[cyphar]

In Physics, I was told that pre-prints were only useful if you had no other way to access the canonical version of the paper (in my university, we didn't have access to a bunch of large journals and there's no way I'm paying hundreds of dollars for papers for a summer research project). In my experience, pre-prints often have different information than that actual published version (though I made sure any papers I co-authored would have identical versions in both).

I obviously found this quite concerning, given that I also work in the world of free software and I know how much of a benefit public access has in a field as opaque as software (let alone something as important to human development as science). But it is the reality, because there is still a view that pre-prints are only meant to be a place to stash away your first drafts of a paper and that the actual journal is where the real paper is stored. Luckily the journal I submitted to (MNRAS) didn't care about submitting pre-prints to arXiv (some do).

(Don't get me started on the fact that some journals still require you to pay them for papers written in the early 1900s. The authors are long-dead, and you've made more than enough commission on Feynman's papers on positrons.)