[betterunix2]

Historically scientific articles were distributed by printing many copies and sending those copies to university libraries, which acted as repositories of human knowledge. Early on scientists realized that it made sense to collect high-quality articles and publish them as bundles (journals) at regular intervals; the peer review process was developed to determine which articles would make the cut. Printing is an expensive and was historically a labor-intensive process, and someone had to pay for it. The arrangement that was ultimately settled on was that the publishing companies, which were often operated by a particular university library, would charge subscription fees to those who received copies of scientific journals.

TLDR: it made sense as recently as 50 years ago.