Traditional academic journals are "closed" in the sense that access to them is only available to subscribers, or through private research libraries. Because academic journals are specialized, numerous, and expensive the new ideas in these journals are only available to large-budget institutions and professionals. This results in the isolation of different fields and the ideas discussed in those fields, and the de facto control of ideas by institutions, many of whom have a vested interest in guiding research and maintaining the dominance of past discoveries. That is, they have a vested interest in limiting information, maintaining control, and encouraging specialization.
Submissions to closed journal are either accepted by an editor who knows your work, or submitted to volunteers advertized as peers who decide whether it's ready for publication based on familiarity with you, or your point of view. Familiarity assures acceptance, unfamiliarity scrutiny by peers of the readership who, in my experience, have limited expertise.
Different closed journals compete for readers from the same group of similarly trained professionals. Since these journals don't share rights to their publications with other journals the only way to reach a larger audience in your field is by having your work endorsed, referenced, and discussed by those authorities in the field who set the research agenda. This puts the whole research endeavor back under institutional purview.
In my current research into the psychological condition known as Attention Deficit Hyperactive Disorder I have learned how norms of truth in psychology differ from those in other fields. Psychiatry, for example, has never had a basis on which to judge the merit of one theory versus another and, as a result, has pseudo-scientific standards that cater to social attitudes, political trends, and economic interests. Psychiatry is less a field of medicine than a taxonomy of attitudes to facilitate social management.
I have seen similar differences in other fields in which I've worked. The "hard" sciences share similar standards, tools and language so that even though astronomers don't seek mathematical proofs and mathematicians pay no attention to observational evidence the two groups can collaborate. In contrast psychology, medicine, divinity, economics, and history have different standards of argument, demonstration, and discussion. What passes for a definitive argument in economics, or an objective conclusion in medicine, or a substantive conclusion in history are incommensurate. These fields have such little contact that they can espouse views that predominated hundreds of years apart, and still experience no conflict. There is not one Ivory Tower, there are many: a separate one for each specialty, and they are disconnected.
Open journals have arrived with the internet. They are driven by the cost-free creation and distribution of content, and the fine-grained cross-indexing that the internet makes possible. Traditional closed journals are trying to migrate to the web but persist in offering content at prices that are affordable only to private institutions. Open journals, being free to everyone, blow this model to pieces.
I don't yet know the full open journal terrain. I know there are some open journals that focus on specific fields, and others that publish articles in a large number of fields. Some review submissions before publications, others do not. For instance PhysNet (http://de.physnet.net/PhysNet/journals.html) lists about 70 open access journals that focus specifically on physics and related topics worldwide.
The Directory of Open Access Journals (http://www.doaj.org) lists over 8,000 free access journals world wide, organized by language and specialty. InTech (http://www.intechweb.org/journals.html) describes its mission to "break down the traditional barriers to becoming published and provide authors with new choices and an equal opportunity to share their ideas and the results of their research with the global scientific community."
http://www.researchgate.net presents a kind of hybrid effort: they index and collate articles of interest to users but do not limit themselves to open journals. As a member, this site will present to me all recent articles that match my criteria of interest, but some (many?) of these will be unavailable to me in their full form. That's not bad given that one does want to know what's being done even if you can't get at it.
Today I found myself at the page of PLoS ONE, a publisher of medical research that offers thousands of free, current research articles. Curious, I asked it to show me current neuroscience articles and, of the thousands that it returned, I scanned through the titles of the first 150. Here are the ones that caught my eye:
Dopamine Modulates Metabolic Rate and Temperature Sensitivity in Drosophila melanogaster,
Towards the "Baby Connectome": Mapping the Structural Connectivity of the Newborn Brain,