[Sigia-l] Rollover Navigation Submenus: Usable or not?

[Benjamin Kahn]

I've always been curious about flyout navigation in web pages.  They
resemble menus in desktop applications but with behavior that seems
designed to make them much harder to use.  I admit that I work with
desktop applications far more than I do with web pages and that I have
never created a web page with flyout navigation.

Since so much research has gone into designing the menu interaction in
desktop applications and since flyout navigation seems so similar to
menus already, why isn't this knowledge applied to flyout navigation on
web pages?

Menus usually require a signal from the user that they are about to be
used.  For example, a mouse click, F10, or an Alt- key accelerator.
Simply hovering your mouse over a menu don't not usually do much.
(Although many menu implementations have have a prelight which lets the
user know that the widget is active.)

Flyout navigation seems to always activate on hover.  This always struck
me as distracting and difficult to use.  I often open menus without
meaning to, and must be very careful where I put my mouse cursor on most
web pages.  

Menus usually require an action to signify "I'm done."  Either an item
is selected, the ESC key is pressed, or the user clicks somewhere off
the menu.

Flyout navigation has significant problems (it seems to me) with
pop-down.  Usually, just moving the mouse off the menu causes the menu
to disappear immediately.  Since the user may have activated the menu
accidently, this might be wise.  Some better implementations appear to
have a timeout that delays closing the menu until the user has moved the
mouse away for some period of time.

Sub menus in desktop applications use two different way to know when a
user is trying to access them.  Macintosh and Linux machines create a
predictive triangle between the current position of the mouse cursor and
the size of the displayed sub-menu.  As long as the user's mouse cursor
stays within this triangle, the sub menu doesn't disappear even if the
mouse cursor is over another menu item.  (This isn't completely true...
If the user pauses over another menu item for long enough, the submenu
will still disappear.)  Windows uses a simpler algorithm -- submenus
wait for some period of time before displaying or disappearing.  

Browsing around, I've found one flyout navigation menu that uses the
Windows style of submenu delays:  http://www.opencube.com/  Why don't
more web pages use this simple (and common) idea?

Do these changes simply not fit in with the rest of the web page
interaction?  Is it simply too hard to implement at the moment?  Is it
just the way everyone else is doing it?

On Fri, 2004-12-17 at 20:26 +1100, Donna Maurer wrote:
> I change my mind frequently on this one...
> 
> I have done user research on many sites that have horizonal flyout navigation and 
> found that it is technically 'unusable'. People have tremendous difficulty hitting the 
> slippery flyout menu 
> (like this: http://www.abs.gov.au/ausstats/abs%40.nsf/ausstatshome?OpenView)
> to the point where I have seen people try to hit a target 6-7 times before they get it 
> (and this is people with 'normal' motor skills).
> 
> I have also noticed that in these situations, people don't pay *any* attention to the top 
> level - they skim through the second level of the navigation looking for something that 
> looks relevant. The top level could be named anything as it is the second level that is 
> used (hey, I wrote about this once too: 
> http://www.maadmob.net/donna/blog/archives/000565.html)
> 
> The real advantage is that the second level can provide the additional explanation of 
> what a top level category includes (if people use the top level). But if this is the 
> solution, perhaps the problem is that the top level is wrong. Perhaps the flyout 
> shouldn't lead to discrete subsections but perform this role in another way - by 
> providing an explanation rather than a set of links.
> 
> I don't know. Sometimes I think that flyout nav (like on this site: http://ato.gov.au/) is 
> just lazy IA, sometimes I think it could be useful.

> On 14 Dec 2004 at 12:14, Jon Nakasone wrote:
> 
> > Forgive me if this topic has been previously discussed.
> > 
> > I'm in the process of rapid prototyping a horizontal navigation using
> > buttons with DHTML rollover submenus.  In the midst of this I was told
> > that this approach is shown to be consistently unusable by many users.
> > No elaboration was provided, other than this had been a recurring
> > comment in numerous usability seminars at Marketing conferences (yes,
> > Marketing conferences).
> > 
> > Now I know what you're going to say and of course I plan to properly
> > test variations of the navigation with our users, but I wanted to
> > collect everyone's opinions and experience with this type of
> > navigation as I've never heard a blanket statement placed on a UI
> > feature made as insistently as this was.