Re: Secure Chrome and Secure MetaData

On 2006-06-21 09:11:30 +1000, James A. Donald wrote:

> I think you are guilty of premature optimization.  One
> question surely is: what problem does one want to solve
> using Secure Chrome and Secure MetaData?

The (maybe relatively modest) proposal here is to end up in a
state in which at least vigilant users are able to reliably and
easily tell in what kind of security context they operate. This
brings up three three questions:

- What information should be presented, as a baseline?  There's
  context to individual transactions (TLS in particular) that
  can help, and there's context in terms of user agents'
  historic memory that might help.

- How do you present that information so people get it?

- How do you keep attackers from tampering with this display?
  How do you keep them from spoofing it?

With respect to usability, this approach to scoping quite
consciously pushes one of the really hard problems to the
sidelines for the moment: How do you get users out of routine?
How do you wake them up, so they become vigilant in the first
place?

Likewise, e-mail authentication is out of scope.

If you think you have a more productive scope and direction of
work to offer, let everybody hear it.  Or comment on the
charter drafts to which I pointed last night, and make concrete
proposals.

But please don't repeat over and over (together with Chris
Drake) that "the problem can't be broken into pieces." This is
not helpful at all.

-- 
Thomas Roessler, W3C   <tlr@w3.org>

Received on Wednesday, 21 June 2006 07:56:34 UTC