APR
9
2003

eWeek: Sometimes, More Is More

eWeek's Jason Brooks gives us his take on the discussion about the complexity of KDE and GNOME in his article "Sometimes, More Is More". "I didn't switch from Windows to Linux on my home and work systems because Windows wasn't easy. I switched because as I came into contact with OS alternatives, I became frustrated with lack of flexibility I found in Windows."
Do you agree? Are we (roughly) on the right course, or should we be trimming away options?

Comments

What works best depend on what kind of user you are. That may be obvious, but people fail to see the equally obvious conclusion that this means that less is more, and this is not a matter of aesthetic preferences.

The reason for this is that probably *more* than 90 % of the users only need a certain level of functionality. Having more options than needed is counterproductive because largely irrelevant information competes with the relevant one.

So, KDE goes deliberately goes for too much compexity the also aim for irrelevancy, either by targeting a fraction of the user base by choice, or by objectively making a less usable product for the larger fraction of users.


By will at Wed, 2003/04/09 - 5:00am

> The reason for this is that probably *more* than 90 % of the users only need a certain
> level of functionality. Having more options than needed is counterproductive because
> largely irrelevant information competes with the relevant one.

This is just plain wrong.

When we talk about choosing defaults wisely, I agree wholheartly. Choose what the majority of users like best.

But when we talk about features that can be turned on by configuration, the more the better. I have yet to see a real user complaining that there is too much configuration for himself.

Also, always look at actual users, not some hypothetical "average" user that is invented by self-proclaimed usability experts.

GNOME2 made exactly that mistake and lost a lot of real, existing users by trying to attract some hypothetical non-existant "average" user.

Please, KDE-team, don't make the same mistake. KDE is the best environment because it is the most complex and the most configurable of them all. KDE is clearly more successful because of that. And the ***ONLY*** reason KDE/Linux is less popular than Windows is because KDE/Linux can't run Win32 programs. Usability has absolutely nothing to do with it.


By Roland at Wed, 2003/04/09 - 5:00am

>But when we talk about features that can be turned on by configuration, the >more the better. I have yet to see a real user complaining that there is too >much configuration for himself.

A user is rarly aware of why an interface is difficult to use, but that doesn't mean there isn't a reason why. Relevance and information economy are very plausible explanations.

>Also, always look at actual users, not some hypothetical "average" user that is >invented by self-proclaimed usability experts.

>GNOME2 made exactly that mistake and lost a lot of real, existing users by >trying to attract some hypothetical non-existant "average" user.

The average users are not hypothetical. Just look out the window and you will see that they are real: office workers, students, state employees, clerks, business people. I believe the "real" users you are talking about here belong to less than 0.1% of the popuplation.

>Please, KDE-team, don't make the same mistake. KDE is the best environment >because it is the most complex and the most configurable of them all. KDE is >clearly more successful because of that. And the ***ONLY*** reason KDE/Linux is >less popular than Windows is because KDE/Linux can't run Win32 programs. >Usability has absolutely nothing to do with it.

But do you think that KDE/linux has a chance at the larger user group if it is less usable than Windows? Windows was once small too - how do you think they got to be big?


By will at Wed, 2003/04/09 - 5:00am

It got big long before it got usable, ask the Mac guys if usability is enough to breed success!

Rich.


By Richard Moore at Wed, 2003/04/09 - 5:00am

I believe the point is that usability is a necessary condition for success, not a sufficient condition. In any case, it certainly did make Mac a success, just not as big as Microsoft, which had the advantage of the more open PC platform.


By will at Wed, 2003/04/09 - 5:00am

"A user is rarly aware of why an interface is difficult to use, but that doesn't mean there isn't a reason why. Relevance and information economy are very plausible explanations."

So you don't have proof - you don't even have a hint that your hypothesis is correct.

So the KDE-team should fuck their userbase just because you say so? Without the slightest proof? Just because of a hypothesis?

"The average users are not hypothetical. Just look out the window and you will see that they are real: office workers, students, state employees, clerks, business people. I believe the "real" users you are talking about here belong to less than 0.1% of the popuplation."

Complete nonsense.

There is no such thing as an "average user". I'm a programmer and I'm probably classified as "advanced", yet I think many office workers will need a lot more features in wordprocessing than me. Now you come along and say: Drop all features only a minority uses. So you piss me off because I can no longer have multiple desktops, etc. and you piss the office workers off because they don't have advanced wordprocessing features anymore. You want to piss EVERYONE off. Students, state employees, clerks, business people - all will have different needs and preferences, everyone will be pissed when forced to use your dumbed down "dream come true" desktop.

There is no average user.

>>>>EVERY USER IS DIFFERENT.<<<<

Get it in your head already.

The "average user", the self-proclaimed usability experts talk about is as real as unicorns, elfs and orcs.

"But do you think that KDE/linux has a chance at the larger user group if it is less usable than Windows?"

First it isn't less usable and second it wouldn't matter because people will use whatever comes preinstalled on their computer.

Sure, KDE/Linux has rough edges - for example drive mounting is awkard - but it's comparable to Windows' rough edges (See a real newbie wondering why he has to double click some icons and single-click other icons to see one of the biggest usability problems of Windows. Another example is drive letters.). SuSE has already put out a distribution with all configuration stuff integrated into kcontrol which solves most of the inconsistency problems some people complain about.

