KDE Desktop Usability Survey

The HASE (Human Aspects of Software Engineering) group at the University of Maryland Baltimore County is conducting two online surveys for KDE users. They expect to obtain results that will lead to useful discussions about the overall usability of KDE. The results of this research will be shared with the KDE community. The surveys take about 15 minutes to complete.

Dot Categories: 


by Janne (not verified)

"People who get confused because they see three text editors in KMenu"

So, that is your definition of "typical end-user"? If that is the case, then shouldn'tt there only be one editor? I mean, the term "typical end-user" points to majority of users (since they are "typical" users"), then it means that most users find KDE to be confusing in that regard. Solution: remove reduntant editors. Or are you saying that KDE is not aimed at "typical end-users"?

Vanilla-KDE is meant for ALL users. you seem to think that vanilla-KDE is meant for some uber-leet powerusers, whereas "lusers" can use the distributors dumbed-down version. I don't want that. I want vanilla-KDE to be kick-ass on all levels. I want KDE (as shipped by the KDE-team) to be the shining beacon of kick-assness, I don't want to see distros holding KDE's hand so they could "get it right".

"Well then the vanilla KDE can't be as bad as you try to make it."

By default, there are flaws in KDE. And the distribution-versions are 98% identical. They might have different windecs, different styles and different icons and the like, but mostly they are the same. They haven't really tweaked KDE that much. So expecting distros to fix KDE simply does not work, since most of them don't do anything major to it! And when they do, people complain (Red Hat & Bluecurve, Xandros and their filemanager, Lindows with just about everything etc. etc.). So, instead of expecting distros to fix the problem (which they aren't going to do), why not fix the problem right at the source? that way ALL users would benefits, instead of just users of some particular distro.

"Well where is your prove that many people complain about KDE's abundane of options?"

That seems to be the nr.1 complaint KDE seems to receive (together with general clutterness, but they both tell about the same problem), from KDE-users and non-users alike. I'm sorry, but I haven't documented, counted and graphed them all, so I don't have any studies or diagrams to show you.

"So you want a minimalistic KDE and I don't. So how should vanilla KDE solve this "conflict"?"

"minimalistic" might be a wrong word, but I didn't have better one at hand. What would I like my KDE to be like? Like this:

1. A set of core utilities that get the job done, and tyey get the job done well. Yes, that means only one text-editor, and not full-featured editor and criplled editor. Instead of thinking "someone somewhere MIGHT use this particular app once or twice, so we better ship it alongside KDE", we should think "do we REALLY need this app? Is it really essential? Couldn't the same thing be done with this other app here? Or maybe we could merge this app with this other app?".

This change would mean that KDE would be smaller, it would take less time to download/compile/install and it wouldn't be as cluttered with apps as it is.

2. Configuration-options that don't scare users away. I have seen new users try to configure KDE. I show them where they do it (Control Center), they open it... And their jaws drop. They are overwhelmed. They click around, look around... And close Control Center. There is simply too much stuff there! they don't know what to do with all of it! It scares them.

Before you say "but those options empower the user to tweak the desktop to his exact liking!"... No. those options bury the relevant options under hundreds of other options. It doesn't empower the user, it drowns them.

As folks at OpenUsability.org said about Kmail:

"The first step is to reduce the amount of settings that a “normal” user is confronted with in the properties dialog. Issues that either require deep technical insight or that is relevant only in 1% of cases should be kept “out of sight”, but still accessible, or put to another place. Reducing (required) information to the relevant makes the user feel safer that he can control his Actions (because he only must set what he understands)."

That could be applied elsewhere as well, besides Kmail. And that is what I have been advocating. Having multitude of options thrown at the user does not "empower" him, it confuses him and drowns him in clutter. No, I do not suggest that we remove features from KDE, if that's what you are thinking.

When the user tweaks his desktop (or some app for that matter), he should feel that he's in command. But if there are loads of options, the user will feel that he not in command. That he can only scratch the surface. If he changes one setting, there are still 347 settings waiting for him, each with several different choices. With smaller set of default options, that would not happen. But if the user REALLY feels the need to change some exotic setting, he could still do it. I think GNOME-folks were on the right track when they introduced Gconf, but I think their implementation sucks.

