News for hackers. Stuff that matters.
Re:Dining Cryptographers Problem by motohagiography on Sunday May 03, 2020 @07:41PM (#23062831)
Naively, this seems analogous to differential privacy, where the distribution of the result is the same as the real one, without revealing the individual elements - or is the dining cryptographers a full proof? I couldn't quite tell on first read.
Re:Dining Cryptographers Problem by nullc on Monday May 04, 2020 @03:17AM (#23065359)
I think your question is a little too vague.

In general case DCnets provide perfect sender privacy (among participants, of course). If there are multiple messages their senders will be completely unlinkable.

Specific implementations might use schemes for collision resolution that break privacy, though there is no requirement that collisions (absent malicious participants) even be possible at all-- with fancy enough technology applied.

Re:Dining Cryptographers Problem by ketzu on Monday May 04, 2020 @10:34AM (#23068527)
Dining cryptographers and their variants are usually fully proven in the sense that when all but one input is zero, you will receive the intended input. You always receive the combination under the operation you are using (i.e. xor of all messages).