<<< Date Index >>>     <<< Thread Index >>>

RE: International Domain Name [IDN] support in modern browsers allows attackers to spoof domain name URLs + SSL certs.



David Schwartz wrote:

> > My proposition is that the argument that they (and their associated webs
> > of trust) are inherently trustworthy because of external pressures is a
> > flawed assumption because they do not have the proposed level of
> > pressure applied to them since most of the people affected by their web
> > of trust don't understand it.
> 
>       They don't have to. I don't understand how my supermarket gets their 
> meat,
> but I trust them to use safe sources because I know that if they didn't
> those who do understand would tell me, and then I'd figure out a way to
> avoid it.

That is not why you trust your supermarket to source good/safe meat at 
all.

You trust your supermarket to source good/safe meat because you live 
somewhere that has strongly enforced regulations, with very stiff 
financial penalties, covering the slaughtering of animals, preparation 
of their carcasses into meat products, and every step of the storage, 
shipping, handling, display and sale of such products.

And, in fact, very similar reasons are why you trust so many other 
conveniences that comprise "the modern Western way of life".

Further, these systems are so ingrained and work so well, most people 
(such as yourself?) have forgotten that the checks and balances even 
exist, taking for granted "safe meat from the supermarket" and so on.

The previous poster, to whom you responded is essentially correct.  The 
difference between CAs and the webs of trust surrounding them and the 
whole CA/certification process do not have the checks and balances 
governing them that they are assumed to have.  This is equally true of 
most other trust issues with computers, such as the most basic ones as 
the assumption on the part of consumers that the OS and standard 
applications for the typical tasks to which computers will be put are 
designed to competently and safely perform those tasks while protecting 
the users from what should, to the technically competent and informed 
folk it is assumed design, write and test such software, be  "obvious 
dangers".

>       No CA wants to find out what market forces will appear as soon as they
> prove to be untrustworthy. There are already many vehicles for immediately
> deploying blacklists. For example, Symantec could release an update for any
> of their security products that removed a root CA. It wouldn't take more
> than a small percent of web users to have a problem with a CA before people
> wouldn't want their certificates to be signed by that CA.
> 
>       The CA market is competitive.

So, why is VeriSign still in the CA business?  Or should releasing two 
code-signing certificates in Microsoft's name to non-MS related folk 
not be considered untrustworthy enough to utterly destroy any rational 
person's or organization's trust in a CA?


Regards,

Nick FitzGerald