What would this KDE be in reality? It would be welcoming towards new users. It would look smooth and uncluttered. User would feel that they REALLY are in charge of their desktop (right now they aren't, due to multitude of options). The desktop would grow with the user. If he wants to change some less-used option, he would still have the tools to do so. But they wouldn't be thrown at him the moment he decides to change something. If the user thinks "I think I would like to change my icons...", KDE replies to him "Good idea! You can also add menubar on top, change launch-feedback, edit keyboard-shortcuts, configure file-sharing, tweak your taskbar, change your splashscreen, configure login-manager, change the spacing between icons in Konqueror...". At that point the user would say "enough!". that should not happen.

The configuration-tools (Control Center) would contain only about 10-20% of what they contain today. They would be easily accessible and the user would know just about all of them. He wouldn't have to look around for the thing he wants to change. He would be in control. The other options would be available through different set of tools (no, not necessarily through Windows Registry-clone like in GNOME).

In short: nothing would be taken away from the user, quite the contrary. By limiting the number of options that the user sees by default, we would empower the user to tweak the UI. If we drown him with options, we do not empower him, we drown him. We confuse him, and we will scare him away.

Sorry for lengthy post.

by cm (not verified)

> > "Well where is your prove that many people complain about KDE's
> > abundane of options?"

> That seems to be the nr.1 complaint KDE seems to receive (together with
> general clutterness, but they both tell about the same problem), from
> KDE-users and non-users alike. I'm sorry, but I haven't documented, counted
> and graphed them all, so I don't have any studies or diagrams to show you.

I got the impression that a *lot* of those complaints are from trolls (not the Norwegian tribe) and GNOME zealots trying to bash KDE.

But I agree, simple and smart defaults and easier config dialogs would be nice. But at the same time the features should stay. The challenging question would then be how to make those apps and options accessible to the user, and how to introduce the cool apps and features to him (how can he learn about them if they're hidden or not installed).

Could this be solved by

- Starting off with a bare-bone KDE that contains
--- apps only for the most common tasks (e.g. web browsing, mail)
--- only one app per task
--- as few apps as possible per mime type
--- toolbars with sparse defaults
--- easy config dialogs with advanced options removed or hidden behind "advanced" buttons
- Offering an app that lets you configure the hidden settings (if any). Good descriptions of the options and a good search facility would be essential.
- Offering a (better) toolbar and context menu configurator with a search facility and good description of the actions (currently I have a hard time guessing what many of the actions do when all the info I have is their label (the one they also appear in the menus with))
- Offering an app in a prominent spot that lets the user discover available apps that solve a certain problem (task-oriented)
--- So you want to write a letter to Mom? -> koffice
--- Need a more powerful text editor? -> kate
--- Want to do web development? -> quanta
--- Want to view binary data? -> khexedit
--- Want to listen to mp3? -> noatun, kaboodle or amarok (with description and maybe a screenshot)
- With that app, letting the user add the apps to the KDE menu (or replace them(!), in case there's already one configured for the task or mime type at hand) either by installing it from some package or just by adding a pre-built .desktop file into the menu structure (if the app is already on disk but not in the menu)

This functionality would border on the one that has traditionally been provided by the distributors. The app would have to play nicely with the platform's package management.

Another question would be how the user can be introduced to new or hidden KDE features like IOSlaves, web shortcuts, desktop-wide mouse gestures (KHotKeys), app plugins...

by Janne (not verified)

"I got the impression that a *lot* of those complaints are from trolls (not the Norwegian tribe) and GNOME zealots trying to bash KDE."

I think that most of them are not some zealots that are out to bash KDE. Besides, openusability have also made similar comments, and I don't think they are trolls or zealots.

"But at the same time the features should stay."

Absolutely! I do not advocate removal of features! But they shouldn't be thrown at the user either (same as with options). The app should have a defined purpose, and it fulfills that purpose perfectly by default. But it could also have more advanced features, that could be enabled if the user so desires. But by default, it should be a straighforward app intented for some well-defined purpose. But if the user is overwhelmed by in-your-face features, even the more simple tasks could e more awkward that they should be.

