[Cryptography] Encryption opinion

[ianG]

On 25/08/2014 12:32 pm, Stephan Neuhaus wrote:
> On 2014-08-25, 12:50, ianG wrote:
>> Phishing is an MITM.
> 
> Except that the M isn't ITM in the case of phishing.  Phishing is not so
> much a Man In The Middle, it's more a Man On The Sidelines That Looks
> Very Much Like Bob, or MOTSTLVMLB, but good luck pronouncing that.


I don't see the distinction.  The phisher redirects Alice's browser to
him.  He then goes to the site and extracts information to perpetuate
the deception.  What's not middle here?

I certainly understand that there is a frustration in taking a term that
is 'captured' by the SSL industry and using it against them.  That is
both the frustration and the point -- wake up and see what it is that is
happening.


>> So it turns out that HTTPS protects against a class of MITMs, not all
>> MITMs.  As the easiest MITM is outside HTTPS, what is the point of
>> protecting MITMs at all?
> 
> Not sure if sarcastic or serious, but HTTPS isn't useless just because
> it doesn't protect against phishing.


No, it isn't useless.  But it isn't doing what it was sold to do.  Until
it is understood what it can do and can't do, and the messages are
aligned, the industry will remain in mobius loops of finger pointing.


>> What it distracts from is:  protecting the user.
> 
> We're very much on the same page here.


So what we actually need here is an honest message.  CAs and PKIXs and
so forth need to go to browser vendors and say:  "sorry, we got it
wrong.  We can't stop all the MITMs.  You are responsible for
outside-HTTPS MITMing and you have to fix it."

But getting any industry player to admit weakness is like slicing off
limbs.  Players protect themselves, not the users.


>> And, do you think that if they browsers had said "we must eradicate
>> phishing!" they would have succeeded?  Of course the would.
> 
> I'm not sure.  It's very hard (at least NOW it's very hard) to come up
> with a way to tell users that a site is probably a phishing site without
> confusing them even more than they already are.


People outside tried and succeeded, there are several plugins which did
the job.  Some of that work is now being appreciated, c.f. 'pinning'.
The original model was quite clear but did not survive to Netscape v1.0
because of what were then known as the real estate wars.  In essence
everyone believes what they have now works, so changing things is as you
say very hard.


>> I don't understand how they did it, but they managed to forget the user.
>> Entirely.  Industry-wide cognitive dissonance allowed everyone from
>> IETF to vendor to CA to audit to proceed happily without addressing
>> phishing [0].  How do you explain that?
> 
> Again, I'm on the same page as you, so I'm not going to "explain that"
> :-) In my original post I merely pointed out that crypto won't stop Eve
> from dressing up as Bob while still showing credentials that say
> correctly that she's Eve.


Indeed crypto cannot;  but a holistic approach can.  And maybe this is
the issue.  We aren't up to fielding a holistic approach to security.



iang