Linked by Kostis Kapelonis on Tue 14th Mar 2006 18:59 UTC
Graphics, User Interfaces The desktop metaphor has served our computing needs well for the last decade. It has started however, to show its age over the last years. For office users it is still adequate but for everyone else it is often awkward and slow. Since a computer is no longer confined in the office, but in some cases serves also as the entertainment hub in our living rooms, new User Interfaces are required. In some areas the foundations are already in place while in others users are silently suffering every day, having to cope with inefficient and unproductive UIs.
E-mail Print r 6   · Read More · 47 Comment(s)
Order by: Score:
v Last Decade?
by JimmyBob on Tue 14th Mar 2006 19:26 UTC
What about CDE?
by Robert Escue on Tue 14th Mar 2006 19:38 UTC
Robert Escue
Member since:

While Kostis mentions Gnome, KDE and Enlightenment, what about CDE? Despite it's lack of popularity it ships with all of the mainstream UNIX variants (Solaris, AIX, HP-UX) and was marketed as a "desktop".

Reply Score: 1

RE: What about CDE?
by grayrest on Tue 14th Mar 2006 23:38 UTC in reply to "What about CDE?"
grayrest Member since:

The problem with CDE is that it pretty much sucks.

I used it quite a bit as an undergrad on Solaris 8 and never liked it.

The CDE implementation of drawers for various apps and actions never worked for me. I consistently had to check all the drawers every time I needed an app I used on an infrequent basis.

There is not a digital clock in the interface by default (the analog is impossbile to read) and my solution of running a xclock in the lower right corner of all desktops never struck me as elegant.

I never did figure out how to set the browser button to launch mozilla.

The interface is ugly.

The docs are unhelpful, describing the obvious (this is a square!) rather than explaining how to accomplish anything (how to change what is launched).

When you minimize windows, they go UNDER your other windows so that you have to shuffle through everything to restore them again or you have to adjust your habits to leave a 30px gap at the left edge of the screen.

In short, there's a pretty good reason for CDE being unpopular. Sun has even dropped it and I believe they invented it. I don't know exactly what the desktop is changing to, but I'm pretty sure it's not CDE.

Reply Score: 5

RE[2]: What about CDE?
by Megatux on Fri 17th Mar 2006 20:28 UTC in reply to "RE: What about CDE?"
Megatux Member since:

mmm, I don't think that Sun invented CDE.
They loose in the toolkit battle with CDE (OpenLook vs Motif, I think)

Reply Score: 1

Krita and Koffice
by yokem55 on Tue 14th Mar 2006 19:43 UTC
Member since:

The author's jabs at Krita and Koffice are unfounded. Krita is NOT a gimp clone. Yes it has many of the same (and many more advanced) features, but it has been designed more for artistic painting, not raster image editing. The difference to those that aren't familiar with graphics programs may seem silly, but it does make a big difference to those used to professional tool sets. While Krita can and is used for image editing, that isn't its primary purpose, and to claim that it is merely reinventing the wheel does a great discredit to the hard work of Boudewijn Rempt and all the other contributors.

As for koffice, it preexisted the open sourcing of openoffice by a couple years I believe, and as of now, it is maturing into a very capable, lightweight (as in doesn't take 2 minutes to star up) office suite.

Reply Score: 5

RE: Krita and Koffice
by Nathan O. on Tue 14th Mar 2006 19:54 UTC in reply to "Krita and Koffice"
Nathan O. Member since:

I agree. I'm not sure what he was getting at with the "reinventing the wheel" thing. Krita and Koffice are very different from their counterparts. *Some* reinvention can be a good thing. I think of it as covering more ground rather than competition.

Reply Score: 2

RE: Krita and Koffice
by l3v1 on Tue 14th Mar 2006 20:50 UTC in reply to "Krita and Koffice"
l3v1 Member since:

I also agree completely. First off, KOffice is the older one so reinventing the wheel doesn't stand. Secondly, KWrite gives frame-based editing with the ability to include any other KOffice part in the document. Very nice, usable and fast.

Also, just take a quick look on the parts of KOffice, and try to search for anything even similar in gnome or else. Yes, you can come up with openoffice as an example but koffice still has a lot more to offer in many areas than

Reply Score: 4

RE[2]: Krita and Koffice
by Mitarai on Tue 14th Mar 2006 22:52 UTC in reply to "RE: Krita and Koffice"
Mitarai Member since:

koffice still has a lot more to offer in many areas than

Like what?

Reply Score: 1

RE[3]: Krita and Koffice
by Emerson on Tue 14th Mar 2006 23:14 UTC in reply to "RE[2]: Krita and Koffice"
Emerson Member since:

Like what?

Speed, gui responsiveness, smaller memory footprint, looks, and system integration.

None of that matters enough for me to use it, but it does to a lot of people.

