Re: djbdns-1.05-epoll + speedup patch

["Matthew R. Dempsky" <mrd@alkemio.org>]

On Wed, Mar 07, 2007 at 05:39:20PM +0200, Sami Farin wrote:
> Well BIND seems to do better, according to this.
> http://www.lurhq.com/cachepoisoning.html

Being able to guess the next result from dns_random() with 30%
accuracy given the three previous values would imply a distinguishing
attack with nonnegligible probability against surf_k. 

However, it doesn't seem like the `calprob' program actually guesses
anything; it just estimates how successful some guesser could be.  I
tried downloading the calprob tool, but the link given in the article
is dead.  I'm interested in knowing how the 30% probability was
determined.