LWN.net Logo

A pair of desktop initiatives

Whether or not they agree that Linux is "ready for the desktop" or not, most observers will allow that there remains plenty of room for improvement. And while some of those improvements will take the form of slick new applications, there is also quite a bit of less glorious work to do. So it is encouraging to see a couple of new efforts aimed at improving the quality of the desktop we already have.

Novell has sent out a press release announcing the launch of the Better Desktop initiative. This effort, part of the OpenSUSE project, intends to provide information to developers which will help them to make the Linux desktop a better experience.

User-oriented proprietary software companies have many techniques for improving the usability of their products. One of those is to film users trying to fight their way through an application, then lock the developers in a room and force them to watch the users struggle. No popcorn provided. Developers know their software, so they will not wander into the traps and dead-ends which confuse the rest of the world. Watching others run afoul of usability problems shines a light on those problems which cannot be denied. Once the problems are seen and understood, the developers can start to think about solving them.

The Better Desktop project cannot lock developers in a room, and it cannot deprive them of the refreshments of their choice. What it can do, however, is provide the films. As a start, the project has posted video streams of several users as they attempt to accomplish a set of objectives. Also posted is a small set of reports drawing conclusions from the videos. These conclusions are relatively simple (users want to see the username and password fields together on the login screen, for example), but they do demonstrate the sort of issues that developers tend not to see on their own.

The research results posted are just a beginning; one assumes that the project will run more experiments over time. Your editor suggests "figure out how to make betterdesktop.org display reports in firefox without popping up new windows" as a nice place to start. As this body of data grows, implementing usability improvements indicated by the results should be a relatively straightforward task. In usability, as in many other areas, the real challenge is figuring out what problems to solve, rather than implementing the solutions.

[icon]
[icon]
[icon]
[icon]
[icon]

The Tango Project has taken on a different goal: get rid of visual inconsistencies between desktop applications, regardless of their source. In particular, Tango has targeted icons as an area needing improvement. So, the project has posted a set of style guidelines on how icons should be created and a specification on how they should be named. If applications adopt both, the result should be applications that look the same everywhere.

The Tango icon gallery gives a good demonstration of the guidelines in action. These guidelines call for bright colors and well-defined perspectives on objects. Not everybody will like the relatively cartoonish approach taken by Tango, but use of these icons will undoubtedly create a lively desktop.

Tango may or may not succeed in the real world. It is important, however, as a cross-desktop effort to improve the overall user experience. If the Linux desktop is to continue to get better, a great deal of usability and consistency work will have to get done. The fact that projects are coming together to make a start on that work can only be a good thing.


(Log in to post comments)

Icons need action ?

Posted Oct 13, 2005 11:01 UTC (Thu) by kenmoffat (subscriber, #4807) [Link]

Surely, the only thing icons need is to be removed from the desktop ;) I'll admit to using icons on the taskbar when I run a kde desktop (which isn't very often, and only to test that it built ok), but what's so great about icons ? I don't know what sort of screens the "average" user is using, but on 1600x1200 I really don't have space for icons on my desktop (one browser window, a couple of xterms, xmms, and almost all the background is covered).

I'm using icewm, I suppose that the things at the left of the taskbar count as icons (menu, show desktop, window list, xterm), but of those the menu and the xterm are the only ones I use. People, we need to leave behind this fixation for random tiny pictures scattered across the background!

If you know what you want to run, ctrl-alt-space to type the program name (icewm) or some other keystrokes on other wms to get a requester where you can type the program name. If you don't know what it's called, use the menus. Of course, if the menus are an impenetrabe maze, that's a different problem.

Icons need action ?

Posted Oct 13, 2005 11:58 UTC (Thu) by smitty_one_each (subscriber, #28989) [Link]

>Surely, the only thing icons need is to be removed from the desktop

emacs

Icons need action ?

Posted Oct 13, 2005 14:19 UTC (Thu) by tzafrir (subscriber, #11501) [Link]

IMHO icons on the desktop are bad: you need to stop what you're doing just to view them.

