The traditional procedure is that scientific work is published in
commercial journals, for which one must pay (just as you normally pay
to buy magazines or newspapers). (A user in a library that subscribes
can read the journals, but scientific journals are usually only in
university libraries, and thus not necessarily convenient.)

Even now, with many/most journals available online, one has to
subscribe or pay by the article.

There is a movement toward a new system, in which publishing is
considered an intrinsic part of doing the work -- with the view that
people (whether scientific researchers in places without a particular
journal or the general public) should be able to see the work. A
particular case is when the research is publicly funded, as with NIH.
So, NIH is proposing that work they fund be made publicly available.
(There are some qualifications.)

This is intended as a brief description of the issue, without any
substantial discussion of arguments about it. But the general trend is
toward more open access.

bob