> Because that doesn't count. Research has to be published in a peer-reviewed journal (or at a peer-reviewed conference) or it doesn't exist.
Well... true.
> You don't get credit for it,
Yup.
> it never gets cited or used by other research, it doesn't become part of the literature.
Nope, not necessarily, depends on the field. I see more and more citations of even blog entries. Some have more citations than the best paper anyone I met personally wrote. Some "not-papers" become cornerstones of entire branches of research, although they tend to be later replaced by real papers by the same author (with nothing new in them). Note that not all publishers will even accept citations like that.
Still, these citations do not count - at least where I leave. You may be the author of the most influential text in the field in years. Your results may have been replicated by multiple peers and cited by almost everyone who matters, making you a real celebrity. However, it's not in a journal on the ministerial list. It is not indexed by the oficially endorsed database (mostly Web of Science here). It gets cited, but these citations do not appear in the database. So, your official parameters (like h-index) are unaffected. It is entirely possible to be - at the same time - one of the most influential researchers worldwide and a deadweight for your institution, lowering its total score. Also, grant proposals you submit will get lower scores, because you're "not influential enough", you're unlikely to produce anything worthwile.
So, you're basically right. You will swallow any restrictions imposed by the publisher if the journal is good enough and wants to publish your paper. Because your evaluation is not based on your real achievements - that's too subjective. It is based on artificial scoring, in which some peer-reviewed journals, mainly from the largest publishers, simply rule.