Re: [Ietf-http-auth] Request for review and consensus -- draft-hartman-webauth-phishing

[Tom Yu <[e-mail]>]

SM <[e-mail]> writes:

> At 09:01 09-09-2008, Tom Yu wrote:
>>     "We assume that users wish to protect themselves, but are willing
>>     to expend only limited effort to combat phishing; they will avoid
>>     an interface if they find it too complicated.  This can result in
>>     the user preferring a simpler insecure interface to a more complex
>>     but more secure one.  Alternatively, a user more fully informed of
>>     the risks may abandon any effort to access a service if the choice
>>     is between using a complex, secure interface and using a simple
>>     but known-to-be-insecure interface."
>
> That's a good summary of the problem from a user angle.  It's a user
> interface design consideration.

I have repeatedly heard, from various people in the IETF, statements
to the effect of "it's a user interface design consideration, so we
don't have to worry about it".  Your statement above might not be an
expression of this sentiment, but I think the point is worth
addressing.

The IETF should certainly care about user interface design, but in a
very particular way: the protocols that the IETF designs place
constraints on user interface designers, and these constraints can
drive user interface design in ways that dramatically affect the
security and quality of the end user experience.

By their very nature, user interfaces involve the user interacting
with or interfacing with something.  I will grant that presentation
details of a user interface, such as graphical layout, are not the
business of the IETF.  Protocol design affects what information is
available to a user through a user interface.  It also affects the
relationships among those pieces of information in terms of time,
space, and dependencies.

For these reasons, I think that the IETF should pay careful attention
to how protocol design decisions affect user interface design.