spatial nautilus concerns

[David Feldman <mailing-lists interfacethis com>]

I realize I'm coming to this discussion a bit late in the game, so I  
apologize if I'm reopening dead arguments. It's been several years  
since I last played around with Linux and I'm pretty excited to see the  
progress the Linux GUI has made...at this point I'd argue it's more  
usable than Windows it most respects. (I'm a UI designer so I think of  
most everything in terms of "usable" :-))
I'm concerned about the discussion to date of the "spatial" /  
"object-oriented" metaphor, and have been ever since I read John  
Siracusa's article. It resonated with a lot of people, which worried me  
because many of his arguments, and the principles on which he bases  
them, are not supported by data and seem somewhat suspect. A few of his  
claims contradict evidence from the cognitive sciences altogether. I'm  
glad to see, looking through the GNOME discussions on the topic, that  
the basis for the spatial Nautilus decision goes beyond just taking the  
Ars Technica article at its word and is based on a direct debate of the  
issues involved. But I'm saddened that, given how fundamental this  
particular interaction is to the user's interaction with his computer,  
it appears that no empirical evidence has been gathered to verify this  
decision.
One thing overlooked in much of the discussion of this issue is the  
existence (and advisability) of hybrid solutions. The Mac OS X Finder,  
despite Siracusa's claims, adheres to nearly all tenets of his spatial  
metaphor. Icons and windows stay where you put them, and there is a  
one-to-one mapping of folders to windows except as overridden by the  
Open in Same Window preference, the Column View, and explicit user  
changes such as Keep Arranged by Name. Because the Open in Same Window  
preference necessarily implies some of the navigation metaphor, Apple  
also provides small forward/back buttons for users who want to take  
advantage of that. To me this all seems like a good compromise, but my  
opinion doesn't constitute empirical evidence any more than anyone  
else's does. The point is simply that it's simply not a black-and-white  
issue, and numerous in-between options are possible that mix and match  
based on the value of certain trade-offs.
Trade-offs are, in fact, the main issue. Most of the spatial metaphor  
seems to make sense in a vacuum, but every change has multiple  
consequences for usability, and is made at the expense of some other  
option. If strict one-to-one mappings of folders to windows is easier  
to learn (something I haven't actually seen hard evidence of yet), is  
it worth the drop in efficiency that might result from having so many  
windows onscreen? If the current GNOME scheme (with browser and  
non-browser versions of the windows) is used, is the added complexity  
of its two-interface system worth the preservation of one-to-one  
mapping in the default configuration? (I think the one-to-one mapping  
thing is really the most important issue - it's certainly worth getting  
evidence that keeping icons where one puts them is the best default,  
but there's less debate about it and I think the issue is less  
complex.)
A couple additional things:
- Without far more drastic changes to the GUI experience, I don't think  
we're ever going to get to a pure spatial metaphor. For example, the  
File Open dialog currently still uses the open-in-same-window scheme,  
and presents the user with her folders in a window other than the one  
shown by Nautilus. Sure, it's a different app, but that's not  
necessarily important to (or understood by) the user.
- One common argument in favor of the spatial metaphor is that most  
people don't find file paths natural. Yet people often put things in  
folders, or categorize them according to nested hierarchies. (Siracusa  
claims that most people aren't good at this, but provides no support.)  
Maybe that's out of necessity, but maybe it's a natural way for people  
to classify things. If the latter, file paths may be natural for  
people, and (for example) it may just be that many users aren't always  
comfortable with the textual representations of those paths, and/or  
with hierarchies that they themselves have not constructed.
I certainly have my guesses at the answers to these questions. As a UI  
designer, I often base designs on such guesses. But I also always  
recommend the use of empirical methods such as usability testing to  
verify my choices. It's important when those choices are simple layout  
decisions, bounded by the standards of the platform. It's much _more_  
important when one is defining the platform itself and dictating some  
of its fundamental default behaviors.
If empirical testing is underway, I apologize for my lengthy  
message...I haven't heard anything and feel pretty strongly about this.  
I started feeling strongly when I read the Ars Technica article, mostly  
because of the number of broad claims it makes about usability without  
any real support. But it goes beyond that. Linux, in large part due to  
the efforts of the GNOME community, is becoming a viable alternative to  
Windows for more and more users. I'm really excited about that, and the  
more it gets right the more excited I am. As more and more nontechnical  
users make the switch, the cost of making fundamental changes to the  
user experience rises, so the time to make such changes is now. GNOME  
has implemented all the functionality required by pretty much any  
permutation or hybrid of the spatial and navigation metaphors, so the  
question becomes one of choosing the right default settings rather than  
a lot of development time. Empirical methods such as usability testing  
work. So why not apply them and make sure we've got it right?
In the meantime, a modest proposal that doesn't change the default but  
provides flexibility to users and developers: Decouple the "open in  
same window" option from the "always use browser window" option. Then  
users (like me) who want to stick with the more minimal window style  
but would like folders to open in the same window can set things up  
that way. (Similarly, users who mostly want the spatial metaphor but  
also want the browser functionality at their fingertips can leave "open  
in same window" off but enable the browser.)
Thanks for listening :-).

Dave

------------------------------------------------------------------------ 
--
David A. Feldman
User Interface Designer
http://InterfaceThis.com