Re:Building supercomputers for autocrats probably isn't good...
by jasonfarnon on Monday June 09, 2025 @12:27AM
(#44221406)
I don't know about the original post being referenced, but the ground truth could be known to good approximation. I assume HN has a good idea of which accounts are the same person using IPs, and could actually evaluate the false negative rate.
How do you know it didn't miss 10x more than it found? Like, that's almost definitionally unprovable.