Allow me to present an example from my KDE-installation: Konqueror. Konqueror is a multipurpose-app that has two functions at it's core: filemanagement and web-browsing. I defined those two as it's main uses. Then I started to think what is the common thing with those two: browsing. Konqueror is not a spatial filemanager (thank god!), it's a "filebrowser". And you browse the net as well. So I proceeded by removing all buttons from the toolbar that does not relate to that task. So I have only buttons that enable browsing on the toolbar (Back, Forward, Up, Reload, Home and Stop, not necessarily in that order). I also have the location-textbox so I can type in urls/filepaths. Sure, some of those buttons have limited use when doing filemanagement (like "stop"), but I dislike toolbars that change, so it was an acceptable compromise. Almost all of those buttons are needed in both of those tasks, whereas by default Konqueror has buttons that are not needed in all it's inteded tasks.

I did that as an experiment in order to see would it limit my usage of Konqueror. It did not. In fact, it made it more pleasant to use, because the UI was so clean!

You might think that that is an overly simplified approach to the issue, but it worked for me, and it worked very well!

"Could this be solved by...."

You are definitely on the right track there :).

"Another question would be how the user can be introduced to new or hidden KDE features like IOSlaves, web shortcuts, desktop-wide mouse gestures (KHotKeys), app plugins..."

We have that problem even today. Many times users are astonished when they find out some cool feature KDE has to offer. Hell, just few weeks ago I told few users about the audiocd-kioslave and how you can encode oggs/mp3's with it. They weren't exactly newbies and that feature has been in KDE since 2.0 (IIRC)! yet they didn't know about it!

I think the key here is user-guides and documentation. And we need a way to actually make the users read them! Unfortunately I have no silver-bullet for this. Interactive guides? Or guides with demo's? Or maybe a comprehensive "Introduction to the K Desktop Environment"-section in kde.org/KDE? That way non-users could check what KDE actually can do. And when the user logs in to KDE for the first time, that presentation would automatically launch. It would contain "What's new"-section, explanation of the basic-concepts, introduction to the features and the like.

by Corbin (not verified)

I just finished taking both surveys and for most questions related to GNOME I answered 'N/A', since I hadn't used that function in GNOME enough (most likely I've never done it once, or only a couple times) to be confident in answering the questions. I never do any searching in KDE so I answered N/A for all those questions (if I ever search for ANYTHING, I'm normally already in Konsole so I just use locate because I know where all _my_ stuff is).

I probably need to try out GNOME, but funny timing since when I upgraded to FC3 most of the GNOME packages were uninstalled :-| (I had been trying to remove them a while ago to save disk space but had no clue what was required to keep GTK apps working). I don't really feel like installing GNOME in either Fedora or Gentoo... I'm probably going to install a GNOME based distro into VMware (or wait till Xen is supported better than just use that), except the last 2 distros I tried to install into VMware failed to install (SuSE and Ubuntu, my god theres so much spam for 'OMFG USE UBUNTU!!!!111!11oneoneoneone' on so many sites).

Wow, a post about a survey ended up talking about ubuntu... I'm scared now... :-(

by Rob (not verified)

I wish someone would do something about the usability of the survey first.

by Flavio (not verified)

1. when starting the survey it opens a popup window!
2. the form is full of randomly positioned radio buttons. Looks more like a Seurat painting than a web form.

by KDE user tasklist (not verified)

Their survey is not "readable" in 1152x864,
since we cannot see B.23 and the "submit" button
without putting the desktop in 1600x1200.

The window is NOT resizable on my machine.

While some people complain about the black on blue.
Blue is not a bad choice since it's a 'calm' color,
but the contrast could be 'enhanced'.
White on black would be fine too, easier to print out.

The survey only covers some of the common task,
but does not cover performing 'actual work'.