The only major problem in KDE I see is the lack of documentation.

" Windows was once small too - how do you think they got to be big?"

1. They were backwards compatible to DOS.
2. Windows was preinstalled on pretty much all computers.

If KDE/Linux would run Win32 apps reliably and hassle-free, we would see millions of users along with computer makers switching. Without that backwards compatibility, the transition will take much, much longer. (But it is happening anyway.)

Actually, Windows was very, very late in offering a GUI. Apple, Amiga and Unix had GUIs much longer - But neither Apple, Amigo nor Unix could run DOS programs on cheap x86-hardware, which was ten times as important as anything Apple, Amiga or Unix offered on the desktop.


By Roland at Wed, 2003/04/09 - 5:00am

Much of what you say can only be settled by empirical investingation, of which I am sure there is exist a good deal, but which I cannot give without doing further research. I am certainly sure that such findings would support my claims, and if they didn't I would certainly retract them.

Problem is: you don't quote any empirical evidence for the contrary claims, so it doesn't matter in the present discussion - and we are back where we are right now: what can reasonably explain good usability. I claim that increasing relative visibilty of the most relevant options (statistically frequent usage) will explain good usability, and you offer no competing explanation. Clearly the burden of evidence is on your side. You would be ill advised to make a choice made on the basis of ignorance.

Concerning the "there is no average user" claim: This is a possible construal - no user has an average use of program functions which in this regard places him in a majority group. I certainly believe you are wrong about that - but only an empirical investigation can settle the question.

You are right about the other factors you mention concerning the success of the Windows platform and the barriers KDE is facing beside usability, but it doesn't concern my point: Good usability is normally a necessary condition of success.


By will at Wed, 2003/04/09 - 5:00am

"Much of what you say can only be settled by empirical investingation, of which I am sure there is exist a good deal, but which I cannot give without doing further research. I am certainly sure that such findings would support my claims, and if they didn't I would certainly retract them."

Being scientific is only possible when you are precise first.

You haven't made any precise claims here (just the usual "omg, KDE is sooo complicated"), so there are no claims which can be supported, so far.

"Problem is: you don't quote any empirical evidence for the contrary claims, so it doesn't matter in the present discussion - and we are back where we are right now"

Wrong. The fact that KDE is more successful than GNOME is proven by loads of polls, studies and articles. Of course this isn't scientifical unbeatable evidence (it could be another reason that KDE is more successful), but it still is a *very strong* hint that configurability and flexibility is rewarded by users.

A comparison with Windows is meaningless because the fact that Windows is preinstalled and runs a much larger software library makes any other factors unmeasurable. KDE and GNOME, however, are comparable because they run the same programs and are distributed among the same channels.

Actually, GNOME is quite in favor: Gimp is GTK-based and pretty much the only serious image manipulation program on Linux, also a lot of users like Galeon and Evolution.

It seems that KDE is so much better than GNOME that a lot of users prefer running a slightly inconsistent KDE with GNOME apps to running pure GNOME.

Sorry, but I don't see a single reason why KDE should parrot GNOME, can you?

"Concerning the "there is no average user" claim: This is a possible construal - no user has an average use of program functions which in this regard places him in a majority group. I certainly believe you are wrong about that - but only an empirical investigation can settle the question."

Such an "average" can only be chosen, not measured - which is by definition unscientific, so NO, there cannot be a scientific investigation about this topic.

Actually it's pretty easy to grasp that such an "empirical investigation" is doomed to fail.

If you have 4 users, one needs full wordprocessing features including headers and footers, another wants all the desktop features (multiple desktops, etc.), a third wants good command line and scripting support and the last wants advanced browser features like tabbed browsing. - Where is the average?

Are all features that are not used by the majority non-average (This is the impression I get from your talk, although it's all very vague and confused)?

In that case everything would have to be dropped: 3 out of 4 users don't need headers and footers in wordprocessing = bloat, dump it. 3 out of 4 users don't want multiple desktops = too confusing, get rid of it. 3 out of 4 users don't use scripting = not needed. 3 out of 4 users don't need tabbed browsing = Away with it.

Great, now you have 4 unhappy users out of 4.

Is that really your goal?

So what empirical studies do you want? When is a given feature worth it? What percentage has to use it?

"You are right about the other factors you mention concerning the success of the Windows platform and the barriers KDE is facing beside usability, but it doesn't concern my point: Good usability is normally a necessary condition of success."

Wrong, again. Amiga had great usability and failed, MacOS had great usability in the past and lost marketshare all the time - and DOS with all IRQ conflicts and DMA problems got stronger and stronger.

I sure agree that usability is important. But a prerequesite for success? Far from it. Usability is a "nice to have" feature, no killerfeature. If my needed program doesn't run on KDE, I will not run KDE, no matter how usable it is. Only on a leveled playing field (for example KDE versus GNOME, where all users can run all programs and the choice between KDE and GNOME is truely by preference and not forced by compatibility) usability plays a major role. And KDE is much more successful than GNOME.

Again, I provided several examples while all you did was put out wild unproven claims without a single link to reality.


By Roland at Wed, 2003/04/09 - 5:00am

>> "A user is rarly aware of why an interface is difficult to use, but that
>> doesn't mean there isn't a reason why. Relevance and information economy are
>> very plausible explanations."
>
> So you don't have proof - you don't even have a hint that your hypothesis is
> correct.

simple example 1: (Information economy)
A certain item can be more easily found if the number of 'standard' *available* items from which you can pick is *decreased* and the wanted item is among the items that is standard shown.

