That's exactly the difference: the scientists don't get paid from the money you pay the journal. They do the work for free and the reviewers do the work for free. If all the journals somehow disappeared, nothing much would change. In some fields the pre-print is what counts anyways, and you don't pay to download a paper on arXiv.
Your comment makes no sense. No one is forced to publish in these journals. It's a choice. So if these journal did disappear I highly doubt "nothing would change".
The choice of publication venue is driven by prestige. Researchers want to publish in “high impact,” high visibility venues. In many fields those are closed journals, but if those journals magically disappeared the research community would probably just take a bit to figure out a new pecking order, and in that sense nothing would change.
JMLR is an interesting example, where the editorial board of a top journal in the machine learning field basically forked the journal to an open alternative, and of course nobody hesitated to publish there because that’s where the prestige was.
For new fields like machine learning, there are free, open access journals that work perfectly well in filtering the highest quality research.
It is helpful to have top tier journals like Nature, but the value is generated by the editors and article submittors. If we shut down the top journals tomorrow new open access ones would quickly take their place, and the behaviour of their owners is parasitic.