For instance, some of the following task needs LOTS of improvement too:

- Modify regional settings
- Modify network settings
- Modify keyboard settings or layout or switch mode
- Modify background
- Modify desktop theme
- Modify font settings
- Search for files and folders (including file content)
- Search for programs and Update KMenu with new entries
- Adjust display properties (move from 1024x768 to 1600x1200 from kicker)
- Adjust volume properties (Wave, CD, mixer)
- Adjust time/date
- Adjust firewall port settings (Deerfield)
- Adjust per application firewall settings (Zone Alarm)
- View recent programs
- View recent documents
- Clear recent programs/documents entry
- Clear browser history
- Clear offline content
- Automatically wipe out after logoff (recent programs, history, cookies, offline content)
- Create a shortcut
- Modify a shortcut
- Manage shortcuts
- Modify/Organize the KMenu
- Drag'n'drop KMenu items
- Automatic layout of desktop icons
- Organize desktop icons

- Browsing files/web pages (Konqui, Firefox, Mozilla),
- Email (KMail),
- Creating/manipulating pictures (KolourPaint, Krita, Gimp, Photoshop),
- Transferring files around (KIOslave nice, but not really usable/intuitive for newbies),
- Taking screenshots,
- Recording videos of running application (Flash/SVG?),
- Text editing (Word/Programming),
- Punching numbers (Excel/Calc),
- Making text document/presentations (PowerPoint),
- Playing music/videos (WinAmp, XMMS),
- Manipulating sound configuration (Volume control),
- Installing/upgrading software (Add/remove software),
- Installing/upgrading device drivers (Add new hardware),
- Burning a CD (K3B, Easy CD Creator)
- Compressing a document (with or without specific settings),
- Compiling a project (without typing ./configure && make && make install),
- Recording speaker sounds into a MP3/OGG format (Advance MP3 Sound Recorder),
- Creating a partition (Partition Magic),
- Instant messaging (MSN Messanger/Trillian/Gaim),
- Integrated webcam support,
- Playing simple desktop games (cards),
- Synchronizing with a repository (KPackage?, TortoiseCVS),
- Configuring hardware configuration (Device Manager),
- IRC clients (mIRC clone?),
- Debugging a project (in trace mode like Visual Studio - DDD/gdb sux),
- Sending a fax (BitWare, WinFAX),
- HTML/XML editor (CuteHTML),
- Database editors (phpMyAdmin, SQL explorer, MySQLGUI),
- Charmap,
- Creating schematics (PCB/Eagle),
- Simulating electronic circuits,
- Hexadecimal editor (HexEdit),
- Publishing/maintaining a website (FrontPage),
- Running an anti-virus (F-prot),
- Updating KDE (AutoPackage?),
- KDE apps automatically when updates are available,
- Configuring a FAX/Modem/DSL connection,
- Automatic Wireless network detection/connection,
- Creating CD labels,
- Creating sticky labels using the printer,
- Integrated PDF/PS viewer (KPDF?),
- Easy-to-use advanced configuration utilities,
- Authentication tools.
- GIF animation creator
- GIF/JPEG/PNG optimizer
- Creating 'installation wizard' (AutoPackage? Zero-Install?)
- Easy drag'n'drop GUI form creator (VB6 or C++ Builder like)
- Defragment the hard disk
- Scandisk
- FreeMem
- Clean wizard for /tmp and temporary files
- Clean wizard for removing unused apps.
- Money management (MS Money, Quicken)
- Map service (Mapquest, Google maps)
- Income tax software

Finally, playing Windows or 3D games (Wine/CrossOver, TransGaming),
accessing Windows resources (see Xandros, Novell/Suse, Samba).

That's mostly what Windows user do with their computers.
Other task could be run via QEMU like games or income tax software.

For instance, currently, we would need some custom apps from Novell,
mixed with some apps from Xandros (Xandros File Manager, Xandros Networks Application, http://www.linuxjournal.com/article/8161)
and some stuff from Mandrake or Linspire or other distro
to make a "possibly usable" desktop.

That makes no sense at all.


by Billy J. West Jr. (not verified)

Most of what you listed has nothing to do with KDE, but rather Linux, or a specific distribution. I hope you don't expect every application available to be ported over to KDE style.

1) please subscribe to the KDE Usability mailing list
2) choose a KDE application, back up your local settings and run it, write down all aspects you find unfriendly (even the smallest report is useful)
3) send that report to the usability mailing list
That's an easy start for all users. If you know more about usability, you can help improve the KDE HIG and be more deeply involved.

