This is unofficial, please do your own math, but it should give you a good idea:

thanks for this. the final values presented here reflect a chain of 8 MN's, correct?
but the default for DarkSend is set at 2 (or less than 8 anyway). is this also correct? obviously 8 is exponentially better than 2.
Please correct any mistakes. thanks.
yes,
probability first row with 2 rounds would be ~0.038^2
second row ~0.057^2 and so on..
if that is too risky for you, you can set it to 8 in preferences
I'd argue the min MN chain should be at least 4 long.
Assuming (with medium paranoia settings) that the "bad guy" has a handle on 10%-20% of MN's (100-250 nodes), then with a 2 MN chain length the above numbers predict between a 1%-4% chance of a de-anonymization. This is far too high IMO.
I have found the last few pages of questions/answers/debate really useful. Please don't stop posting that kind of thing - it is very stimulating compared to the troll-wars here a while back.
in pursuit of anon