Reply Score: 2

RE[4]: Krita and Koffice
by Mitarai on Tue 14th Mar 2006 23:28 UTC in reply to "RE[3]: Krita and Koffice"
Mitarai Member since:

Speed, gui responsiveness, smaller memory footprint, looks, and system integration

It is easier to open Office to archive those goals than for KOffice had the features OO.o has.

The points you mentioned are useless if it doesn't have the features users need.

Reply Score: 1

RE[5]: Krita and Koffice
by superstoned on Wed 15th Mar 2006 11:46 UTC in reply to "RE[4]: Krita and Koffice"
superstoned Member since:

that's SOOOO wrong... Koffice is some 13 mb in source, OO.o is bigger than the linux kernel!!! cleaning up OO.o would take hundreds of manyears, while adding the features Koffice misses compared to OO.o would take tens of manyears.

this is typical for KDE - a better framework, just not 'finished'. take a look at krita - its 1.5 release added to the last release several features Gimp has been trying to add for YEARS. adjustment layers and full colorspace independence - these 2 are seemingly impossible to add to Gimp, and where easy for Krita. the power of a good framework shows.

lets say, you want to add a spreadsheet as a layer in krita. you can. now you want 2 effects - do it with an adjustment layer. you can always change the spreadsheet, or even replace it by a kword document... OO.o and gimp could never do stuff like this.

again, cleaning a few hundred mb's of pure CRAP or adding features to a clean and powerfull codebase of 15 mb - i think any developer can take the right pick.

apple did... they choose Khtml over Gecko... quallity over quantity, any time.

Reply Score: 2

RE[6]: Krita and Koffice
by prokoudine on Wed 15th Mar 2006 12:48 UTC in reply to "RE[5]: Krita and Koffice"
prokoudine Member since:

> take a look at krita - its 1.5 release added to the last release several features Gimp has been trying to add for YEARS. adjustment layers and full colorspace independence - these 2 are seemingly impossible to add to Gimp, and where easy for Krita. the power of a good framework shows.

You are misinformed. Krita's success is the result of close team work. Framework is barely related to it.

Reply Score: 1

RE[7]: Krita and Koffice
by superstoned on Wed 15th Mar 2006 14:37 UTC in reply to "RE[6]: Krita and Koffice"
superstoned Member since:

i'm sure they would have had a much harder time if they had to work with c and gtk...

Reply Score: 1

RE: Krita and Koffice
by someone on Tue 14th Mar 2006 21:16 UTC in reply to "Krita and Koffice"
someone Member since:

I agree...

Krita is not even aimed at the same crowd as Gimp. Gimp is aimed at the photoshop crowd, while Krita is aimed at the Corel Painter crowd. Krita also offers 16bit support, which is still absent from Gimp (and GEGL is still in the early stages even after all these years)

As for KOffice, it began long before Sun open sourced OO.o. In addition, OO.o's codebase is almost impossible for an outside contributor to understand, which means Sun is still responsible for most of its development. KOffice is much cleaner than OO.o and some of its concepts (the DTP-infused Kword) are novel compared to OO.o, which is basically a MS Office clone.

Reply Score: 3

RE: Krita and Koffice
by grat on Wed 15th Mar 2006 02:26 UTC in reply to "Krita and Koffice"
grat Member since:

I'll agree that the jabs at Koffice/Krita were unfounded, but I have to also take exception with your shot at OpenOffice for taking "2 minutes to start up".

On my Athlon XP 1800+, 512mb, OpenOffice Writer 2.0, it's around 20 seconds, with no preloading (SuSE 9.3, if anyone *really* cares).

Now, if you'll excuse me, I'm off to look at Krita... even though my primary graphics thing is image manipuation, not painting. ;)

Reply Score: 1

Desktop is here to stay
by Jamie on Tue 14th Mar 2006 19:49 UTC
Member since:

Interestig article but the author failed to state or deliver any visions of what might be a better replacement (ignoring the start trek ideas of voice recognition et al).

For me, the desktop metaphor does not need changing but refining. A more task/project driven interface and smarter use of metadata that allows your desktop to adapt to your needs is possible and this will bring a higher level of efficiency to all users.

Fewer mouse clicks and less micro managing the environment will increase productivity and ease of use.

Creating profiles for games, office, home etc where your desktop adapts to suit your mood and needs should also be possible with a single mouse click.

Reply Score: 5

Things are improving
by ronaldst on Tue 14th Mar 2006 19:56 UTC
Member since:

Vector UIs, etc... Steps are being removed from doing everyday "computer chores" which in the end will make them easier. The most problematic is going to be dealing with people's habits. Some people don't like going to retrieve money from ATMs and we can't change that. Apps are easier to write nowadays with .Net. People are more focusing on what the user needs then what they might need.

