My incomplete understanding is that publishing in big-name journals provides prestige and improves funding prospects for academicians. Since academia is very competitive, researchers will do whatever it takes to publish in the most presitgious journals. In other words, they provide a brand that researchers want to associate with; analogous to how rappers mention luxury brands (sometimes with but often without being paid) in their songs.
I don't know what benefits reviewers receive, but they are gatekeepers to the journal's brand, so conceivably they are able to obtain some benefit to themselves.
As a reviewer, you get to read relevant new research in your field several months before it gets published. This doesn't work in physics and maths, thought, where the whole field has the habit of pre-publishing manuscripts in arXiv, so everyone gets to read everything before it's published.
Is there a reason arxiv doesn't directly manage what would be an analog to paper reputation?
My impression is that journal selection is doing some work of signaling how awesome scientists think a particular paper is, either actually or aspirationally, and so is capturing some sense of group regard. It seems like just keeping track of views, downloads, and "likes" on arxiv might serve much the same function although would clearly require a lot of work to get right to be credible.
Overlay journals and the reputation graph as the result of citations. The more reproducible your work is, the higher the reputation should be. Part of getting an undergraduate or masters degree should be in reproducing research.
More easily gamed, I'd think. Also, if you remove the curators (journals), then people looking for research in the first place (who are driving the ones arxiv stats) won't know where to look first.
I don't know what benefits reviewers receive, but they are gatekeepers to the journal's brand, so conceivably they are able to obtain some benefit to themselves.