The 4-5 most important programs should be used as taskbar icons. Beyond that, I can't figure anything better than a menu. In a menu icons od help to find an item faster assuming you have already found it before.

Sadly, the command-prompts on icewm (alt-ctrl-space) has no form of completion and/or history.

Icons need action ?

Posted Oct 14, 2005 2:20 UTC (Fri) by coolian (guest, #14818) [Link]

I guess this is an either-or question.

I use icons on the desktop all of the time.

Who wants to watch something slog away? I minimize things...Crazy, I know.

many kinds of icons... instant visual recognition is the key

Posted Oct 13, 2005 15:13 UTC (Thu) by jabby (guest, #2648) [Link]

Icons refer to more than just the launchers on the desktop. As you point out, they are used to identify items in the application menus. They are also used to identify applications which are minimized (iconified). When using the cross-platform Alt-Tab key combination, applications are usually most prominently identified by their associated icon. Then there are icons in pull-down menus and on toolbar buttons inside of applications... It goes on and on.

Icons are everywhere in the UI, and there's a good reason. The human brain is wired to very efficiently process and recognize visual patterns. Colors and shapes in certain arrangements can be analyzed and understood without using the higher cognitive brain functions. So, these visual shortcuts make biological sense.

It follows that having a consistent set of visual patterns will make the act of using a GUI even more efficient and intuitive for the user. The goal is instant visual recognition of the function of a graphical object without having to read any text and then comprehend it within its context.

many kinds of icons... instant visual recognition is the key

Posted Oct 13, 2005 18:45 UTC (Thu) by kleptog (subscriber, #1183) [Link]

Icons are everywhere in the UI, and there's a good reason. The human brain is wired to very efficiently process and recognize visual patterns. Colors and shapes in certain arrangements can be analyzed and understood without using the higher cognitive brain functions. So, these visual shortcuts make biological sense.

For you maybe, I have a terrible time with icons and have removed them also everywhere. I don't know whether it's years of using an old text based IDE but popup a menu and I will find the words "Firefox" far quicker than some non-descript orange circle type logo. Especially if it's sorted.

I understand that "graphics" may be quicker for some (maybe even most) people, but words are "visual patterns" with "colours and shapes" and don't need "higher cognitive brain functions" to understand. Besides, they have the added advantage that they're not all the same size. How's that for quick differentiation.

BTW, I'm one of those people who, if a film jumps the camera to a new scene every two seconds, can no longer follow it. This makes many recent "action films" a complete waste of time. Why can't they keep the camera still, dammit!

many kinds of icons... instant visual recognition is the key

Posted Oct 13, 2005 23:38 UTC (Thu) by phgrenet (guest, #5979) [Link]

I have only one thing to say: kleptog, you are not alone.

many kinds of icons... instant visual recognition is the key

Posted Oct 14, 2005 7:26 UTC (Fri) by hppnq (guest, #14462) [Link]

Besides, [words] have the added advantage that they're not all the same size. How's that for quick differentiation.

Icons are great for people who know what they mean. Like words, really. ;-)

many kinds of icons... instant visual recognition is the key

Posted Oct 14, 2005 7:51 UTC (Fri) by zblaxell (subscriber, #26385) [Link]

"I don't know whether it's years of using an old text based IDE but popup a menu and I will find the words "Firefox" far quicker than some non-descript orange circle type logo."

Someone once asked me what the Firefox icon looked like. I said "It looks just like the one beside the text 'Mozilla Firefox' in the menu named 'Internet.'" Interestingly enough, on my machine at this moment, it's a blue-green circle type logo (I think it's supposed to be a stylized picture of the Earth), and has no trace of any color similar to orange. I think that's wrong...judging from mozilla.org I'd say it's supposed to be blue in the center and orange around the outside. So there is definitely room for improvement in Linux desktop software--even if I don't in principle agree that icons are a good idea, I do think that whenever an icon must be used, it must also be used *consistently*.

