4]
Wikipedia states and cites << [Elsevier] also claimed that its profit margins are "simply a consequence of the firm's efficient operation" >>
It doesn't sound like decreasing margins to me, and it doesn't seem that way to the universities paying for these margins, who can't afford it.
This is purely a problem of legacy. If you launch a new OA journal there isn't the same print expectation. And granted, it's a problem that will go away eventually. But I continue to be surprised that we can't simply stop doing all printing of journals.