changing the default for --keyid-format

[Robert J. Hansen]

On 5/29/12 1:18 PM, Werner Koch wrote:
> Frontends should handle this problem.

The problem is that most people developing front ends are making them
pretty darn user-hostile.

A few years ago while taking some HCI courses, I did a usability study
on the most common certificate interface -- the tabular widget.  It
turned out to be just beyond-Godawful.

<rant follows>

Tabular data is the Right Thing To Do in two major use cases.

The first is when you have a noninteractive display of identical
field(s) for multiple pieces of data.  Consider a printed almanac: if it
wants to convey a list of countries and populations, the best way to do
it is with a table.  Different records (countries), identical fields
(population), and since the paper is noninteractive, the table is a win.

Now consider if instead of an almanac you have Wolfram Alpha.  Typing
"population of Switzerland" immediately yields *just* the data you want,
and you don't get confused by your eye accidentally jumping a row and
reading the population of Sweden instead.  A table widget is more prone
to misreadings.

The second Big Win for tables is when data must be contextualized by
other data.  Consider a spreadsheet showing profits and losses for
different divisions of a business: if all you know is that a given
division made $X, you don't know if that's your most profitable
division, your least profitable division, or what-have-you.  The other
data is necessary to put the data you're interested in into a larger
context.

Now consider the tabular widget as used in PGPkeys, GPA, the Enigmail
key manager, etcetera.  The certificates don't need to be
contextualized: all the data necessary to evaluate a certificate is
present in the same record as the certificate.  And since it's a
graphical application the interface can be interactive, which means the
other major use-case isn't applicable here.

Enigmail tries to have its cake and eat it too by prominently featuring
a large search box at the top of the window.  But this isn't a very good
solution.  In terms of screen real estate, about five-sixths of the
screen is taken up by the tabular widget.  The search box takes up a
relatively small portion.  The human eye tends to view large things as
more important than small things -- so the center of attention is
naturally drawn to the tabular widget, not the search box.  Further, the
human eye tends to view complex things as more important than simple
things -- so the eye is drawn to the tabular widget again, not the
search box.  I'm grateful Enigmail has a search box in the certificate
manager, but I doubt if new users even notice it.

According to Google's HCI guys [2], 90% of the U.S. internet-using
population doesn't know how to use Control-F to find a word in a
document or a page.  That's the level of skill most people have with
user interfaces -- awful.  And if you count up the number of widgets on
the screen in your average certificate manager, you'll find that there's
more visual complexity there than in Microsoft Word.

</rant over>

Anyway.  If people are interested in what I found out about effective
user-interface design with respect to certificate managers, say the
word.  Otherwise I'll crawl back under my rock and leave the subject
alone for another couple of years.  :)