Changing preferences

[Robert J. Hansen]

Faramir wrote:
> So probably it is a mistake to try to explain in a logical way
> something that is, by definition, non based on logic.

I don't have any problem with people having their own personal likes and
dislikes.  I like Blowfish; I use it, although I don't recommend it to
others.

I have a _big_ problem with people arguing that their personal
prejudices are actually reasonable conclusions to draw.  Like Mark Twain
said, "a lie can be halfway around the world before truth has pulled its
boots on."  In the internet era it's even faster.

When people who sound like they know what they're talking about say
things that are not factually true, newbies remember the sound bites a
lot more than the facts.  The facts: 3DES is ugly, slow, and the most
trusted cipher in the OpenPGP arsenal.  But from the way you're talking
about it, it's a nightmare of engineering rather than a triumph.  Which
do you think the newbie will remember?  Which do you think they _should_
remember?

> When people is lazzy and don't want to spend time and energy to make
> a proper solution for a problem, and just take what they have and
> adapt it in a sloppy way, they use to get solutions that are bulky,
> non aesthetic, and with a lot of disadvantages... a bit like you
> described the performance of 3DES.

What disadvantages?  It's slow.  That's irrelevant for most OpenPGP usage.

Also, if you really want to call Don Coppersmith and the rest of the DES
design team 'lazy,' well, go ahead, but expect a lot of people to look
at you funny.  Don Coppersmith is widely considered to be one of the
brightest cryppies ever -- he ranks up there with Abraham Sinkov.

> But when I saw an article about 3DES, and I understood (or 
> _misunderstood_) it was just to apply DES 3 times, that arose the
> same _feeling_ that I feel when I see a sloppy job.

That opinion puts you in an enormous minority.

> Once I saw a shelf attached to the wall by no less than 24 screws. 
> When the shelf was removed, the wall looked like it had been attack
> with a screw-shooting machine gun. Sure, the shelf was firmly
> attached to the wall, but it would have been better to use bigger
> screws, or maybe to add "legs" to it to support its weight. Or maybe
> some other solution.

And when you look at the Roman Coliseum, do you think "gee, they really
overengineered that, the design must be lazy and shoddy, and this
doesn't look anything like an I.M. Pei or a Frank Lloyd Wright design,
it's ugly"?

The fact is that the surviving buildings of antiquity have taught us a
great deal about engineering.  They are still deserving of respect, not
to be written off as sloppy and aesthetically unpleasing work.

> But it is not the same than to discover a painkiller can also reduce 
> the risk of heart strokes.

In DES and 3DES's case, this is almost exactly what we're talking about.
 DES was the cipher that allowed us to discover differential
cryptanalysis, for instance.  Essentially every single attack that's
been devised in the last thirty years was first tested on DES and
discovered not to work.  Then they went to apply it to FEAL, MacGuffin,
or any of dozens of other ciphers, and watched it destroy them.

> I don't know if the article I read was not clear enough about that 
> point, or if I failed to notice it. If 3DES is the application of a 
> theory that was not considered before, then it is not what I thought
> it was. If 3DES is built using DES, as a wall is made using bricks, I
> don't have anything against that.

Both are correct.

3DES is an application of a cryptographic theory which did not exist
prior to DES.

3DES is built using DES as a building block.

> Yes, but I figure these patches are carefully designed to solve the 
> problems without causing new ones. And if a bug is introduced, there
> are efforts to remove it ASAP.

Yes.  Just like 3DES.