As far as I know, there are 4 distinct ways to publish articles;

 

Traditional journals; peer-review (behind closed doors), the authors lose some rights by publishing, access is limited. Publishing takes months.

 

Open access journals; The same as traditional journals, but the access is not limited and sometime, the authors keep more rights. Publishing takes months.

 

arXiv; no peer-review but a system of "endorsement".

 

PLoS One; open peer-review process, articles are never rejected because of perceived lack of importance, they're rejected only if the content is not scientific (i.e.: creationism). It's easy to write answers and criticism about an article. Publishing is fast. The authors keep all their rights.

 

Many scientists in the past had troubles publishing because their work was unorthodox, it's the case of Feigenbaum, Pontryagin, Belousov and many others... Also, the ability to access information and to criticise it is, IMO, an important aspect of science. In traditional journals, answering to an article is sometime so difficult that a scientist will need to find a way to fit a small, but sometime important remark, in a stand-alone article.

 

It seems to me that PLoS One is much closer to the ideal of science than anything else because.

I have to edit my post, as I was referring to the wrong PloS journal. However, quite a number of papers are being rejected even in PloS one due to the same reasons as in other papers. The things, the editor's name gets published, but not necessarily the other referee's he/she has been consulting. Lack of importance (or lack of novelty) is quite a common reason not to publish something for basically all journals.