Problem: which items should standard be shown?

simple example 2: (Relevance)
A certain item can be more easily found if is shown *higher up* in the list of 'standard' available items from which you can pick and the wanted item is among the items that is standard shown.

Problem: which standard item is the most important?

simple example 3:
Exposing a user to an interface that is known to the user eases the transition from that system.

problem: How far should you go in adapting your interface to the system known to be the one that is most used worldwide, but which is flawed -- according to your opinion.

These are challenges that KDE currently faces and where there has to be made a choice, since it is easy to grasp that the interface usability will degrade if you shown zillions of options.

I could go on for ages with examples like these, but there is nothing new here. Go read in any book about usability and discover them yourself. Go play with Gnome themes and see what's great about the HIG of Gnome.

BTW I use both Gnome and KDE (Mandrake 9.1 right now) and see that they both have advantages:

- Gnome 2.2 is notably faster than KDE 3.1 when starting applications specific to that desktop. (p.e. Nautilus is faster than Konqueror, but it has less features at the moment).

- Both desktops are very easy to use once you have spend a certain amount of time on each desktop.

- KDE has the better experience for users transitioning from Windows (any version).


By Richard Bollinger at Fri, 2003/04/11 - 5:00am

"average" is a very dangerous concept in this sort of context. For example, let's say there are two people in the world; Alice and Bob. We want to design the shoes that are the best fit. Alice has size 5 feet, and Bob has size 12 feet. The average shoe will fit neither.

Despite there being a large number of desktop users the "average" does not necessarilly get any more accurate. Especially as the possible number of configurations is fairly enormous as well. Again to illustrate - let's say there are two types of desktop user - those who like 20 options per widget and those who like 2 options per widget. You end up pleasing nobody if you put 10 options per widget.

I also don't agree that options confuse new users. I've put KDE + Linux on my girlfriends computer. She isn't interested in the thousand options available to her in KDE, so much so that she hasn't even gone looking. They aren't confusing her, but the day she wants something different she can have it. As long as options aren't "in your face" there is no problem for newbies.


By Andy Parkins at Thu, 2003/04/10 - 5:00am

I don't think you're entirely right. I am a technical user (software engineer, use Linux + KDE 7 days a week 365 days a year) but I can never find anything in the KDE control center. ALmost every time I need to change an option I have to hunt through most of the options trying to find the setting I'm after.

Having done informal tech support for friends who are learning Linux, KDE usability does need improving - some things just don't work the way new users expect and some things take more effort than they should.

Just because (you think that) Gnome have screwed up, don't assume that there usability experts have nothing to offer, or that something you can use because you're familiar with it cannot be improved.


By cbcbcb at Wed, 2003/04/09 - 5:00am

Well, OK you didn't find something in kcontrol right away. So what time did it cost you to find that option? 30 seconds? Maybe 2 minutes?

If we listen to people like "will" there will be only one default and you wouldn't lose any time searching for configuration option because there would be none.

Actually this was all my point. There is no default perfect for everyone.

So we have essentially 2 choices: Force a default down user's throat or offer a default and let users change the behaviour in kcontrol.

I don't know about you, but a hard to find configuration option is far better than no such option.

"Just because (you think that) Gnome have screwed up, don't assume that there usability experts have nothing to offer, or that something you can use because you're familiar with it cannot be improved."

Wait, I never said that.

The only thing I said was that removing configuration options on purpose is making things worse, not better.

I don't see how you can see that KDE is already perfect or that nothing can be improved in that statement.

There are many things in which KDE can improve, for example tabbed browsing still leaves a lot to be desired in KDE3.1 or KDE still doesn't support 4 mouse buttons ( http://bugs.kde.org/show_bug.cgi?id=48062 ) bookmarklets would be great ( http://bugs.kde.org/show_bug.cgi?id=30302 ) and other things - But destroying configuration options will accomplish nothing.


By Roland at Wed, 2003/04/09 - 5:00am

Bookmarklets are largely there, thanks Alex Kellett. See:
http://lists.kde.org/?l=kde-cvs&m=104881296429430&w=2


By SadEagle at Wed, 2003/04/09 - 5:00am

"Well, OK you didn't find something in kcontrol right away. So what time did it cost you to find that option? 30 seconds? Maybe 2 minutes?"

This is a perfect example of why people like you are full of it. You told will that you'd never heard a user complain about there being too many options, and then as soon as some user complains you tell him he's wrong to complain.

Simplicity is beautiful. If you went and grabbed a thousand computer users off the street and asked them what they want from a computer, not a single one of them would say, "I want my interface to be more complicated. What's really missing from computers these days is complexity! Expose more of the complexity to me!" What alternate universe are you living in where complexity is justified by user demand?


By guest at Thu, 2003/04/10 - 5:00am

Personaly I think 99% of al users will fawor spending 2 minutes shearching for a way to do what they want, commpared to spending the same amount of time before beeing told they can't do it.


By Morty at Thu, 2003/04/10 - 5:00am

Most stuff are only configured once. So 2 minutes lost to gain better usability because you can make the system behave the way you like *every day* is a tradeoff I think most sane people will take.

Now there might be better ways to arrange the configuration settings in kcontrol. - But removing opions for the sake of "simplicity" is just not worth it.


