Re: SSH revised

[Marcus Brinkmann]

At Tue, 28 Mar 2006 22:05:50 +0200,
Bas Wijnen <address@hidden> wrote:
> 
> On Tue, Mar 28, 2006 at 08:25:30PM +0200, Marcus Brinkmann wrote:
> > Anyway, I think user-based authentication would be severly crippled by
> > a confinement requirement.  One feature that would be useful is if
> > they can access arbitrary devices that are not supported by the
> > system, and for that alone you can forget about confinement here.
> 
> Hmm, that makes sense.  So you suggest the user needs to select an ID first,
> after which an explicit confirmation is needed that it's the right one (for
> example showing the fingerprint picture)?

Yeah, well.  We need a user interface expert here.  If we are serious
about this, I would try to contact somebody like Ka-Ping Yee
(http://www.sims.berkeley.edu/~ping/sid/).

Two confirmations make sense, but I was also considering an
"optimization": The "fingerprint picture" could be displayed as soon
as the user name is entered.  Then both could be confirmed at the same
time.

I have a cuter name for the "fingerprint picture" already: The user's
mascot.  The pictures must be so cute that the user will log in just
to see the mascot :)

In principle, the mascot could be chosen by the user, as long as every
user has a distinguishable one.  That is a challenge for the
administrator, though, of course.  So, it would be an interesting
research field to automatically determine "distinguishability", or to
automatically generate many distinguishable mascots algorithmically.
Of course, there are many other possible policies to choose the
mascot, for example by using photos taken from the user (like on photo
id cards), or words from a lexicon, typeset in different fonts, colors
and sizes, ...

Of course, the whole issue is highly user-interface specific: For
accessibility reasons, you also need other solutions.  Audio files?
Quotes from the literature?  User data like residence address?  Lots
of possibilities here.

Thanks,
Marcus