Be honest, do you not go by journal title when you have to choose whether or not to read a paper? It's an imperfect metric, but it's the one we have. 25 years ago I attended a talk (by a representative from Pergamon) about the future of scientific publishing. It was out of control even then, and when in the Q & A session someone suggested ditching journals altogether this Pergamon (!) guy made the same remark: do you not read the journal title? Pergamon, whose best-known journal was Tetrahedron. Pergamon. Uttered by a Pergamon guy.
In some fields arXiv may have improved matters, but most fields still rely on journals.
Sadly, I did go by journal title, but not the way you think. In antibiotic research and microbiology I finally arrived at the point where I assumed anything published in Nature, Science, or Cell was likely to be so fundamentally flawed that the only reason to read it was so I could explain why we shouldn't waste our time in journal club with it.
Lately I have been questioning whether or not we need large-scale peer-reviewed journals anymore.
For example, if I knew that Leonard Susskind read a paper and cited, shared, liked, or bookmarked it, then that could easily replace my fundamental need of having the paper stamped by a trusted authority. If scientists could share, cite, and discuss papers on an open source platform (like Facebook, but for Academia), I believe that we wouldn't need these gatekeepers anymore, and they would become obsolete. The real authority has always been the scientists who are doing good, promising research, not some contrived group of scientists who are solely reviewing papers for the sake of assessing their quality.
An (possibly flawed) analogy: do need CNN/ABC/FOX when we have YouTube, Facebook, and Twitter to propagate news?
In some fields arXiv may have improved matters, but most fields still rely on journals.