One thing to think about is what happens to academic fields that are not well funded? Plan S makes sense for funded research in STM disciplines. If we accept that publishing does cost some amount of money then it seems reasonable to have the funding of the research also cover that cost. Obviously we can argue about what a reasonable cost is ($100? $500? $5,000?), but it's certainly non-zero (even though yes, you can find specific examples of zero-cost publishing, but not on a scale that works for the system as a whole).

But what about the humanities and social sciences, which are typically not funded by government or foundation grants? We currently have a system in which the expectation of the academics is that they can publish for free because the universities pay for that cost via subscriptions. Changing to an author-pays model, which Plan S seems to push the industry toward, doesn't work for a lot of academic fields. There are certainly alternatives, like university libraries converting some of the funding they currently use for subscriptions to cover publication costs, or entire governments covering all publication costs for every academic within their borders. But it's not as easy to see exactly how the non-zero cost of publishing is covered outside well-funded disciplines.