From: doug@netcom.com (Doug Merritt)
Date: Wed, 13 Oct 93 09:42:03 PDT
To: cypherpunks@toad.com
Subject: Re: Breaking DES
In-Reply-To: <pmetzger@lehman.com>
Message-ID: <9310131642.AA01905@netcom4.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain


>When I use a hash table, it is never a substitute for storing the
>actual value of the thing I'm hashing. Its always just a way of
>rapidly FINDING the underlying object. I have to store the underlying
>object in order to compare to it. As an example, in a hashed symbol
>table, I store the actual symbols.

Instead of storing the underlying value, I am assuming that it is *thrown
away*, and recalculated whenever there is a collision. This cuts down
on the expense of this disk drives, at the cost of increasing runtime
by roughly a factor of 7 (log_base_256(2^56)). I didn't include that
cost in my previous estimates because I was doing a very rough back of
the envelope calculation, but I accept that it should be included.

>> Impractical? Your response to Karl implied that it was *impossible*.
>
>The two are very similar in our field. Cracking RSA with a 2000 bit
>key is merely impractical, not impossible, where "impractical" is
>defined as completely beyond human ability.

It's a question of where you draw the line. A budget of one hundred billion
dollars and a runtime of say a year, I'm willing to call "impractical".
A budget of 10 trillion dollars and a runtime of 100 years, I'd be
willing to call "impossible".

The 2000 bit key is over everyone's threshold.
	Doug