Here's my favorite mainframe to average compuuter joe improvement: remote services, now called web services. IMO 5250 sessions are the great. I was hooked the first time I crossed an IBM AS/400. Later on, when I started to play my first MMORPG, AC, the first thought I had was how great it was since I don't have to mess with everything, setting up stuff, moving files to another PC. Just log in and get to work (or play in this example). I could just go visit my friends and install the game then my content would come to me. Web service. No fussing around.

Gmail is a great example of this. I know plenty of have stopped using Outlook Express replacing it with gmail. We now have a FREE with ads email client inside our web browsers. The content follows us. Requires NO installation other than a shortcut on the desktop or on the favourites bar. Web service.

Microsoft did the same with "remote desktops". I was completely turned off at VNC. Clunky and slow, it was horrible. However once I tried XP's remote desktop I couldn't stop using it. It's a little gem that Apple needs to look into. Web service.

There is so much to improve. Everything is to improve.

Reply Score: 2

RE: Things are improving
by Peragrin on Tue 14th Mar 2006 21:24 UTC in reply to "Things are improving"
Peragrin Member since:

Microsoft did the same with "remote desktops". I was completely turned off at VNC. Clunky and slow, it was horrible. However once I tried XP's remote desktop I couldn't stop using it. It's a little gem that Apple needs to look into. Web service.

Um it's called Apple remote Desktop and has been there just as long as XP's. Both are basically enhanced VNC setups.

I was hooked the first time I crossed an IBM AS/400. Later on, when I started to play my first MMORPG, AC, the first thought I had was how great it was since I don't have to mess with everything, setting up stuff, moving files to another PC. Just log in and get to work (or play in this example).

I felt the same way when I learned to use Unix on my college's network. I could log into my files and setup no matter where I was on the network. When i switched terminal's to SGI's I got new desktop options that weren't available on the other thin clients. This was in 1996. I spent 7 years trying to get similar functionality out of windows, and finally settled of OS X. It comes closer to ideal usability.

Now for an On-topic rant. The Author basically does a plus/minus of the various popular DE's. He doesn't address the question of the desktop and is it still useful. My personal opinionthat modern DE's focus to much on the mouse and not enough on the keyboard and actually entering Data in.

But that's my opinion. I live with an open terminal window on my Powerbook. between that and spotlight I can launch most apps without ever taking my hands off the keyboard.

Reply Score: 1

WM concepts...
by jimaz on Tue 14th Mar 2006 20:13 UTC
Member since:

I just wrote down some ideas I had a long time ago reg. a wm which at least allows separate desktop icons in each workspace, something the popular wms don't do today. That way the desktop could be split into separate workspaces for each 'task', one for coding, one for graphics etc..

(Warning - web site still under development)

Reply Score: 3

RE: WM concepts...
by NicolasRoard on Wed 15th Mar 2006 00:06 UTC in reply to "WM concepts..."
NicolasRoard Member since:

I just wrote down some ideas I had a long time ago reg. a wm which at least allows separate desktop icons in each workspace, something the popular wms don't do today. That way the desktop could be split into separate workspaces for each 'task', one for coding, one for graphics etc..

Actually, we want to implement something a bit related for étoilé..
Roughly, something similar to Squeak projects, but implemented as virtual desktops. Eg, manage (partly) your environment around the notion of "projects" rather than folders and files. A project would be in fact a virtual desktop, plus some persistance mechanism that reload the proper apps, opened documents, etc.

In addition we also want a tabbed shelf ala OPENSTEP beta, but that's another thing ;)

Edited 2006-03-15 00:07

Reply Score: 1

RE: WM concepts...
by rcsteiner on Wed 15th Mar 2006 16:04 UTC in reply to "WM concepts..."
rcsteiner Member since:

That's actually what I do under Windows 95 OSR2 with an older desktop replacement called QuikMenu 4. I have a main desktop screen containing mostly game icons, and a couple of other desktops on subsequent "pages" which contain common utilities and applications.

It helps to hide a lot of the clutter...

Reply Score: 1

RE: WM concepts...
by rcsteiner on Wed 15th Mar 2006 16:08 UTC in reply to "WM concepts..."
rcsteiner Member since:

One of the things I still use heavily under OS/2 is the Workplace Shell concept of Workgroup Folders. You can organize data files and programs in a folder, mark that folder as a "Workgroup", and then open/close all of the apps and data files as a logical group by opening/closing that folder.

It's a slick way to group a programming IDE, help files, and various other notes together into a single execution unit. Let the desktop do the work for you. ;-)

Reply Score: 1

by netpython on Tue 14th Mar 2006 20:17 UTC
Member since:

The driving force behind this article is the fact that the latest buzz in user interfaces is to use 3D and transparency everywhere. Many users are excited and feel that the way they use their computer is going to change in some magic way. They cannot understand that eye candy is completely different from usability improvement.

It's not only eye-candy.That would be a great understatement denying there's a lot of development and innovation behind projects like Xgl,etc.

The desktop methaphor serves it's purpose,being an suitable symplification of something being transferred in whatever format to whatever location.

