Good arguments in the article: move to a publish then filter model from a filter then publish one. Why? Peer review needn’t catch all good papers or weed out all bad, it panders to the bias of the reviewers and doesn’t have much upside for science. I need to think on this.