← Back to team overview

unity-design team mailing list archive

Re: Global menu in Oneiric Ocelot (11.10)

 

On Thu, May 26, 2011 at 12:47 PM, Adrian Maier <syraxes@xxxxxxxxx> wrote:
> You are right about the lack of verdict.  The fact that some people dislike
> it   doesn't necessarily mean that the default behavior must be
> immediately changed.

Two things we need to discern: the concept of a global menu and the
Unity global menu.
The Unity global menu (visible only on hover) is objectively bad
design. The main reason to switch to the global menu was speed of
access (and not saving screen estate as some want to make you believe:
maximized windows can do that Unity magic independently of the global
menu or even a top panel). However with the hover design this
advantage was lost. Especially for new users (the focus group of
Unity) it is empirically (see the mail by GonzO above) worse than the
alternative.

The second aspect is the concept of a global menu in general. It's not
so easy to dismiss it completely but it also doesn't come without
drawbacks. But I think I have sufficiently demonstrated by now that it
only has a single advantage (speed as per Fitts's Law) which isn't
undiluted (because of multi-tasking and other issues) but on the other
hand introduces _several_ new issues that didn't exist before. This
isn't my personal opinion (as I said, I'm not dogfooding Unity and if
I was I could patch it to do whatever I want), it's about objective
and logical conclusions. The only thing missing is empirical hard
data.

My suggestions:
First get rid of the hover behavior, by default. Then do some user
testing, by some I mean extended. Unity with and without
indicator-appmenu and as a bonus: Unity without a panel at the top and
browser tabs on top.
The user testing should include desktop and netbook resolution, how
fast people operate the system, their feedback (did they feel lost,
was it enjoyable to use?) and report on any problems that did arise
during the testing.

My question: How can we make this happen?
How can we measure behavior in the millisecond range? How should the
test be setup in detail, has anybody done this before?
I've heard of two Unity usability tests, can we get them on board?
Anything I'm missing (I bet there is)?

> But there is a solution:    make the thing configurable .
>
That's not a solution, that's a band aid. Once we have found the best
settings which are going to be used as the default we can always add
some options of course. But we are far from that stage.



References