The short answer is because it is a way for someone to make money. Not trying to be trite, but it is the same rationale that is used to pay ICANN for domain na

The short answer is because it is a way for someone to make money. Not trying to be trite, but it is the same rationale that is used to pay ICANN for domain name registration or GoDaddy for an SSL certificate: They are perceived as an authority (ICANN exclusively, GoDaddy as one of an elite few). It's possible that arxiv.org would be the eqivalent of Let's Encrypt, both lack the "extended verification" of the paid service. For reputable journals, the peer review process is far from trivial. This last year, I reviewed ~8 computer science papers. I only recommended 2 for publication. For all of them, however, I spent several hours going through the paper, the references, the algorithms, mathematics, experiments and methodology, etc. For all of them I gave a detailed writeup to support my recommendation. For all of them, there were 3-5 people doing the same. We are all experts in the field. Most submissions are rejected, but the review process is the same. The authors (who worked several months on the paper) will probably take a failed submission and rewrite parts of it, taking the suggestions and criticisms into account, and re-submit it elsewhere, where the same effort is repeated by different reviewers. All of this is to say that there is a lot of overhead to get a paper published. Some journals charge a publication fee if your paper is accepted. If the journal is an open-access journal (which is what you are looking for), then the fee can be $2,000-$3,000! Please note that I did not get paid a single cent to serve as reviewer for any of the papers. The money goes somewhere, but not to me, and not to the authors. Finally, being in CS and having a background in server management, I can honestly say that the technology side is not that expensive, either, so that is not a valid justification. It should give everyone pause: Why does a conference cost $1,200 to attend (not including the hotel stay or transportation) when the research presenters don't get paid anything? Why does it cost such exorbitant fees to publish a paper? Why does it cost so much to buy access to a single paper? The authors and reviewers don't get paid any of the money. Somebody is making money on it, though.