The "without affiliation" aspect is a dealbreaker in many fields. Lots of journals won't even acknowledge you if you're not at an academic institution.
It's not a dealbreaker in computer science, where the norm is for blind peer review. And in any case, you can just list your undergrad institution or a company.
Actually, I think you'll find there are sufficient number of mid or lower tier journals that will send these out to referees. I've refereed papers from non affiliated researchers before. And even Nature will send you an rejection letter.
Most journals in the fields dominating the arXiv (physics, math, CS, and probably more) allow you to submit to the arXiv concurrently. So unless you refuse out of principle, you can both get free refereeing and release the paper open access.
I'm aware that most subdisciplines of physics and CS are sympathetic to arXiv, but that is most certainly not the case in bio and chemistry, although the rise of biorxiv may start changing things.