OpenLibrary (relatively less known project by Internet Archive) is doing good work on collecting identifiers for books and works, which LearnAwesome leverages. But if you check in app/utilities/book.rb, you will notice that we have our own data enrichment pipeline to work with multiple sources.

Goodreads is owned by Amazon, and a lot of the data in various digital sources (both real and correct and random or mistaken) seem to come from people listing physical books for sale through Amazon. (Library catalogues are another common source)

The ISBN codes seems to be in some weird legal gray area.

The codes can be broken down so you can tell (to some degree) which language, geographic region and publisher was involved and whether the ISBN is valid (as in meets a checksum constraint) and has been "issued" (a continually updating list) managed in a decentralised manner but since it's an older system, it wasn't ever really designed for this kind of usage.

It also hits the problem, that these days you care a little less about the actual physical object, which ISBN was designed around, but certainly a trove of data hidden there.

If you mean ISBNs in general, that's not available. However plenty of libraries around the world do publish their own list of ISBNs, and you'll find dozens of datasets and search engines that combine those into one.

You're gonna have to move away from western languages entirely to find a book that's not available in one of those.