By Roland at Thu, 2003/04/10 - 5:00am

Did you not notice the second tab? It's not like there are more than 3 ;-)


By SadEagle at Wed, 2003/04/09 - 5:00am

This is crucial. Glad to see I'm not alone. There are so many entries and things are so randonly strewn around that there is VAST ROOM FOR IMPROVEMENT. Make it a usability engineering study. Do some research post the findings in a group and hash it out ... then decide and make the perfect default (offering several config options to cange it of course ....). This is the next step in OSS. Getting HID and UI students and pros to contribute.

They have been trimmed but they need more trimming and organization. Just as XP seems to have improved the config interface heirarchy and simplie]fied the start menu (but then they wrecked the whole menu by making it way too huge ... when the normal menu size is set it's OK - the only other problem is that it is called "start" ... hehe).


By KDE Fanatic at Thu, 2003/04/10 - 5:00am

I have trouble finding things in the Windows control panel.

I used to use GNOME, but GNOME 2 just changed everything I liked about GNOME. I no longer had a good customizable UI, but one that was fairly concrete and not too my liking. I thought usability had gone down quite a bit. I understand the GNOME Project's stance on customizability, but they are alienating their users.

KDE is good, and I really like it. It has its flaws, but I consider it superior to every other desktop or window manager I've ever used (with the exception of that hacked thing I put together, but thats just personal pride).

Now, finding some options in KDE is difficult, but we aren't talking about a small program here. We are discussing a massive project with millions of lines of code and thousands of features.

And about KDE not working the way new users expect: Its different. Deal with it.


By crichards at Sat, 2003/04/12 - 5:00am

I think that it would be MUCH better to actually discuss what could be done to improve the average usability in both KDE and Gnome. I want to have full controll over my desktop, but not all... So what could we do here? Make basic options available and then add all maybe more advanced options under some advanced button? I don't think that's a good solution...


By Dancing with th... at Wed, 2003/04/09 - 5:00am

"I have yet to see a real user complaining that there is too much configuration for himself."

Well, congratulations, today you'll meet your first one: I am a real user of KDE nearly every day for the past 4+ years and I am annoyed that there are too many options most of the time. Open up Konqueror and right-click on a file. Too many options. Open up "Settings" -> "Configure Konqueror". Too many options.

Furthermore, your formulation makes the simplistic assumption that the costs of having many options are readily apparent to users. Every option exacts a cost in terms of speed, memory usage, usability, and developer effort. If a user were presented with an accurate description of the cost/benefit ratio of options vs. improvements in app startup time, stability, and progress on important features, more users might very well choose fewer features. The fact is that most users---including, it seems, you---are ignorant of this cost/benefit tradeoff, which is why they never complain about having too many options: the costs are hidden. Would I rather have the authors of KOffice getting WYSIWYG printing right, or dithering around with yet another checkbox in the configuration dialogs? Would I rather have yet another window deco or a desktop that actually works? Duh. But nobody thinks about it that way.


By guest at Thu, 2003/04/10 - 5:00am

I agree with what you are saying, as a user of GNOME2 since it started being usable in CVS. So many options were taken out that it became much harder to configure it to what I wanted. There were a large amount of complaints about this, yet most developers merely stated that "average" users would just be confused by more options.

In my opinion, we do not need two similar desktop environments. GNOME is working very hard to have a lack of configurability, and should work well for the seemingly non-existant "average user." Thus, I think KDE should work toward a very customizable system. In this way, there would be no question as to whether you should use KDE or GNOME - if you want to configure things, you would use KDE, and if you are confused by too many configuration options (i.e. if you are an "average user"), then you can use GNOME.

Please don't go the same route as GNOME. GNOME 1.4 was my favorite system, far better than KDE at the time, in my opinion. But the lack of configurability in GNOME has meant that I have been forced to switch from GNOME2 to the far more configurable KDE. Still, KDE 3.1 doesn't seem to be quite as configurable as GNOME 1.4, but it is significantly better than GNOME 2.

C. Evans


By C. Evans at Fri, 2003/04/11 - 5:00am

> Usability has absolutely nothing to do with it

Good comment.
We must understand that optimum usability/productivity can only be achieved
for a specifically targeted user-group. It is not possible to target all
office-workers exisiting on earth and also achieve optimum usability.
Though, this may be wise economically though... :-(

So the bottom line is always that KDE will have to be taylored
for the user-group in question.


By KDE User at Sun, 2003/04/20 - 5:00am

I've heard the argument many times before - and I'm afraid I can't quite see the point. I think at this time in the desktop user interface history, the "less is more" argument is similar to the idea of premature optimizations in code. By doing less or having fewer features we are trying to optimize a system that isn't fully defined or completed. Who knows which feature will be the most important -- that in the long run will be invaluable.

Until Linux, Gnu, and Gnome or KDE are on 25% of the worlds desktops and there is substantial feedback from the non-tech population - we can only speculate which of the "extra" features make the system less elegant or useful.


By Paul Seamons at Wed, 2003/04/09 - 5:00am

>I've heard the argument many times before - and I'm afraid I can't quite see the >point. I think at this time in the desktop user interface history, the "less is >more" argument is similar to the idea of premature optimizations in code. By >doing less or having fewer features we are trying to optimize a system that isn't >fully defined or completed. Who knows which feature will be the most important -- >that in the long run will be invaluable.

