Re: Delivering on talk

From: Jean-Luc Cooke (jlcooke_at_lager.engsoc.carleton.ca)
Date: 12/31/03 

Date: 31 Dec 2003 02:53:16 GMT

Michael Wiener <michael.RemovEwiener@thesesympatico.cawords> wrote:
> The final collision search will be a little easier if you report the number
> of steps taken to the DP.

Thought about this a bit more.

Number of steps between DPs would need to be fairly large. The average
distance would be 2^32 for this implementation. But I'd need to
account for larger values.

We'd expect to need 2^32 DPs before we find a collision
 2^32 * 2^32 = 2^64 = sqrt(2^128)

2^32 * 4 bytes = 16GB of extra space.

A DP search would take at worst case, 4*T time to find (T= time to find
one DP). It may be worth just paying the price later.

Unless you can think of another use. CounterScriptKiddie tactic perhaps?

JLC 

--

Relevant Pages

  • Re: Delivering on talk
    ... >> The final collision search will be a little easier if you report the ... > Number of steps between DPs would need to be fairly large. ... I agree that the savings may be small (unless you receive a lot of bogus ...
    (sci.crypt)
  • Re: Delivering on talk
    ... Michael Wiener wrote: ... > The final collision search will be a little easier if you report the number ... JLC ...
    (sci.crypt)