[tor-dev] Effect of padding on end to end correlation false positive rate

[dan at geer.org]

While considering whether A & B are alternate personas
of the same person, you might like this paper

"De-anonymizing Programmers via Code Stylometry"
www.princeton.edu/~aylinc/papers/caliskan-islam_deanonymizing.pdf

as seen at the 2015 USENIX Security Symposium.

--dan