It's a convienient symplification for the average desktop user.Should we suddenly place the steering wheel on the rear seat?(Maybe:-)

Though everybody is free to improve how they feel fit's their expectations well.I'm sure there's still a lot of headroom for "innovation".

The prime is:"Does it work (well enough)?".

Reply Score: 1

Member since:

Pages long of opinions - can't call them more since most of them lack any real support -, praising Apple's engineers and the lack of clutter in Gnome (I... will... not... flame... gosh, it's so hard) and where do we get: we need open minded devs and new concepts.

Please, next time ask your article writers that if they put a question in the title, then at least make a minor effort to try finding something which could sometime become something similar to an answer... ;)

Reply Score: 5

Member since:

I use everyday Debian pure compiled everything.
For average joe EASE OF USE and interconnected applications are the most important than fancy looks of the desktop. Right click and email word file with outlook..Right click and do anything....
Desktop is just like a shirt put on the human body. In this modern age we need not only well built healthy body but also shiny looking shirt on it also.

Linux/OSS has historical opportunity of developing an intutive and revolutionary desktop before introduction of XP/OSX But as many other genius people can't get ideas into reality, OS developers also wasted this golden opportunity to take their own lead. Result, in 2007 they are still chasing to mimic XP/OSX desktop functionalities and ease of use.

KDE/GNOME Copycats may have arguments supporting to lure MS users into linuxland but naked truth is less than 5% marketshare and willbe same in 2010.

Reply Score: 1

Member since:

... wait and see.

I don't think we'll be seing a major change in the way we interact with computer in the next 5-10 years. (3D technology has been with our PCs for 10+ years, and we only consider using it for the desktop *now*?)

We'll see an accumulation of features, various dead ends, and in time the next major step needed will be obvious.

skip the next paragraph
Right now I could use a good, reliable replacement for the tradional file system with directory hierarchies. (Desktop search is good but i still have to save those files somewhere). Drag'n'drop everything, less right clicks menu, more (optional) keyboard shortcuts, more eye candy, not too much of it. Sane defaults. More than 1 set of them, so that I don't have to either configure everything by hand or go with a default I don't like. But that's just me.

And that's my own non-vision for the future of computer desktops.

P.S. To Kostis: leleeee! ;-)

Reply Score: 1

by viton on Tue 14th Mar 2006 22:30 UTC
Member since:

Heh i rarely use desktop for things it is designed for.
Usually i just launching applications.
The default file managers in windows or linux makes me sad.
These tools are useless for me because i need something more than just basic file operations.
I hardly find it interesting to drag the stupid icons over the screen.

Edited 2006-03-14 22:31

Reply Score: 1

by bytecoder on Tue 14th Mar 2006 23:09 UTC
Member since:

After skimming through that, I was hoping to see the author's intended solution. Unfortunately, I did not find one.

My opinion on the subject is basically that the desktop needs to go. Why in the world putting files behind windows so you can't get to them was thought to be a good idea is beyond me. I must point out, however, that I still find a hierarchical (not relational) filesystem best for pretty much all purposes, though.

To sidetrack for a minute, I'm perplexed as to why people seem to like Mezzo so much. Aside from it's rather rough implementation, the concept just seems to be incredibly inefficient and backwards for the user. Again, we see content being hidden by windows, and the only floating elements utilized are 4 little buttons on the corners. I'm also rather against the task/app-based nature of it, which tends to be slower to work with and less consistent between different tasks.

Finally, I'd like to point out how little the filesystem is utilized as a UI element. First, let us derive the basic purpose of the filesystem. Basically, it's a hierarchy of folders which contain abstract items, of which the only requirement is that they're either a file or directory. Using this definition, it can be easily seen that folders are essentially lists of basically anything that can be fit into a file or folder.

Now, let's look at a few other common list-type objects. Possibly the most common thing would be an application menu. Not only could the filesystem subsume the functionality of said menu, it could also do a much better job at it. Probably the biggest reason why people don't think this would work is because the usual layout of an application folder (e.g. Program Files) is terrible. Application files are loosely spread throughout the filesystem, and actually executing said application manually requires trudging through its directory and trying to find the executable. Of course, it doesn't have to be that way. Like I said before, almost anything can be made to fit into a file or folder, and if it falls under the category of a single entity, it generally should be. Since an application is a single entity, it can be found that an application can and should be fit into some sort of contained file/directory; enter application directories. Now, not only is executing applications completely consistent with other list-like tasks, you can also move and delete them from a single interface. The best part of all, of course, is that you don't even need to write an application to do it! Just use the file manager of your choice.

Reply Score: 2

RE: disappointed
by Ronald Vos on Wed 15th Mar 2006 00:11 UTC in reply to "disappointed"
Ronald Vos Member since:

I just want to say I agree with Bytecoders ideas here.