This is real contribution to the KDE project! This is real USABILITY work!

"1) please subscribe to the KDE Usability mailing list
2) choose a KDE application, back up your local settings and run it, write down all aspects you find unfriendly (even the smallest report is useful)
3) send that report to the usability mailing list
That's an easy start for all users. If you know more about usability, you can help improve the KDE HIG and be more deeply involved."

Thanks annma. After a long and fruitless discussion on that very same KDE usability mailing list, getting my head down, learning about KDE and coding is exactly what I'm going to do over the next three or four weeks. Trying to argue your case can often be totally fruitless.

If nobody wants usability thinking (as I'm getting the general impression of that - not from everyone mind you), everybody wants incomprehensible settings that non-programmers cannot use and nobody wants a default desktop that you can put infront of an ordinary non-programmer non-KDE experienced person (why KDE was started in the first place) then I'm just going to start small and do it myself.

With the right technology, which KDE has, and the right default presentation to an end user it could blow anything else out there away. We can also get distributors doing less customisation and using more default KDE technology, which is something I think should be aimed for.

As for the KDE HIG, I'm sceptical about most developers wanting anything to do with it but I'm willing to be hopeful.

by HASE (not verified)

We would like to thank everybody who took the survey and especially to the KDE project for posting a link to it in the Latest News section. We noted that, immediately after that, some individuals tried to mislead the others about the motives and the scientific basis of the survey. We prefer not to answer them because such discussions are not productive. The survey is being conducted by a research group at UMBC, HASE, and as noted before HASE has no vested interest in its research and findings. We think that any contribution is valuable.

The survey is still open. We would like to encourage you to take it if you haven't done so.


by Aaron J. Seigo (not verified)

while i respect your research efforts, i must say that i'm completely unclear as to your goals with this survey.

it certainly isn't the sort of study that can lead to _any_ useful usability related conclusions whatsoever. it will be able to tell us what a self-selected audience thinks about their satisfactions levels with KDE, but from a usability perspective that's not particularly useful data.

so assuming you haven't simply bungled this survey, i'd be interested in knowing what you really are trying to measure here or how you see it fitting into efforts to improve usability on the Open Source desktop.

by Fast_Rizwaan (not verified)

It is really very unsatisfactory to find my files, which are scattered at many places.

If I want to find my file which has a substring "ate"

1. I have to add 2 extra characters '*' before and after ate to get that thing done. KFind must by default consider the search string to be substring with '*' before and after the string.

2. The Interface of Kfind is somewhat not useful. like the Results are displayed in 'list' format only, which is not a good integration with Konqueror which has multicolumn, icon, detailed view.

3. The KFind window in Konqueror simply takes too much window space. A KFind Sidbar Search panel would be great like in Windows.

4. And there is no Default key assigned in Konqueror to find files (like CTRL+F) as in Windows. Yes, a user can assign a shortcut to that action but that's too much of a work for thousands of users. Plz assign default 'Ctrl+F' to find files in konqueror.

by aleXXX (not verified)

Known problem, it is being worked on.


by Leo Spalteholz (not verified)

I'm all for taking surveys, and I would have completed this one, had it been even reasonably easy to complete.
The format is hard on the eyes and the colours are bad, but what got me to quit, was that you either did not proofread your own survey, or have a poor understanding of how to conduct one. Question 3 is the exact same as Question 6 in the tasks survey. I understand that you might want to check consistancy, but having identical questions half a page apart is just going to confuse people, and make them copy down their previous answers.
If you can't be bothered to properly design a useful survey, I (and many others) can't be bothered to take it.

by annma (not verified)

I expect the DOT to publish the results from these 'surveys' within 6 months. Surely a high quality and detailed report will be freely available for KDE to make use of your findings based on scientific expertise.
Thanks in advance!

by Richard Dale (not verified)

"..your findings based on scientific expertise"

Reducing everything to numbers is not scientific. A scientific theory must be testable. In what way will the results from this survey be testable? I filled in the forms, but I haven't the faintest idea about the difference between 'I agree' and 'I strongly agree' over the issue of whether I use the right mouse button a lot.

