@lauren We can approximate it, in principle, by the size of Wikipedia plus the size of the Google Scholar cache plus the sizes of the databases of the various journal publishers who lock up their wares behind paywalls that exclude even Google. We can approximate it by counting pages of journals and multiplying by the size of each page, by era—modern, born-digital pages are likely to be much smaller than older, scanned pages, and for that use you wouldn't want to rely on sketchy OCR, especially for all-important equations, tables of constants, etc.
Not that I know what any of those sizes are, but it is, at least, calculable. And in Project Hail Mary, we're dealing with a planet-wide effort on what is literally an existential crisis. (That laptop was one of the less incredible parts of the book, with astrophage being first…)