The article, besides not pointing out a solution or a better metaphor, also doesn't touch the heart of the matter. Which is workflow. Metaphors are nice and all for learning to work with a system, but they fail badly once you get functions implicated by the metaphor but not actually present (the infamous 'metaphor sheer' as coined in ), and metaphors also fail when a piece of functionality isn't covered by the metaphor. Two famous examples are documents you work on aren't actually there for you to continue working on unless you save them, and networking, respectively. Both are mentioned in the article, but only touched on briefly.

Another aspect of using any metaphor for user-interface design, is that they guide systems-design towards a certain structure. And that structure can sometimes be very unproductive.

For example, the 'desktop'-space itself. Aka: that place that clutters with icons. The idea was that this was supposed to be your working space. Then it got used for quicklaunch-icons (Great idea, instead of merely putting icons in my quicklaunchbar/dock/whatever and my startmenu, also put them in a place that requires minimizing every window/application currently up before you can reach them!), links to essential system components ('My computer' wouldn't actually be so bad if it actually gave quicker access to required functionality and had a less condescending name) and random files a user might have chosen to put there (For quick access..after you minimise everything you're working on again).

If projects like Mezzo would focus on workflow instead of how easy it is to reach a button when it's in the corner, they would require a lot less mouse-movements and cause less RSI.

Simple example: how come every system I used (and know of) allows you to copy and paste only one thing at a time? (Someone please correct me if there is something out there.) Plenty of times have I wanted to alternately paste 1 thing, then another, depending on context. Or a link first, then a piece of text.
Present systems require me to take one piece of copied data, like a link, to another location, and then going back again to get the other thing I wanted to copy-paste, like a piece of text. And then forth again, to actually paste it. This takes either 4 alt-tabs going from program I'm working in, to location of data, back again, forth again, back again, or 8 mouse gestures (Mezzo or not).

NeXT's dock was nice in decreasing the pain in doing this, ION is even better since getting 2 windows next to eachother so you can work from alternating programs is a *lot* easier. You can do stuff like that in Windows too, for example by making Firefox occupy half your screen, and your favorite filemanager the other half. But then you run into the problem that Firefox is designed with the desktop metaphor in mind, expecting to get the full screen, and rendering pages centred in the screen with wide margins for most sites.

Another problem with desktops isn't per se the metaphor, it's that they're application centric. Handing over a piece of control of the system over to programs is bound to leave users with a slightly frustrated feeling when a program selfishly hogs screen-estate, ransacks keybindings or most of all: doesn't work well with other programs. That's something BeOS had an edge in: datacentrism. They first came with clever and easy ways to handle data, and then they let users and programs toy around with that. Which led to an indexed filesystems with extended and flexible attributes which allowed for great versatile data management, and an object oriented desktop that allowed scripts and programs to transparently control other programs. Bottomline: you could easily invent ways of boosting your own productivity. And I'm sorry to see only OS X coming close in replicating this (and not being well-advertised there)..
Vista is coming closer though, with it's object oriented CLI able to control programs, and WinFS maybe being implemented some day.

Bytecoder's suggestion of integrating program-manager with the file-manager is one I love, and one I meandered about in the past. Removing programs shouldn't involve digging in a submenu of the startmenu, looking for the right icon in a sea of icons, and then finding a program buried somewhere in a list, only to then have to check if the icon on my desktop, quicklaunchbar, start menu and programs menu (different things!) have all been removed. Or that part of it wasn't removed because I installed/moved/renamed icons myself (programs I use often but not often enough to put in the quicklaunchbar I usually put outside of their folder in the programs menu in Windows, so I only have to enter 2 nested submenus instead of 3 when looking for it).

Anyway, to cut a long ramble short: merely going for another metaphor isn't a be-all end-all solution. Different paradigms aren't necessarily it either. For those proclaiming the 'zooming interface' holy, try out the BluebottleOS livecd. It has some awesome interface thingies, works, and is usable as a production system (somewhat, considering it's limited # of apps), but I was completely dumbfounded when I wondered where I had just saved my sourcecode file, and getting to the right zoom-level in the right spot without losing what I was working with drove me nuts. The first is an issue of metaphor, the second is an issue of interfacing and workflow.

Reply Score: 2

RE[2]: disappointed
by cerbie on Wed 15th Mar 2006 12:30 UTC in reply to "RE: disappointed"
cerbie Member since:

I do like Mezzo (not necessarily the web-centric ideas, though), but also agree with you. The problem with a project like Symphony is that so far, they've thrown the baby out with the bathwater. Fitts was right is good, but sometimes you have to tell Fitts to go away, because feature density is more important.

The desktop metaphor does need to die. That does not mean something radically different needs to take its place--it means we need to stop being mentally boxed in by the limitations of the metaphor, and see the interface as a set of features sitting on top of a filesystem.