Usability is more like art, someone with a good sense of aesthetics must make judgements on what is best.

by Aaron J. Seigo (not verified)

> Usability is more like art

so was science when it was just "natural philosophy". and then some people Got It(tm) and start up this whole "scientific principle" thing. most usability is testable and it is verifiable.

usability is not "art" (as opposed to "science"). esthetics can help create interfaces that are more approachable and enjoyable, but that's not what "usability" is about.

by Richard Dale (not verified)

I'm thinking of Karl Popper's definition of science. If a theory isn't testable, then it isn't a theory and it isn't scientific. That's roughly what Isaac Newton invented - ie the scientific method. In what way is usability a testable scientific theory? Some users might hate the right mouse button, and want to remove all the options from it. Other users, such as myself like using the right mouse button. How do you make the compromise between users with different requirements? How could you ever have a 'theory of right mouse buttons'?

The important thing is to have peer reviews about the options, so that the opinions of people who have been shown to have good judgement in the past will be given more say in any current discussions. For instance, people like annma and yourself. I don't think the results of this sort of survey should be given much weight.

On the other hand, I believe (unscientifically) that peer review is much more scientific than this sort of survey by numbers, even if there is no underlying testable theory.

by Aaron J. Seigo (not verified)

yes, from a usability perspective this survey is thoroughly useless and unscientific. that much we certainly agree on =)

but you most certainly can measure usability related issues, such as "how do people tend to group these specific options?" and "how many levels of depth can an individual manage to deal with?" and "what sort of context menus are easy for people to use and which aren't?"

there is no single answer to many of these questions that cover every user perfectly, but there are statistically significant answers especially when you target particular audiences.

i personally consider usability to be like architecture: there are certainly esthetic principles involved, but there is also a lot of hard science in it too. these scientific principle range from studies of human psychology ("how do people prefer to or naturally occupy a space?") to physics ("how strong do these beams need to be?" or "what are the aerodynamics of a building 80 stories high?") to matters of practicality ("how many bathrooms are necessary on each floor to ensure they are convenient enough that people aren't inconvenienced beyond a certain point?") ...

by Richard Dale (not verified)

Yes, I agree with you here, and think this survey is useless.

I did a degree in 'Philosophy with Cognitive Studies' 30 years ago, and have been fascinated with this sort of thing ever since.

by Richard Dale (not verified)

Thinking about this more, I think your comparison with usability and architecture is spot on.

The trouble is that we make artificial distictions between art, science and technology/engineering.

I'm not a painter, but from what I've read about how painters go about painting it does involve significant amounts of science/technology. For instance, Carravagio did very clever things with mirrors to implement his very realistic style of painting. Without state of the art technology for the time, he couldn't have done that. When a 'non-painter' just looks at a picture, they assume 'it just happened'. But a painter has to work hard thinking about the techniques needed to implement the painting to achieve that 'suspension of belief', and that mechanism must be carefully hidden, in order to communicate their message.

A good user interface is one you don't actually notice. When I'm editing in vi, I'm aware of it the whole time, and I hate the thing. On the other hand with an editor like Kate, all I'm thinking of is the program code I'm working on. I don't care if I need a few more keystrokes to do something, as long as I can be thinking about my program at the same time, instead of whether mistyping 'J' instead of 'j' is suddenly going to ruin my edit.

by annma (not verified)

hmm, are you challenging what I say or going in the same way? I don't follow you. For me a survey should be conducted with a scientific base, for example the fact that I can take it twice from the same machine is, hmm, non scientific. The fact that the user is not asked how many desktops he use/used is also non-scientific considering the first survey questions. The scientific expertise someone would expect is facts about people who took the survey. When you are polled in real life, you have to agree to be asked lately to certify you indeed were polled and in what conditions. Scientific did not apply to usability in general in my sentence but how from the questions usability facts are extracted.

And your usability view, well, implies that we can have different usability behaviors just as we have different forms of art. Not quite what we want to achieve in KDE.
Anyway, as I stressed, real usability work is done in KDE and that is what matters.