Everything worth talking about needs to have names, or people can't talk about them. Consider the many epithets for the anonymous "command" key on Mac keyboards, or the short and elegant synonyms for punctuation characters known to all experienced Unix people (or at least people who have read the appropriate sections of the Jargon file). If no names already exist, names will arise spontaneously whenever people need to talk about something, because most people can't pronounce images.

If something has a name, it would be nice to label it with that name, so that everyone knows what you're talking about. And once the object in question is labelled with a name, there's no further need for the icon, so it can just be dropped. Unfortunately, for some reason, people keep making the text *larger* than the icons, so I end up keeping the icons and turning off the text labels because wasting screen space is the worst of all.

Icons are death when you're trying to do any kind of remote technical support, especially with non-native-English speakers. If I'm speaking to someone 5000 miles away, I'd much rather say "Click Applications...Internet...Mozilla Firefox" than say "Click the logo that looks like a foot...it might be at the bottom or top of the screen...then look for the button that looks like a shiny plastic globe...then pick the globe icon which is slightly deeper blue without the shiny highlights but does not have the gears or ship's-steering-wheel on the left and bottom sides, and whatever you do, do *not* click on the globe with the human hand closing around it or the smaller globe with the colored ribbon circling...no, I'm not making this up! Why do you need me to wait while you write this down? No, I mean why are you writing this down at all when it's all in the computer? What do you mean the computer's in the other room?" The conversation usually goes on longer, but some of the words that come next *are* best represented with icons, at least on "professional" web sites.

Even with a remote connection, a data stream full of icons is just slow for no good reason.

Icons as a user interface feature just don't scale. How many household consumer devices have more than a dozen icons? How many have less than half that number? Most use only a few: one for "power", a pair of icons that means more or less of something (e.g. faster/slower, locked/unlocked), one for "this part might hurt you," one for "opening the case voids the warranty and may cause personal injury and property damage," and occasionally one for "we really mean it, put the screwdriver away before you kill yourself and burn down your house."

Anything that doesn't fit neatly into one of those categories gets a text label ("card reader" or "channel" or "remote speaker" or "Video 3") or no label at all. I have one or two devices where icons are used that don't fit these categories, but the icons are so meaningless that it's necessary to experiment with all of the buttons to see what they actually do, as if they had no labels or icons at all.

My car--by far the non-PC consumer device with the most complex interface--has a dozen icons on the driver's side of the dashboard (and, ironically, an LCD text display), and another dozen sprinked throughout the passenger space. There is a cluster of icons under the hood which means approximately "you have opened the hood, you ingenious fool you [icon], now go read the manual [icon] before you burn [icon], blind [icon], cut [icon], dissolve [icon], or electrocute [icon] yourself, try to avoid hitting your head with the hood [icon] as you close it, and make sure you close it properly [icon]." Road signs add another hundred or two icons. Most of these are variations on a theme, e.g. "curve left" is the mirror image of "curve right" and "air conditioning" is different in color but exactly the same shape as "maximum air conditioning," so the number of visually distinct icons is only a few dozen.

My laptop currently has 1114 entries in /usr/lib/menu. That's almost 100 vistually distinctive, recognizable icons on my desktop for each distinct icon a driver will see while driving a car. If they all had consistent, visually distinctive, and mnemonic icons, users would still require a lot of help just to find anything in that pictographic soup. Help like...text labels! Yes! Why didn't we think of this before, just put names on everything! ;-)

"BTW, I'm one of those people who, if a film jumps the camera to a new scene every two seconds, can no longer follow it. This makes many recent "action films" a complete waste of time. Why can't they keep the camera still, dammit!"

Because most modern action films would just look like a bunch of people trying really hard to avoid getting hurt if they didn't confuse the viewer with lots of editing. Hopefully there is continuity to be found in the soundtrack, otherwise I just wait it out and hope the film isn't directed by Jean-Christophe Comar.

many kinds of icons... instant visual recognition is the key

Posted Oct 20, 2005 7:49 UTC (Thu) by mcm (guest, #31917) [Link]

> Someone once asked me what the Firefox icon looked like. I said "It looks
> just like the one beside the text 'Mozilla Firefox' in the menu named
> 'Internet.'" Interestingly enough, on my machine at this moment, it's a
> blue-green circle type logo (I think it's supposed to be a stylized picture
> of the Earth), and has no trace of any color similar to orange.

