The current publication system is very odd. Mostly government money is being used to fund research, then professors are asked to review a manuscript (who may just hand it off to a grad student or postdoc) without any compensation for their time, and the final result then locked behind a paywall of some private journal.

It seems that the incentive of a journal is to maintain an appearance of legitimacy, rather than actually enforcing it. This is why, as the article mentions, journals tend to be fairly quiet about retractions and issues of misconduct.

There is a better incentive by those that are funding the research (the taxpayers/government) that their investments are resulting in legitimate works. This also goes in line with the idea that these final manuscripts should be freely available to the public. Now with that said, I also acknowledge that the idea of having the NIH, NSF. etc. operate the editorial and review process would be nightmarish.