I signed up as a reviewer with the electronic Journal of Health Informatics and after a few weeks received my first request to do a review. The eJHI uses the OJS e-journal system and the review workflow process is largely automated.
The editor selected me as a reviewer and the system then sent me an automated message inviting me to review. I clicked on a link in the message and was taken to the journal web site, where I clicked to agree to do the review. I then downloaded a copy of the article for review, and looked at the offered guidelines on what to do.
After I wrote my review I uploaded it to the web site. Shortly after I got a message from the editor complementing me on the quick response. This all worked very smoothly and quickly. Perhaps I should not be surprised when an automated system works, but it was a relief.
There is, of course, a danger that the speed of the automation will encourage quick reviews and lightweight papers for on-line journals. But that is something which can be countered in other ways. As an example the OJS system has provision for the editor to rate the each review. Also some form of public assessment of the papers by the readers, perhaps after a traditional review process, is possible. OJS doesn't have a way for readers to rate papers, but they can post comments.
One way to counter the virtual nature of on-line journals would be to associate them with events. In the paper publishing world conference proceedings are separate to journals. Attempting to incorporate the papers from conferences in a journal is difficult as journals have regular fixed deadlines and page limits. Conferences need their papers ready to suit the conference and to be as big as necessary. Using e-publishing remove many of the constraints.
OJS have a companion OCS for conference publishing. But it is designed for one-off events. What is needed is a blending of the two. I have been doing a little of this by changing the wording of the OJS to suit conferences. Some of this can be done using the internationalization features of the system, without changing code.