It is also valuable for scientists as it is often a 'directors cut' version of the paper. Journal submissions are heavy edited and shortened to fit into the page limits.
In CS, unlike in many other fields, the best work comes out as conference papers, not journals, and they do have strict page limits. Most arxiv papers I read become conference papers, not journal papers.
ACL conference papers usually cap the main body at 8 pages, with unlimited appendices. KDD has a hard limit of 9 pages plus 3 for references and appendices.
look at essentially any proceedings of any conference (in crypto we dont really do journals). see EUROCRYPT for example https://link.springer.com/book/10.1007/978-3-031-91098-2 in there, every paper will be cut down and referring to full version for proofs etc. which are typically on eprint.iacr.org
We usually do conferences in cryptography/security, and most of them have page limits: CCS, USENIX, NDSS, S&P, CRYPTO, EUROCRYPT all have page limits (some allow appendices, which reviewers are not obligated to read).
When that’s the case, the preprints would be just as short. We don’t really like unnecessary pain so we write short manuscripts from the beginning, if we plan to submit in such a journal. Usually, the longer versions get published somewhere else anyway.
By the way, one of my favorite pastimes is to download the latex source for papers on arxiv and read all the commented-out stuff.
% we should make sure this theorem is actually true