Traditionally, the peer review process is one where neither party knows the identity of the other (double-blind peer review), but this is lately being challenged with more open reviewing systems such as in BioMed Central journals where identity is not only know, the reviews are published alongside the papers. BioMed Central have prodiced a rigorous system that is leading to the publication of high quality papers but, at the extreme end of the online publishing industry, there is post-publication review where manuscripts are published in whatever state they are submitted, and reviews follow online. This strikes me as a very public way of making your mistakes and, of course, if the reviews do not lead to revision of the articles, then this process - unlike more 'traditional' reviewing routes - does not lead to improvement of the published articles.
Therefore, the routes for reveiwing include:
- Open reviewing - the identities of reviewers and reviewed are mutually known
- Single-blind reveiwing - the reviewer knows the identity of the reviewed (usually) but vice versa is theoretically possible
- Double-blind reviewing - neither the reviewed not the reviewer know the identity of the other.
Within all of the above the variations include the number of reveiwers - clearly, the bare minimum is one, otherwise review has not taken place, but two is common and some journals request many reviews - I think the record I have heard in a North American psychology journal is around 10. In addition, some journals ask authors to suggest reviewers - some requesting that they are not pre-warned by the author; others requesting that they are to avoid asking reviewers who will not respond. Journal may use these reviewers exclusively; others will use them - or some of them - and also some of their own, thereby introducing an element of uncertainty about who the reveiwers are.
If I have not exhausted the possibilities here then please leave a comment.