For copying and pasting, I know of nothing worthwhile that is system-wide. MS Office's clipboard manager thingie does a good job, though...within Office. I would love to see C&P stay as they are at the first level, but use a stack that can then be accessed by the user (so single copy and paste operations work as normal, multiple ones follow with the lastest being first, then you can get a list with info on them, and reorder and drag and drop them). With this, it would be simple to use the keyboard or menu options for one or two operations, and then easy to use the GUI to handle more (a menu for more would not be great, because with different kinds of data copied, like images, getting a thumbnail, application that the copy was performed from, etc. could be very handy).

Reply Score: 1

Last decade?
by AppleFollower on Tue 14th Mar 2006 23:09 UTC
Member since:

Let me take a moment to correct this statement:

"The desktop metaphor has served our computing needs well for the last decade."

To the more accurate:

"The desktop metaphor has served our computing needs well for MORE THAN the last TWO decadeS."

Edited 2006-03-14 23:10

Reply Score: 1

The Desktop is dead, long live the Desktop
by grayrest on Tue 14th Mar 2006 23:21 UTC
Member since:

Whether the desktop is dead or not depends on your view of what makes up the desktop. The original desktop metaphor broke in Windows 95 with the introduction of the start menu and windows explorer as opposed to the mac spatial file manager. You could even argue that it died earlier with the introduction of aliases, I think that was MacOS 4.

The modern desktop is not supposed to be a desktop, but rather a slow evolution from the desktop metaphor. I believe that something like the Mezzo desktop is the next step in this evolution. I do not believe that non WIMP interfaces like Raskin's THE or wmii's frames metaphor will ever catch on because the interaction model is fundamentally different from most people's current experience.

In the longer term, I expect the desktop to primarily become an interface for interacting with web pages. I would argue that the current concept of applications is changing and will change dramatically in the next few years with the coming ascendancy of web based applications. Web based apps have a huge number of business advantages (no distribution, no install, no patching, no piracy) and technical advantages (auto cross-platform, choose your server implementation language and environment). Web apps have been big even with heavy user experience flaws (full page refresh every action).

As javascript toolkits mature and browsers pick up SVG the platform will become considerably better as a windowing layer. Moreover, I see the current developments in web page interaction as a chance to introduce new interaction conventions. It's an exciting time to be an interface designer.

Rapid app development frameworks (seaside, rails, turbogears, django) on the server side remove many of the hassles of dealing with putting apps on the web, allowing devs to focus on business logic. Even faster development is possible if your app is just babysitting a database (think access). Take, for example, the jot wiki or dabbledb. Both allow you to slam out a db app in a few minutes while having it look pretty good and both knock the socks off traditional development even in their immature state.

Back to the discussion at hand. I don't see the development of 3d environments as providing anything but eye candy without some new input method. The desktop/WIMP interaction style works because we have a 2d input device (mouse) and a 2d display to view mostly 2d data (text,video). Neither the mouse nor keyboard really facilitates efficient 3d motion (though the scroll wheel makes one dimensional z motion possible) and without such a method, I don't believe usable interaction with a 3d environment is feasable.

An aside:
I feel I should mention that I'm a wmi user. I'll add that I think wmii as currently implemented has severe and fundamental usability issues, namely that it moves your stuff around behind your back. This is the point of a dynamic wm, but it kills consistency and learnability of switching between windows.

Reply Score: 2

by Kroc on Tue 14th Mar 2006 23:24 UTC
Member since:

"It has started however, to show its age over the last years."

Er, no it hasn't. It's only just started to become usable to average people now, there's still tons more to explore. Apple's Tiger is an example of the desktop metaphor still having tons of freshness left.

Reply Score: 1

People are...
by Tuishimi on Wed 15th Mar 2006 03:22 UTC
Member since:

...visual. We like graphic representation of objects. That's why GUI's exist. The desktop is not going anywhere.

Reply Score: 1

by Jankdc on Wed 15th Mar 2006 05:21 UTC
Member since:

This article gave my a chance to look at the Mezzo desktop. I really like it.

Reply Score: 1

Something Round
by seishino on Wed 15th Mar 2006 07:07 UTC
Member since:

One way to make a wheel that rolls well is to find a series of wooden slats, calculate their relative lengths at a particular distance, bend them together using a heating element of some sort, then add all sorts of slats / supports / and other structures to keep them in place.

Or, you can find a big rock, and chisel it down into something round.

As the article points out, we've been layering UI on top of UI for years now, until nobody knows whether the application they're looking for is in the start bar, the program bar, the quicklaunch bar, the desktop, or on their hard drive. Or which of the many configuration utilities they should use in Linux/Debian/KDE.

It's time to start cutting that down. It's time to create User Interfaces that have ONE or TWO ways of doing something that always work, rather than 10 or 20 ways that sometimes do.

Good design doesn't come from knowing which features to add. Good design comes from knowing which features to cut.

