They're pretty bad for meta-scientific research. That is, they are very much a for-human-eyeballs-only format.
Agreed, I've download parts of the sci-hub archive and want to analyze the articles but I can never even get a decent solution for extracting multilingual text. I've tried the well-known open source tools for this and none have really been satisfactory.
Try narrowing down to papers in English only ? Why try to deal with a problem which has no feasible solution for now ?
I hate PDFs. Hard to read, don't own a printer, weird layout.
I hate hating PDFs. Hard to not read, don't not own a printer, unweird layout. Disagreeing is fun. Did I do it right?
I actually meant what I said, though.
HTML (in some reasonable form) is my favourite format for reading text.