FWIW, mozilla.org only allows official builds to carry the official firefox icon (stupid, i know ...)

Words vs. Icons

Posted Oct 20, 2005 16:52 UTC (Thu) by kellymjones (guest, #33224) [Link]

Amen! I can't believe that we've developed advanced writing systems over thousands of years only to throw it all out and use pretty little icons. I like a visually pleasing desktop as much as anyone, but as I tell my young son, "Use your words!"

Icons need action ?

Posted Oct 20, 2005 18:44 UTC (Thu) by renox (subscriber, #23785) [Link]

The thing is: generaly an user tend to use only one application at a time, so it doesn't matter if the screen is full of window: a click to iconify all the windows and a click on the icon and voila.

Much easier than navigating into a maze of menus, and no need to reach for the keyboard.

Besides, icons are just an additional option, that you're free to ignore, but that many people like..

Icons need action ?

Posted Oct 22, 2005 8:46 UTC (Sat) by jospoortvliet (subscriber, #33164) [Link]

have a look at kde-artists.org, they are working on a replacement for the
menu's for starting applications! some very cool ideas there...

A pair of desktop initiatives

Posted Oct 13, 2005 15:33 UTC (Thu) by bronson (subscriber, #4806) [Link]

From the Better Desktop recommendations:
2. Split the Evolution menu entry into multiple entries for the various components, and make them more prominent in the menus.
Why not just split Evolution itself? How many times do we need to learn this lesson??! Witness FireFox et al vs. Mozilla. Look at OpenOffice staggering under its own weight. It's obvious! Gaahhh!! *eats hat*

A pair of desktop initiatives

Posted Oct 22, 2005 8:45 UTC (Sat) by jospoortvliet (subscriber, #33164) [Link]

you mean like kontact, which just consists of seperate applications?

A pair of desktop initiatives

Posted Oct 13, 2005 15:37 UTC (Thu) by maney (subscriber, #12630) [Link]

Your editor suggests "figure out how to make betterdesktop.org display reports in firefox without popping up new windows" as a nice place to start.

Either they've fixed it quite promptly or one solution lies in setting Firefox not to popup new windows. When I dug down to what I assume are the reports in question and clicked on the first one (login), it opened a new tab, not a new window. Which seems appropriate to me: if I'd been browsing without any expectation the link was anything but vanilla, I think that would be the point where I'd have hit the middle button to intentionally get the report page in a new tab. In fact after it had opened in the new tab I wasn't absolutely certain that I hadn't done just that, and felt the need to double check it by reopening the report. :-)

Usability Films

Posted Oct 18, 2005 23:11 UTC (Tue) by crouchet (guest, #1084) [Link]

Usability labs work. Developers tend to have blind spots about their own software and end users don't know what is possible so they typically have no idea what to suggest. But watching those end users work through your application is a great way to spot the flaws and areas where the users just don't think like the developers.

As a member of a team developing a large, complex application I can say we have found this method very valuable. It is important to select your end user sample carefully since selecting those who know too much or who are too incompetent is not useful and can even be misleading. The ones you select should accurately represent your target audience.

Watching those end users try to find their way though your app can be a bit painful and humbling so you need to leave your ego on the shelf when you watch.

We usually have our team members sitting together in a location remote from where the user is working through a set of scenarios. We watch using a projector. Since we are far out of earshot we can discuss what we are seeing, make suggestions, brainstorm, etc as the user works. We make lists of problems/improvements/suggestions to be worked later. We have a controller who will help the user along if they get really stuck or misunderstand the scenario but should otherwise be silent and let the user struggle through it as much as is feasible. Our controller will also interview the users at the end and get their impressions, suggestions, likes and dislikes.

Of course, this does little or nothing for under-the-hood issues but that is not the point of this sort of testing.

JC

Copyright © 2005, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds