Self-archiving refers to an author depositing their work in an online database for open and unmediated access. The barrier to entry is extremely low: an account with the database-keeping organization is all that’s necessary in most cases. This makes it very easy to share all research, regardless of content, scope, or importance, with interested readers.
The problem is that, unlike classic journal publishing, there’s no gatekeeper in self-archiving. (Although some organizations subject content on their servers to certain conditions like topicality.) What if bad science gets in? What’s to stop unethical researchers from compiling spurious research to support financial interests: for example, to promote a certain pharmaceutical that is profitable yet poor in efficacy? Negative consequences don’t even have to be intentional: methodological mistakes and sloppy errors can slip through, even in work with the most honest of intentions. If you cite a factually incorrect or poorly supported study in your own research, it can severely damage its credibility, not to mention waste your time.
A related approach that combines a low barrier to entry with scientific rigor is post-publication peer review. The practice is best summarized as “publish first, ask questions later”. Authors submit their paper for publication in a journal: after a cursory check for topicality, the paper is published in its original form. Later, the paper is subjected to scrutiny from readers given special permission by the journal or database, or even opened to commentary from the whole community of readers.
This approach has many advantages. Your paper gets input from all interested readers in your field, rather than 2-3 ordained peer reviewers. More input means more insights into how to improve your study; it means the most interested members of the community will act independently to criticize or defend your work, rather than a couple of peer reviewers who are responding to an unpaid request from the journal. Finally, by reducing the time taken for a research discovery to be published, your findings can be disseminated in the field immediately: there’s no risk of your paper being published after other researchers have moved onto something more topical.