>Until Linux, Gnu, and Gnome or KDE are on 25% of the worlds desktops and there is >substantial feedback from the non-tech population - we can only speculate which >of the "extra" features make the system less elegant or useful.

This an important insight - the right way for OS-developed software to develop user-friendliness is by concrete user feed-back. However, you would be wrong if you attempt to conclude that KDE can ignore information economy now, because it makes for a less usable product, which in turn would make it less likely KDE to reach 25% of the users. The only thing anyone can do is to make as good decisions as can be, and which those are, is presently under discussion - which consequently do have a point.


By will at Wed, 2003/04/09 - 5:00am

I wonder why this argument allways comes up -- Why should the majority of those who have already realized the potential of a non MS desktop wait for those who needs a office binder to tell them what to think, to judge what is an elegant and useful desktop?

The less is more idea is more of an individual preference, I think. You are still quite likely to find emacs and vi users having this discussion.
My view on this subject is that you can have lots of stuff if it's arranged in a way that you will allways know where to look for what.
A nightmare, in my oppinion, is when you have a lot of programs installed in Windows and they are organized by who made the program, not by the function of the program as in the K-Menu. ie. who is interesting in knowing that Ahead made Nero, and how obvious is it that Nero is a program for burning CD-ROMS compared to a program which is named arson :)

There is absolutely nothing I miss from Windows after I've logged into my wonderful KDE. The ability to run Windows programs is not something I will wait for, OS/2 had this feature and I guess that was what made it just a half.

I prefer a desktop that is made by and for intelligent people :)


By Luguber at Wed, 2003/04/09 - 5:00am

>I wonder why this argument allways comes up -- Why should the majority of those >who have already realized the potential of a non MS desktop wait for those who >needs a office binder to tell them what to think, to judge what is an elegant and >useful desktop?

The reason is that many people want a free alternative to Microsoft which is equally usable. You are certainly entitled to your opinion that KDE should be for a small minority of users, but IMO you should stop to wonder why people disagree with you about that.


By will at Wed, 2003/04/09 - 5:00am

I don't think that KDE should be only for the rich, intelligent, famous or attractive minority or any other minority for that matter. What I'm simply implying is that if so many users already think KDE is superior in usability and flexibility etc. Why should we wait until Bill Gates comes to tell this? I've reinstalled the computers of most of my family members and quite a few friends with Linux/KDE. So far not a single one of those have complained that they wanted their Windows back. I even gave a computer to my wife with Linux/KDE on it, she has virtually never used a computer before and the first time she had problems and called me was when she visited a friend with this strange Windows installed, that she had no idea how to use :)
If you think that the majority stays with Windows because it's easy to use I think that you are really wrong. Most of these users stays with Windows because Linux-people keep telling them that Linux is more difficult, which I think during these two/three last years have been proven quite wrong.


By Luguber at Wed, 2003/04/09 - 5:00am

Right on!

I personally love to customize each individual part of my workspace. With Windows, I had to edit the registry to use/change some of the "hidden" features. Pain in the arse if you ask me. With kde, the only options that seem to be kept hidden are some of the developmental features where it hasn't been implemented for kcm. In fact, I was rather peeved when kde removed a feature from the background select when they changed the background kcm module (using custom monochrome patterns other than TigerT's Nightrock and Stonewall 2).


By Brendan Orr at Wed, 2003/04/09 - 5:00am

I'm going to take the middle view, guaranteeing that I will be equally despised by both sides. Oh well.

Less is not always more. Oftentimes less is just less. A desktop, not even considering the underlying OS, is a very complicated system. Removing configuration options does not make the complex simple, but merely removes elements control over the system. For users that do not want this control that is not a big deal, and may even be lauded. But for users that want that control, a less-is-more interface becomes a pain to use.

The opposite is just as bad. When you can choose (as a hypothetical situation) ten different "mouseover" looks in a theme's settings page, then you have gone too far. When 80% of the options are used by only 20% of the users, then it's overkill to include all 100% of them in a dialog.

In my opinion, GNOME hasn't gone so far in their "less-is-more" campaign to make an unusable "usable desktop". And KDE hasn't gone so far in their "give-me-options" philosophy to make the desktop confusing. In comparison to a microwave oven GNOME is complex. In comparison to a nuclear reactor control panel, KDE is simplicity.

We have a competent usability team at my work, and I've learned some few things from them, though I am hardly a usability expert myself. One thing I learned is that lab analysis of usability is generally useless. You need to see real users in real situations using the system in real life. Bringing in people to sit in a lab and follow scripts is next to useless. Another thing I learned is that logic and gut instinct are irrelevant in this area. Sometimes the system with the more "complex" interface is easier to use, while that with the simpler interface pleases only the marketroid. The third thing I learned is the most important: every user is different. Our customer base is probably the most homogenous you can get, yet every time we've tried to cater to the "average" user we've been bitten in the butt.

Make the desktop easy for the new user, but only to the point that you do not take anything away from the experienced user.


By David Johnson at Wed, 2003/04/09 - 5:00am

I despise you! :-)

Seriously, the problem of walking the middle way is that the message becomes less relevant as an means to rectify shortcomings...