Reply Score: 1

context menu
by axel on Wed 15th Mar 2006 07:39 UTC
Member since:

me, i really don't care what metaphor it uses. even with the filemanaging/webbrowsing/whatevering multitasking i still use the OS/Desktop as little more than a glorified application launcher/switcher everything i do is still interacting with applictations not the desktop, which brings us to workflow

what i want is better context menu's
-beos' tracker has a nice one, you don't need drag and drop really thanks to both copy to and move to menu's which i'm a fan of, all drag and drop means to me is more windows, dnd isnt abysmal but i really don't see it as any more convienient than copy paste/cut paste
-i also want an 'open with' menu that allows me to load anything with anything, not just things the system thinks i should open it with (obviously there's a time and a place for this so like an open with sub menu with both several logical programs and a sub menu that is essentiall just the application menu)
-copy and paste like vos said easy access to multiple independant pasteboards. your standard copy past ctrl c ctrl v sure but in the context menu have copy and submenu copy2 copy3 copy4 copyN, as well as paste and submenu paste2 ... (preferably with a small title for contents). as well as a copy to option that lets me copy say a bit of text to a file, just tacks it on to the end of the file (see mozilla extension quicknote), could work for images too be a bit less useful though, or at least a bit weirder.
-also an omni present root menu. flux box for instance has the root context menu not just the application menu but also the WM configuration, background switcher, with the right plugin it's also an RSS reeder and i think can show contents of a mail box (which would be easy enough to do with a beos every mail is a file) i want that functionality present whether or not i can find the root window, again a sub menu.

organize it right and you add maybe four or five lines max onto an applications normal menu and you never have to find a corner or desktop space again (if you don't want too)
course having the whole filebrowsing/copyingpasting saving context menu thing requires a file browser that supports that kind of thing so as does the omnipresent system options. so you'd have to program apps with the goal allowing it work might be a lot of work.

on the filemanager/program manager thing, am i missing something or aren't you just describing the osX/appdirs concept
and as launching and app manually from the file manager:
everything that has a .desktop file and it'll launch from a file manager.
beos is what /beos/system/apps
osx is applications
it's just windows thats bit behind the curve.

Also on Mezzo.

am i the only one that realizes it is EXACTLY THE SAME as every other DE out there? it's sum total innovation is having the system decide what icons are on your desktop instead of you
past that we still have a corner menu for launching applications, and a couple buttons in other corners (all of ten seconds work in gnome or xfce4) for file browser system and trash.
wow amazing.

Edited 2006-03-15 07:47

Reply Score: 2

nothing innovative
by wkornew on Wed 15th Mar 2006 09:45 UTC
Member since:

This article is only about the well-known systems. The author did not even mention projects like Archy or Croquet. Of course, ZUIs like Croquet are not the final solution. They are ineffective for many every-day tasks, but we can learn from their concepts and maybe combine ZUIs with other ideas to make them more effective. For example, "location" is not a (textual) path, but a real location in space. This is more natural and easier to remember, but more difficult to organize (micro-management of locations, big mess, difficult to reorganize). Also, "files" are not opened, anymore (simply zoom into the file) and stand-alone applications can behave like simple documents (no more distinction between "application" and "file"; many aspects of "applications" become functionality extensions/add-ons).

There are concepts of new input methods. E.g.: your desk becomes a big touch-screen where you can use both hands (and all your fingers) to get more freedom in expressing actions (more powerful gestures). This has also been combined with the ZUI concept and it looks very interesting. Other ideas work with projections (instead of touch-screen) and either your hands or physical input devices (direct interaction with projection, not indirect like mice) are used to interact with the UI.

There are people who think that everything should be query-based. No more folders for organizing files. Tags do this job. Maybe with taggable tags you can create relationship networks between tags, so our computers can still create complex structural visualizations on demand (e.g.: when developing software paths are too inflexible and one-level tags not powerful enough).

For the distant future we could imagine organizing data with semantic relationships (the EU already started a project based on Linux). Computers will begin to "understand" (to a certain level) the meaning of your data and allow for more intelligent representation.

We should think about better ways to organize our data (i.e.: automate this process!) and how to improve collaboration instead of how to make UIs "cooler" or how to re-invent the Windows wheel (look at KDE, GNOME, etc.).

Reply Score: 1

RE: nothing innovative
by grayrest on Wed 15th Mar 2006 19:46 UTC in reply to "nothing innovative"
grayrest Member since:

This article is only about the well-known systems.

This disappointed me as well, the author built up this big intro and wound up with a rather weak survey of the current destkop environments rather than a survey of unusual ones. The redeeming value for the article is the mention of the Mezzo desktop (which I had missed) and the opportunity for discussion.

your desk becomes a big touch-screen where you can use both hands (and all your fingers) to get more freedom in expressing actions (more powerful gestures).

LogicWorks tried to do this but unfortunately they're out of business. Part of the problem with an approach like this is that touch typing is simpler with tactile feedback. I should try to grab one of their gesture pads before they become impossible to find.

