http://www.nature.com/news/open-access-the-true-cost-of-scie...
Profit margins are amazing though, it really pays to get into publishing. Elsevier (not all open access, and rather dodgy to many) is one of the most profitable businesses in the world: https://medium.com/@jasonschmitt/can-t-disrupt-this-elsevier...
This may depend on the field, but in CS I have never seen this. There is essentially no change made by the publisher to papers, and when they do make changes, more often than not it is to introduce errors.
In any case, the huge profit margins of publishers indicate that the cost of this dubious service is much lower than the charge.
But the most important point is this: doing minor copyediting shouldn't be grounds to own the journal and own copyright on the published articles. The journals should be owned by scientists, or by a nonprofit scientific society. Even assuming that research needs to pay for copyediting services, these should be contractors of the nonprofit (with competition among them), not owners of the journal and articles.
In principle it's great that articles are uniformly in XML -- you could easily write reader software for tablets or phones, data-mine the text, extract references, whatever -- but I haven't seen much software taking advantage of it.
However! Once published and out the door, the shiny XML is available for scraping and transformation, and, if OA, there is no reason the XML can't be made public: https://github.com/elifesciences/elife-article-xml
An excellent bit of software that takes advantage of it is Lens: https://github.com/elifesciences/lens
Yeah, as in forcing you to submit in an artificially crappy template with horrible double spacing (which is awful as a reviewer) to then apply their standard template and voilà! The paper looks good.
give it a thorough last read (finding for me several errors that peer review never found - missing or switched references, for example)
They do that (with varying quality) but it's quite aggravating that they charge thousands of dollars mostly for that, while peer review, which is much harder, is done for free.