I think you are absolutely right that usability choices should be made on the basis of actual user observation and not speculation, I don't intend to advocate that (much as I hope the pro-choice people wouldn't either). I would be happy to let abstract principles go and leave it all in the hands of competent usability research for the specific program. All that matters is that program becomes as productive as it can for as many as possible. (Clearly a part of the problem targeting a user group).

I am curious: As a sidenote, I have heard that in usability studies, usability *shortcomings* in software is often quickly recognized because they the users generally experience the same problems. What does your usability team say about that?


By will at Wed, 2003/04/09 - 5:00am

I think that what KDE needs is not a reduction of options, but a reduction in the visibility of options. KDE can have its feature cake and eat its usability too, if it adopts a configuration-editor sort of program. Something like about:config in Mozilla or the GConf editor. All the options in the world make no difference in usability in an interface like this, because when you use the configuration editor you are most often looking for a specific option that you heard about and got instructions on how to change (like a website that said "here's how to get focus-strictly-follows-mouse in KDE" or a guy who told you "oh yeah, you can set the font used in the KMail message display with foo-bar-font-option in KConf"). Then uncommon options could be removed from dialogs and still be accessible to a user motivated enough to find them.


By not me at Thu, 2003/04/10 - 5:00am

A more "usable" solution is to do what has already been started: place "advanced" options under an advanced button in the dialog. This is simple and elegant. The newbie need never see confusing options like "show tooltips" or "font size", but the intermediate and advanced users can still access them.


By David Johnson at Thu, 2003/04/10 - 5:00am

> What works best depend on what kind of user you are.

Exactly.

> That may be obvious, but people fail to see the equally obvious conclusion that this means that less is more, and this is not a matter of aesthetic preferences.

People fail to see this "equally obvious conclusion" because it's the obvious wrong conclusion. On the one hand you proclaim that it depends on the kind of user what works best, on the other hand you say reducing flexibility is the solution and you yourself know already which reduction will hurt the fewest users. That's contradiction at its best.

> The reason for this is that probably *more* than 90 % of the users only need a certain level of functionality.

You forget to note that while indeed more than 90 % of the users only need a certain level of functionality they all need a completely different set of functionalities. Implying that your "more than 90 % of the users" all use the same set of functionalities and reducing functionality based on that set is simply hogwash.

> Having more options than needed is counterproductive because largely irrelevant information competes with the relevant one.

Indeed, so we'll need a GUI with brain-reading capability so only the relevant information is shown at a time. Reducing functionality for everyone on the other hand is definitively the wrong way.


By Datschge at Thu, 2003/04/10 - 5:00am

A big debate these days seems to be focused on how configurable the Linux desktop should be. KDE has always taken the approach that users will have different preferences on how they like to work so the UI should be as flexible and configurable as possible. Gnome 2 has taken the direction that "less-is-more" and that the configurability in Linux desktops, including Gnome 1.x, was clutter and confusing to the end-user. This has resulted in some pundits calling for KDE to remove some of it's configurability.

The thing all people who argue for less configurability in Linux desktops have in common, whether they are Sun usability teams or Linux editorial pundits, is that they all presume to know how you should work and what a GUI should look like. If you believe your way of doing things is the "one true way" then making things configurable is a waste of time and space. Some pundits in particular have insisted that KDE developers should listen to them, do things their way, then remove configuration options to "clean up clutter in the UI".

KDE does not make such presumptions. Some people prefer single-click, others can't work without double-clicking. Some Mac users prefer window close buttons on the left, Windows users expect it to be on the right. Some people like icons on their desktop for devices, others don't. Some old school Unix people hate to work without window focus following the mouse, but this would confuse the hell out of non-Unix people. Don't even get me started on button order. Some KDE people have already started talking about making it configurable in dialogs - probably much to the dismay of those who believe they know "the one true way" despite that different users have expressed opposite opinions.

While there are those who have argued that having the extensive configurability of KDE is "clutter" and "tries to include everything but the kitchen sink", I view it as flexibility and trying to satisfy the user no matter what his or her tastes are. It is not up to a Sun usability expert or some Linux columnist what the best user interface is. It is up to the user. KDE tries to make the desktop as enjoyable and usable to as many people as possible. Doing that requires being configurable and flexible. The only people who like less configuration options are those who believe their way of doing thing is "the right way". Everyone else is, well, screwed.

Another argument is that configurability equals bloat. This is simply not the case with KDE. On the performance front due to constant optimization of code KDE has managed to actually improve performance while increasing the amount of things users can configure to make KDE match how they work. That's a good deal.

On a more aesthetic front some people would argue configuration options make user interface bloat and are confusing to end-users. This argument doesn't really hold up. You can productively use KDE without ever knowing about how you can configure, for example, menu drop shadows or window decorations. Many people who probably will never know what a "Desktop Border Snap Zone" option is have been enjoying KDE just fine. But I can guarantee the people that do know what it means and use it are sure glad it's there! KDE doesn't *force* you to configure anything. It's just there if you need it. You don't see all the window decoration or panel options unless if you explicitly go looking for them. If that is the case you should get as many options as possible in order to make them most closely match how you want your desktop. Otherwise you can use KDE and be blissfully ignorant of their existence.

As a matter of fact, many people are now setting up KDE for things like their families and rarely do they complain about things like too many configuration options. This is a complaint that you rarely see from end-users. I haven't seen many emails saying, "I want less options and features!". It is something "usability experts" and pundits say. People who believe they know the "one true way" user interfaces should behave... Again, if this is how you feel then configurability is a waste.

