[cap-talk] Re: First point of consensus

[Ian G]

Ka-Ping Yee wrote:

>On Wed, 9 Feb 2005, Ian G wrote:
>  
>
>>Web sites running phishing attacks on themselves ... by which
>>I mean there are a number of cases where hackers have slipped
>>into a website, and injected HTML that will attack the users.
>>    
>>
>
>I assume we can consider such cases out of scope for this discussion.
>If your site has been hacked, the problem has to do with securing your
>website, not with spoofing.
>  
>

In security work, we would generally not
assume any threats out of scope of
discussion.  Rather what we would do is
catalogue them and put them in the "Hard"
basket, which later on becomes the "Do
not cover" basket as the model gets closer
to completion.

I know that sounds bitchy, but repeating
the errors of the past isn't going to be
worth anyone's time and effort.

One way to tell if a model is bogus is if
it doesn't have a "Does not cover" basket
or the designers say that it is completely
secure.

In this case, you know that the authors
assumed away all the things they couldn't
cover, and hid them, and hoped nobody
would notice.  This is partially why SSL is
how it is, certain real threats and imaginary
threats were assumed out of their real
proportions, then the assumptions were
buried.

And here we are.

iang

-- 
News and views on what matters in finance+crypto:
        http://financialcryptography.com/