Re: CWD: "Jacking in from the "One that Got Away" Port

At 10:19 PM 6/3/96 -0500, Declan McCullagh wrote:
>(By Brock Meeks / brock@well.com / Archived at http://www.cyberwerks.com/)
>
>
> CyberWire Dispatch // Copyright (c) 1996 //

> A virtual tactical nuke was hurled into the arcane subculture of
> encryption technology Monday when RSA President Jim Bizdos revealed
> that his company's Japanese subsidiary had developed a monster chipset
> capable of scrambling voice and data real time with a so-called "key
> length" of up to 1024 bits.

I don't mean to be overly critical, but why not provide for the possibility 
of longer keys?  The RSA key only has to be exchanged and decoded once per 
call, presumably, which suggests that there shouldn't be a problem to do 
2048-bit keys.  Admittedly, 1024 bits will be good for many years, but...

> That key length stuff is just so much gibberish to those playing
> without a scorecard, so let me drill down on it for you.  Basically,
> the longer the key length, the harder it is for a message to be broken
> by "brute force" automated attacks.  Current U.S. laws prohibit the
> export of any encryption device with a key length longer than 40-bits,
> or roughly the equivalent of  Captain Crunch decoder ring. For hardcore
> math types, I'm told that a 1024-bit key length is 10 to the 296th
> power more difficult to break than 40 bits.

I sure do wish they'd get things like this a bit more accurate...  Oh, well, 
I suppose it doesn't really matter...


Jim Bell
jimbell@pacifier.com

Yeah, we fucked up here and are getting rightfully flamed.

A more accurate way to say this would be something like:

1024-bit RSA is as hard to crack as three nested iterations of 56-bit DES.
The key length for symmetric-key cryptosystems isn't comparable to the key
length for public key cryptosystems. 

And now that I'm at it, I'm now told that it was a Captain Marvel decoder 
ring.

Apologies, all.

-Declan


> > That key length stuff is just so much gibberish to those playing
> > without a scorecard, so let me drill down on it for you.  Basically,
> > the longer the key length, the harder it is for a message to be broken
> > by "brute force" automated attacks.  Current U.S. laws prohibit the
> > export of any encryption device with a key length longer than 40-bits,
> > or roughly the equivalent of  Captain Crunch decoder ring. For hardcore
> > math types, I'm told that a 1024-bit key length is 10 to the 296th
> > power more difficult to break than 40 bits.
> 
> I sure do wish they'd get things like this a bit more accurate...  Oh, well, 
> I suppose it doesn't really matter...
> 
> 
> Jim Bell
> jimbell@pacifier.com
>

Declan McCullagh writes:
> The key length for symmetric-key cryptosystems isn't comparable to the key
> length for public key cryptosystems. 

You should have stuck with that.

> 1024-bit RSA is as hard to crack as three nested iterations of 56-bit DES.

Unknown. Cracking 3DES and 1024 bit RSA are both hard, but the
algorithms used for brute forcing both of them are very different. I
would say that making comparisons between them is probably in general
a bad idea, especially given that over long periods of time the
techniques used improve at different rates.

The conservative attitude is, in any case, always "encrypt until it
hurts and then back off a little bit."

Perry