Read other letters about this article
Open source journals needn't be profitless. BioMed Central (http://www.biomedcentral.com/) has totally free access. Instead of the reader paying, the authors pay, and make their research available for free.
This works well for anyone publishing highly specialized research. If you want to make your work broadly available, publishing in a super-specialized subscription-only journal may bury your work in a journal that few libraries get. However, if your work is out there in a freely available journal, anyone with an internet connection can get it.
Elsevier is fighting a losing battle. As we scientists get used to the idea of having free access to published research, we're going to demand it from the fewer and fewer journals that don't provide it (many journals now provide their publications for free within 6 months of publishing - it's only the privately held journals that don't).