We should think about better ways to organize our data and how to improve collaboration instead of how to make UIs "cooler" or how to re-invent the Windows wheel (look at KDE, GNOME, etc.).

You seem to be ignoring the value of past experience on usability. Using any UI is a learned behavior and most research projects deliberately refuse the evolutionary path and instead re-think things from the ground up. This is well and good when there aren't accepted norms, such as the developing ajax web experience, but it makes these research projects very difficult to push into common use in established environments.

Take mouse gestures as an example. They don't require special hardware or a new OS, they can significantly speed up mouse interaction with a program, but they still very niche on the desktop with the main user base being Opera and Firefox extension users. Every time I show them to non-geeks it takes a few minutes of explanation to update their mental model of how a mouse is supposed to work and why they'd want to use these gestures.

Reply Score: 1

nice article
by superstoned on Wed 15th Mar 2006 11:54 UTC
Member since:

it was a nice read, tough the writer clearly prefers gnome over KDE - which results in some unjustified stabs at KDE. especially when he says KDE is not trying to clean up - he never saw a KDE release after 3.2??? many cleanup work has gone in KDE already, redesigned dialogues, menu's and toolbars. he should have a look at kubuntu 6.04 when it's out...

anyway, i think for the coming years, KDE is FOSS' best shot at getting ahead of the Other OS'es like Windows and MacOS... the KDE dev's are fully willing to innovate for KDE 4, and every idea is welcome. they want to keep it usable, and they have the resources and technology to do it (automated usability testing will be used in KDE4's development! a first!).

Reply Score: 1

My view
by D-J-P on Wed 15th Mar 2006 12:42 UTC
Member since:

I think in the future we start to have more ipod like, mediacenter like, frontrow like interfaces.
Interfaces that are easily controlled by touchscreens.

The startscreen is simple, what do you want to do?
-play dvd
-watch photo's
-listen to music
-go on the internet etc.

This next video also shows how the future of desktop computing could look like. With a multiple touchscreen which is in front like a book on your desk.

Reply Score: 1

RE: My view
by grayrest on Wed 15th Mar 2006 19:11 UTC in reply to "My view"
grayrest Member since:

This style interface has been tried for years (The first machine I played with after my Apple IIe had an interface like this) and it really doesn't work well for a general purpose machine. Once you go over a few activities the interface bogs down and you wind up with deeply nested menus, which are not immensely discoverable. It does work on purpose built machines like media center pcs, web tablets, xbox 360s, ipods, ATMs, etc. I'm not saying that interfaces like this won't happen, but the interface expectations and form factor of the machines it will show up on is considerably different from the traditional desktop.

Reply Score: 1

Metaphor Definition is Suspect
by RGCook on Thu 16th Mar 2006 03:50 UTC
Member since:

The article defines the desktop metaphor as such:

By desktop metaphor we mean the concept where the Graphical User Interface represents an office desk on screen.

I'm not sure that is correct. Every window manager represents a unique abstraction of the computer itself - accommodoating its inherently unique interface and operational properties from that of a simple office desk. This would explain some of the breakdowns in the metaphor presented by the author.

Perhaps the biggest flaw in the metaphor is that it is not a suitable metaphor after all. In the object-oriented world of software development, it is understood that a model must be based on a suitable abstraction of the problem. The resulting interface and functionality of that abstraction may not even approximate the real world objects on which it is based. Not because it is a bad model, but due to limitations imposed or placed on the model (on purpose). This is true of games as well. After all, we don't click a mouse to fire a rocket launcher do we? No, there is usually a trigger involved in a real world gun. Does this mean the metaphor is flawed and we must rewrite the interface to more accurately describe the real-world object (circumstance)? Of course not. We implement the unique interface exposed by the hardware (our laptop, desktop computer or PDA) and abstract the real world object or problem to accommodate the needs and limitations of the hardware.

Each attempt to rewrite the so called "desktop metaphor" is simply going to result in a different abstraction of the same problem (interfacing with the computer) based on the inherent limitation assocated therewith. Therefore, if you want to change or improve the user experience or make the computer easier to operate and fun to use (what's the difference?) think about what us visual creatures need.

For example, you can't "feel" the glow of the warm sun telling you its going to be nice and sunny out today and you need to get off your butt and get off the computer and get some Vitamin D. The monitor is limited in its ability to radiate heat (or cold) in concert with that desktop widget. And God forbid it spray water on you as a means of telling you that it is raining outside. But these are tongue-in-cheek instances of a more accurate abstraction. Does it improve the interface or present a better computing experience? Probably not in my silly weather example. So the metaphor is flawed here too, even as we move closer to real life.

So where is the interface to be improved to expand on the abstraction in an intuitive manner? I think this speaks to a level of maturity we have reached with current windowing systems. The polish that is now going on is a result of efforts to get the last squeal out of the pig - if you will.

Reply Score: 1