Paper X, on the ArXiv, keeps getting rejected. Years later, Paper Y comes up and makes progress on X, or does something closely related to X. Y may or may not cite X. Y gets published. Now X cannot get published because the referees do not see what the contribution of X is, given that Y has been published and that in light of Y X is not new.
The solution in my opinion, following a series of earlier posts the last one of which is this, is to move the emphasis away from publication towards ArXiv appearance. Citations should refer to the first available version, often the ArXiv one. Journals/conferences can still exist in this new paradigm: their main important job would be to assign badges to ArXiv papers.
Obviously, this plan does not work for the entities behind current journals/conferences. So they enforce the status quo, and in the most degrading way: by forcing authors to fish out, maintain, and format useless references.
It seems that the Hokuto No Ken videogame that should have been made decades ago is finally being made. Thanks to Marco Genovesi for sending me this link. (More about Marco later on this blog.)
I consider watching the Hokuto No Ken series (excluding the more recent garbage) one of the most significant artistical experiences of my life, something that also makes me understand how can some people be so passionate about Dante or Homer. And, if you grow up in Italy there is a special treat for you. You can watch a version where the words are dubbed, but the soundtrack and the screams are from the original Japanese. By contrast, the English-speaking audience can either watch the Japanese version with subtitles — and I always hate subtitles — or they can watch an English dubbed version. I once happened to get a glimpse of the latter and I was horrified: The masterful soundtrack has been replaced by a very cheap synth, not to mention the screams. Compare this to this.