Lots of people upload their research to ResearchGate (even though I really dislike the site for other reasons), make it available from their research group's web page, or in some fields upload it to ArXiv or BioRxiv. It's usually not against the terms of paywalled journals to do any of these things. Google Scholar or even the regular Google search engine can be decent at finding these PDF links. As a last resort, if you're serious about wanting to read a paper, you can contact the corresponding author, who will usually email you a copy and be glad you are interested in his/her work, though this becomes a bit impractical if you need access to multiple references.
Good points, but there is another use case: AFAIK (and in my limited experience), much of research involves reviewing many papers, to find the right one and to glean information here and there. It's not always a hunt for one specific special paper.
How is that use case addressed when the cost of finding each paper seems to be so much higher?
Many people work at home now, even if they work at wealthy institutions
That's not really the issue. You would set up a proxy, log with your institutional account... actually even with access the sci-hub UX is vastly superior.
Almost all papers are legally available from the researchers’ websites anyway. I’m not sure it’s the big deal everyone makes it out to be in the first place. I very very rarely find a paper I want is paywalled in the first place.