4 ms·
Naively, this seems analogous to differential privacy, where the distribution of the result is the same as the real one, without revealing the individual elemen
by motohagiography 7y ago
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.
- nullc 7y agoI 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.
- ketzu 7y agoDining 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).