Re:A case study in PDF forensics: The Epstein PDFs
by Eisenstein on Wednesday February 04, 2026 @02:30PM
(#46890471)
If you have a large enough set to test against and a specific person you are looking for, this is totally doable currently.
[ Parent
| on HN ]
Re:A case study in PDF forensics: The Epstein PDFs
by fluoridation on Wednesday February 04, 2026 @03:14PM
(#46891004)
Of course it's doable. The question is how reliable the results are.
Re:A case study in PDF forensics: The Epstein PDFs
by ted_bunny on Wednesday February 04, 2026 @05:45PM
(#46892968)
It just needs to find the needles in the haystack. Humans can better verify if they're truly needles.
Re:A case study in PDF forensics: The Epstein PDFs
by andai on Thursday February 05, 2026 @04:39AM
(#46897744)
I wonder if it works on zoomers too. I have noticed a slight mode collapse among this population ;)
Re:A case study in PDF forensics: The Epstein PDFs
by Cthulhu_ on Friday February 06, 2026 @03:40AM
(#46910481)
Not to mention, I'd argue that most people have a (subtly?) different writing style depending on where they post and to who they talk to.
Re:A case study in PDF forensics: The Epstein PDFs
by hansvm on Wednesday February 04, 2026 @06:01PM
(#46893136)
Not just a test set, but enough of a set to search through and compare against. Several pages of in-depth writing isn't anywhere near sufficient, even when limiting the search space to ~10k people.