Re: PRIV - Crypto-smashing

Hara Ra (harara@shamanics.com)
Sun, 13 Jul 1997 04:20:04 -0700


FringeWare News Network wrote:

> In January 1996, a group of leading cryptographers calculated the time
> and cost to break major encryption codes, using either standard
> computers, or faster, custom-designed, code-cracking machines.
> (list of examples follows)

> > From: jim@hosaka.SmallWorks.COM (Jim Thompson)

> The really interesting thing to watch is that going from 40 bits to 56 bits
> makes the time go up by a factor of 70,000 (with today's technology.)
>
> e.g. adding 16 bits (16 doublings of the keyspace) makes the effort go up
> by a factor of 70,000X
>
Which is about 2^16, which implies since the search time is about
2^keylength, that most of these estimates are based on brute force
searches for keys instead of any genuine analysis. Bigger, Faster, Just
as Dumb as Always.

If we work backwards, the keylength which will resist ALL brute force
searches is computable. We have 10^80 particles in the Universe, 10^18
seconds age of the Universe, assume a clock rate of 10^12 ops/second,
giving 10^110 ops, which is about 2^390, so a 390 bit key will suffice.
Note that this presumes a computer with a processor for every atom in
the universe, running at a teraflop speed, for the age of the universe.

So a 512 bit key should do quite nicely. (and is a power ot 2)

O---------------------------------O
| Hara Ra <harara@shamanics.com> |
| Box 8334 Santa Cruz, CA 95061 |
O---------------------------------O