No question the actual publication costs should be covered.
Which they are when a paper is published Open Access. They are covered once, which is ok, because the publication costs also occur once for the most part. There is no reason for a publisher to be allowed to perpetually charge ~40$ for a pdf download.
One other often mentioned criticism is that universities are forced to subscribe to bundles of journals with a mixture of desired journals fudged with journals no one would want to read. Nominally the subscription fee covers a large swath of journals and papers, but essentially the actual value is much more narrow.
To be honest there is no need for journals anyway in our completely digitized world. If all the papers were published Open Access (paid for once) the users could simply filter for the keywords relevant to them. Why would I subscribe to a Journal? This is technology from the last century.
Arxiv, research gate, scihub— they all seem to work well together with the commercial publishing system.
I think the world without a lot of money going into scientific publishing would be worse off.