I agree such meta data should be captured, but if the papers are converted to IPFS, it'll be easier to copy, ship, and then re-serve the data at end points.

Think of it as a [sneaker|dark]net CDN enabled by content addressable storage.