There are some things KDE has done to make things simpler. KPersonalizer - the wizard that runs when a user first starts KDE - allows users to configure things in a general manner, (such as do I want things to generally operate like a Mac or like Windows), without mucking around with a lot of configuration values. Many KDE Control Center modules have also started moving less frequently used options to an "Advanced Settings" window. But all the settings are still there for you to muck around with to your heart's content, and should remain there. KDE is about making the Linux desktop operate how *you* want it to work, not how anyone else feels you should work :)


By The debate abou... at Wed, 2003/04/09 - 5:00am

>>> The thing all people who argue for less configurability in Linux desktops have in common, whether they are Sun usability teams or Linux editorial pundits, is that they all presume to know how you should work and what a GUI should look like. If you believe your way of doing things is the "one true way" then making things configurable is a waste of time and space. Some pundits in particular have insisted that KDE developers should listen to them, do things their way, then remove configuration options to "clean up clutter in the UI". <<<

I couldn't agree more.

I just want to add that these pundits are actually contradicting themselves. If you have found "the one true way" and have set the default accordingly (so far I agree - let's find the best default which is so good that a minimum of users want to change it), there can't be any configuration "clutter" because if everybody was happy with the default, nobody would try to configure it differently and nobody would ever see the "clutter" in the configuration panels - so it would be irrelevant.

The pundits want to eliminate configurability, because they KNOW that not everybody will be satisfied with the "one true" default and they want to FORCE those to use them anyway.

>>>As a matter of fact, many people are now setting up KDE for things like their families and rarely do they complain about things like too many configuration options. This is a complaint that you rarely see from end-users. I haven't seen many emails saying, "I want less options and features!". It is something "usability experts" and pundits say.<<<

This sums it up perfectly.


By Roland at Wed, 2003/04/09 - 5:00am

"I just want to add that these pundits are actually contradicting themselves. If you have found "the one true way" and have set the default accordingly (so far I agree - let's find the best default which is so good that a minimum of users want to change it), there can't be any configuration "clutter" because if everybody was happy with the default, nobody would try to configure it differently and nobody would ever see the "clutter" in the configuration panels - so it would be irrelevant."

This is a ridiculous straw man. Nobody's saying that one configuration is best for everyone. They are suggesting some degree of moderation. Do you think KDE would be easier to use if it had a thousand options in the control center? How about a million? A trillion? The idea that "more configurability == better" in all cases is patently absurd. Clearly there's some limit. The "less is more" people simply disagree with the "more is more" people about where that line should be drawn.

Furthermore, your description of the "less is more" line of argument is wrong. The people who say the number of configuration options should be decreased are saying that there are some configuration options that are often changed (e.g., desktop wallpaper---everyone I know changes their desktop wallpaper) and others that are almost never changed (e.g., Autostart path), and it's a sound usability engineering decision to get rid of the latter in order to make the former easier to find and change for most users.

If you did a usability study---in the lab, or in the real world---you would find that Mac OS classic is far more usable for far more people than KDE. KDE is objectively harder to use, and one reason is that there's too many toggles and switches.


By guest at Thu, 2003/04/10 - 5:00am

"Furthermore, your description of the "less is more" line of argument is wrong. The people who say the number of configuration options should be decreased are saying that there are some configuration options that are often changed (e.g., desktop wallpaper---everyone I know changes their desktop wallpaper) and others that are almost never changed (e.g., Autostart path), and it's a sound usability engineering decision to get rid of the latter in order to make the former easier to find and change for most users."

What ridiculous nonsense.

Did it ever occour to you that it's possible to make the often used configuration options available easily without "getting rid of" the other options?

For example you can configure the wallpaper by right-clicking on the desktop. So you say that removing the autostart path configuration will make that more accessable?

If you *still* think that destroying the autostart path configuration will make configuring the desktop wallpaper easier, well, I guess you were either doing a satire or trolling.

"If you did a usability study---in the lab, or in the real world---you would find that Mac OS classic is far more usable for far more people than KDE. KDE is objectively harder to use, and one reason is that there's too many toggles and switches."

I own a Powerbook and have tested both MacOS9 and MacOSX. If you define usability by the first impression after let's say 1 hour, you are right. MacOS *looks* easy, looks great and the animations are cute when you first use it.
But in day-to-day work, it's useless compared to KDE/Linux. I want to get work done, not look at animations. KDE let's me do that, I can have many programs open simultaneously, because I have multiple desktops. - On MacOS or Windows I would no longer be able to keep track of more than say 20 windows. In KDE I can do a lot with the mouse, Unix-style copy-paste even let's me edit text fast and efficient. Also, MacOS suffers from single/double click inconsistency like Windows (actually the only reason Windows has such a moronic behaviour is because they copied it from MacOS).

But OK, maybe some people prefer such things. If you choose "Windows-style" or "MacOS-style" in the personalizer, you have a pretty similar environment.


By Roland at Thu, 2003/04/10 - 5:00am

>If you did a usability study---in the lab, or in the real world---you would find that Mac OS classic is far more usable...

Each time I've used the mac it has been an exercise in frustration. The top menu changes all the time. You are forced to move the mouse too far, wearing out mouse pads. It crashes, although with a cute bomb (os classic).

The Mac enforces a way, which some like and some don't. They do some things very well, that put others to shame, ie. application installation, hardware detection, etc.

