Message-ID: <1074711226.2255.116.camel@coruscant.weisserth.net>
From: tobias at weisserth.de (Tobias Weisserth)
Subject: Who's to blame for malicious code?

Hi Paul,

Am Mi, den 21.01.2004 schrieb Paul Schmehl um 06:53:
> ...
> > The two examples I gave in my initial answer to you actually contain
> > that. I wonder why you didn't comment on them. What's your opinion on an
> > enabled RPC port by default in consumer OSs?
> 
> Precisely the same as my opinion of shipping the OS with inetd running and 
> chargen, finger, et. al. enabled.  It's stupid.

We agree. But shouldn't you point out that the Unix system administrator
as targeted end user of Unix servers or any server OS in general has a
different level of education than the consumer end user who is
confronted with the same problem? The administrator _does_ know what
kind of attention security deserves. The end consumer doesn't even know
what a "port" is. That's what you missed here.

> But we know that *now*, don't we?  We obviously didn't know that a few years ago, or all the *nix 
> vendors wouldn't have done that years ago, right?

Again: consumer end users != Unix system administrator

And yes, we know by now. Then why is it so hard to demand "secure by
default" from MS for millions of consumer end users?!

> > Don't you think the simple
> > measure of shipping Windows XP Home without such a service enabled would
> > have stopped the spread of Blaster cold? I do.
> 
> Of course it would have, but so would have appropriate OS maintenance.

No. By the time Blaster and its variants were on the way there didn't
exist a patch. Besides, you didn't even have to _do_ something to catch
it. I had a case where I couldn't even reach the MS update site before I
already had it again by sheer presence on the Internet.

Patch maintenance is good but it doesn't replace "secure by default"
settings.

> The only machines we had that got infected by Blaster and friends are those 
> that ignored my many warnings *and* refused to participate in our 
> push-patching program (either through ignorance or belligerence.)  So, 
> while Microsoft may be criticized for shipping RPC on by default, you 
> really can't blame them for the results of the Blaster worm, simply because 
> it was possible to be unaffected by it by updating properly.

You still don't understand the issue. If software requires action on
behalf of the end users to avoid disaster and if this same disaster
could have been avoided by proper software delivery (secure by default)
then of course it is the vendors fault.

> We have thousands of Windows machines running RPC, and none of them are infected 
> because they've all been patched.

Well, then explain to me why Blaster was such a big hit on the net then?
It can't be the end consumers fault. It's never the end consumers fault.
Blaster could have been avoided by simply delivering Windows XP/2000
with a disabled RPC port by default.

> It's high time for us to stop making excuses for stupid behavior simply 
> because Microsoft is an easy target.

There is no stupid behaviour. When a user blindly runs an email
attachment or forgets to patch his machine then this is not the users
fault. The fact that such an uneducated user can actually use the
product this way is to blame on the vendor. Products have to be
fool-prove. It isn't the end consumers who have to be fool-prove.

> *None* of the famous exploits and worms (Code Red, Nimda, Slammer, Blaster, Nachi, et. al.) would have ever 
> happened had people simply updated their machines in a timely and regular manner.

Well, why don't they do it? Ever wondered if this whole problem isn't
being caused at the users end?

> We expect people to change the oil in their cars regularly.

People return their car to the garage when the little red indicator
inside the cockpit blinks. I haven't seen a little red indicator in any
MS product the way it is shipped by default. There isn't one single
indicator inside MS Windows XP Home that tells consumers they are
working with administrative access and that they have an open RPC port
they should close.

> Why don't we expect similar behavior in the computer world?

It isn't the same. That's the problem. Look at liability. A car
manufacturer is always liable up to a certain amount and within a
certain period of time. Software manufacturers aren't liable. They
dismiss liability in numerous different EULAs.

> Would you blame OpenBSD if a user got hacked because he hadn't bothered to 
> patch?

OpenBSD isn't aimed at the consumer, it is aimed at the system
administrator. The point why I brought up OpenBSD is that even if the
Apache ports package shipped with OpenBSD causes the risk of system
compromise due to a bug then this isn't tragic because only those users
actually running Apache have to care. Other users don't bother since
OpenBSD comes with minimum enabled services. That's what makes it
different from MS. You fail to recognise that.

> I'm not arguing that Microsoft has done the right thing or even that their 
> OS is secure.  (It isn't, and I refuse to use it as a server unless forced 
> to.  I prefer to use FreeBSD whenever possible.)

This is a totally different matter, but I agree with you ;-)

> I'm arguing that you can't blame Microsoft for malicious code that takes advantage of weaknesses 
> for which they have already issued patches, sometimes 12 months in advance 
> of an outbreak.

But sometimes weeks after first exploits have shown up. There are still
numerous unfixed flaws in IE6 and beneath that can be exploited.

And I have to say it again: if an exploit could have been avoided by
simply shipping the OS with minimum enabled services, reducing the need
for immediate end user involvement, then the vendor has utterly failed.
There is no compromise to this.

>   *That* is a problem directly attributable to users.

No. Users are never wrong. Get that into your heads techies. THEY are
the customers, WE have to supply products THEY can use WITHOUT making
these mistakes. If THEY fail to use OUR product the way WE intended to
then it is OUR fault not THEIRS. It's as simple as that.

I'm not talking about administered company LANs here, I'm talking about
the consumer end user who wants things as smooth as possible. Products
have to be fool-prove not the end user. This is an ideal and may never
be reached. But nevertheless this has to be the essential design
philosophy of ANY product.

> What you're trying to argue is that, if OS vendors would simply do the 
> right thing from the start, users would be protected despite their lack of 
> patching, and I am saying that is preposterous.

It's not 100% that way but I stand by my statement that more than 75% of
viral and worm outbreaks could have been avoided if attachments weren't
executable out of email applications, if email applications didn't use
the IE's buggy rendering engine together with scripts enabled and
AvtiveX and if ports of unnecessary services were closed by default.

The latest thing, Win32/Bagle-A wouldn't be possible if email client
vendors and MS wouldn't allow the execution of attachments by default.

There is no user "too stupid" to do the worst possible thing with his PC
yet not realising it. This must be in developers mind when designing
software, nothing else. MS has failed utterly at this.

> *No* OS is so secure that you can simply leave it on the Internet, never patch it, and still be 
> secure.

I am not propagating against patching. I am propagating against "opting
into security".

cheers,
Tobias

Powered by blists - more mailing lists