Because how trivial would it be to modify a PDF maliciously and change data, forge results, etc.?

The same reason why we encrypt traffic using TLS, even if it's "mundane" traffic like tweets.

That's a big ask for a site that's essentially 'morally justified' piracy.

If you want the guaranteed authentic version you should buy it from the original journal.

How do you know the authentic sci-hub isn't doing that?

You don't, though if they were, it's very unlikely they would be doing it in bulk for every file, since someone would notice at some point, either by inspecting the file or notice that a file when downloaded to a canary system misbehaves.

Targeted attacks might be possible based on the file contents or using a side-channel to ID the downloader; meaning only specific targeted users would get the malicious files.

Either way, being able to hold Sci-Hub accountable as this source would be important.

Especially since pdf can embed almost everything ...

> Why would it matter? If you're using it, you already don't care where you're getting the paper from, only that you're getting it.

How can you exclude that case that Alice does care where she gets the paper from, but she simply cannot afford it from the original source?

Or how can you exclude the possibility that Alice wants a version of the paper that does not embed a watermark that makes her or at least the setting (e.g. university, IP address) that she downloaded the paper from identifiable. I can witness myself that Elsevier embeds such information into paper downloads.