Personally, having used os/2, windows, mac, *nix with kde is like coming home.

So is some usability expert to tell me I'm wrong?

Derek


By Derek Kite at Thu, 2003/04/10 - 5:00am

>>While there are those who have argued that having the extensive configurability of KDE is "clutter" and "tries to include everything but the kitchen sink", I view it as flexibility and trying to satisfy the user no matter what his or her tastes are.<<

As a matter of fact let me just point out that KDE is in the process of integrating the kitchen sink, or actually the KitchenSync. You will find it mentioned at:

http://pim.kde.org/development/meetings/20030103/summary_report.php

Sorry... I just could not resist the temptation of pointing this out...

;)

I wish people would just stop trying to make KDE just like Gnome or just like Win or Mac or whatever else. It's KDE... (that does not mean we should strive to improve things, but let's improve things our way, not the Gnome, Win, Mac, Next, OS/2, QNX, BeOS... way)


By Phantom at Wed, 2003/04/09 - 5:00am

> As a matter of fact, many people are now setting up KDE for things like their families and rarely do they complain about things like too many configuration options. This is a complaint that you rarely see from end-users. I haven't seen many emails saying, "I want less options and features!". It is something "usability experts" and pundits say.
----------------------

I'm sorry to say that but that would be the *weakest* argument in this whole configurability debate!

If a newbie is confused by the tons of options in KDE you can be sure that he won't send the KDE team messages about the fact that there is too many options! Chances are that he didn't even realized he was using KDE in the first place! He'll probably just tell everyone that Linux is too complicated and that it sucks and reboot into Windows. The fact that nobody complains about something doesn't means that everyone like it.

I don't think that the problem is with the number of configurability options, maybe more the way they are presented to the user. Reducing interface clutter doesn't systematically means removing features. A good example of this is with the trash contextual menu. You had options to browse images, burn files, etc... appearing in this menu... a user had to look hard to see the most important option, empty trash. Is the browsing images a good feature, yes, is putting it also on the trash context-menu a good idea? No. I'm aware that this problem was fixed recently (i think).

To play devil's advocate, why has been this fixed at all? End-users never complained about the fact that the context-menu of the trash contained too many options! ;)

The goal of "less-is-more": ergonomy, not stupidity


By Louis-Philippe ... at Wed, 2003/04/09 - 5:00am

Trash too complicated? There are 3 options in the context menu of my trash bin:
Open
Empty Trash Can
Properties

I never edited and services menus, its just the kde default (deleted the .kde directory in my home directory, compiled from sources)

I personally don't see how the options can be confusing, then again I've been compiling programs since 3rd grade. Everything for me is layed out in a neat fashion: everything dealing directly with the ui goes in "Appearance & Themes" File and Network operations go in Internet & Network (though I can see where someone wouldn't register that would include local filesystems). etc.


By Brendan Orr at Wed, 2003/04/09 - 5:00am

oops, didn't read the "fixed recently till now" sorry, just omit the first 5 lines :)


By Brendan Orr at Wed, 2003/04/09 - 5:00am

Mine shows 11 options including "open with", "open terminal here", "create data cd with k3b", etc...

Your example and mine shows exactly what can be done to remove UI clutter without removing functionnality and flexibility. Are there many peoples that create data cd from their trash contents? The KDE developpers are certainly agreeing with me because this has been fixed.

Now fix the other similar issues, if the options are simply stupid, remove them, if some peoples want them, make them configurable and *remove* them from the default behavior!

Many people here raves about KDE configurability... well it seems that KDE is not configurable enough because i can add new features but cannot remove many of them. (like the "up" context-menu item in Konqueror).


By Louis-Philippe ... at Wed, 2003/04/09 - 5:00am

did you compile from source or use binaries?


By Brendan Orr at Wed, 2003/04/09 - 5:00am

From sources but not by hand. I use Gentoo so everything is compiled from sources.

Why?


By Louis-Philippe ... at Wed, 2003/04/09 - 5:00am

Uh.... everything is compiled from sources in some way.
Since you are using the gentoo scripts, and not your own options, it is pretty much the same as taking binaries, really, unless you customized them.


By Roberto Alsina at Thu, 2003/04/10 - 5:00am

"Uh.... everything is compiled from sources in some way."

lol

tell me about it, being a programmer it's my day job! ;) Of course you know what i meant.

"Since you are using the gentoo scripts, and not your own options, it is pretty much the same as taking binaries, really, unless you customized them."

That's why i pointed out that i wasn't compiling "by hand". However i'm still not following you. What does it have to do with what i was saying?


By Louis-Philippe ... at Thu, 2003/04/10 - 5:00am

I'm talking more about red hat's packages, They took varies liberties with respects to some of the menus


By Brendan Orr at Thu, 2003/04/10 - 5:00am

> Mine shows 11 options including "open with", "open terminal here", "create data cd with k3b", etc...

KDE-cvs has exactly 3.

> Now fix the other similar issues, if the options are simply stupid, remove them, if some peoples want them, make them configurable and *remove* them from the default behavior!

The problem is... this has already been done for a lot of things.. the result is the large amount of config options that exists today.

The solution is of course to add a second way of configuring that is tasked based. Somewhat how WindowsXP has both task and traditional methods of configuring.


By fault at Thu, 2003/04/10 - 5:00am

Pages