Twenty years ago, being "published" meant something. Now, you can polish turd theories until no one but a scientist working in the field can refute them. Protip: Proceedings of the 8th International Symposium on Nuclei in the Cosmos isn't too great of a scientific journal.
Nature and PNAS publish bullshit, too, on occasion. I'm hopeful that the whole journal model falls apart in the next decade or so in favor of an all open access system, hopefully one brokered by NIH/NSF or perhaps the PLoS system. Subscription journals are a relic, a gift to industry, and totally out of date in the digital age.
I think progress is inevitable, but the publishing companies still maintain their grip on science for now. The other day I had to send out engineering drawings for signatures... physically print out E-size drawings and take them around the building. Why? Because old men, that's why. We will migrate to digital signatures, just as peer review will migrate to a digital review format, but neither will happen as soon as they should (now).