16 June, 2010

Flameworthy LinuxTag Notes

As I kind of promised in my previous blog - here some notes from LinuxTag, worked out in an opinion :D. Now this blog is different from my usual ones. While I do write about strategical stuff, I try to stay away from controversial things. Not this time, sorry.

KDE in the early days
At LT I spoke with quite a few interesting people, including Georg Greve (the dot will feature an interview with him soon) and Matthias Ettrich (who started this whole KDE thing). The latter had a fairly interesting opinion about his 'kid'. After we started talking about stuff not for this blog (mostly politics and economics) we came on the topic of why he started the KDE project.

Basically, it was to give the common user access to the freedom and power of a better platform (Unix/linux). When Gnome started he was fine with that as some in the KDE community moved over there and added a whole bunch of features he didn't want in KDE anyway. But at some point, Gnome and KDE started a feature race - which was lost by Gnome, so they re-focussed themselves on user experience with help from Sun. Basically, Gnome 2.0 was what Matthias had envisioned for KDE in terms of user experience. He decided things didn't go his way, and instead of fighting or following the new direction he stepped out and started working on Qt.

KDE now
Since then, a lot has changed. That doesn't include the difference in focus between Gnome and KDE - Matthias complained Gnome builds a great user experience on inferior technology, KDE creates super technology while not doing enough on usability. He's not wrong.

So KDE has created a very open culture which results in innovation, experimentation and new technology. The user experience, while more of a focus than in the KDE 3.x times, imho still ain't what it needs to be, might never be the way we currently work. At least, the finishing touch is boring and hard to do in such an open meritocracy.

How to create usable software
You'd need a strongly design-driven development for a finished, consistent experience, were developers follow a common vision laid out by a few brilliant designers. And follow through on the boring stuff. This simply doesn't fit how we work - developers decide on what happens to their applications and what they spend their free time on. So different applications have different ways of solving certain issues (like with the + hover button on folders). This leads to innovation, and in time to better solutions - but while these are fleshed out (and there is always SOMETHING going on) it makes the whole set of apps less consistent, less stable, less usable.

Another issue is the large influence users have on how we work. Yes, that's right, I'm blaming active users here for unusable software. Many of our vocal users are powerusers and demand features easily accessible which only 1% of the world needs. And we cater to them, which often leads to a more bloathed piece of software. Especially removing functionality often leads to a big flamewar. Gnome simply ignores those who don't agree with the vision they have, and the end result is at least more consistent.

The last mile
Many in our community hoped the distributions or other commercial parties would be able to pick up where we left, and finish that last mile. They didn't - well, they tried, but every distribution which successfully did so either didn't contribute upstream (xandros) or went belly-up, was bought or had to change focus (mandrake/mandriva, suse). Ubuntu does the last mile for Gnome but Kubuntu lacks resources to do it for us.

What do we do?
Now we could start some top-down effort and force these things. Kill innovation, chase away a large proportion of our brilliant developers, having less fun. Please let's not. KDE, as it is, is great. We have a lot of fun, we innovate, and most of our current users love our products. So I am not advocating any change in how we work here. But we need to create a more usable and stable product if we want to grow beyond 1% of the market. So I think change needs to come from the outside. From a new start. How? I do have ideas, but I'll first put on my flamesuit and enjoy the heat...

What do you think?


  1. What you are saying about Gnome is exactly what I don't use it. Developpers are thinking that the know what is the best and that bring something like the so-called "spatial navigation" a big mistake. I love KDE having plenty of option and let me choose what I want.
    Having a clearer or more define default settings ok but don't go the gnome direction please.

  2. I am a recent KDE-to-Gnome-convert. Seriously don't want to start the-3.5-was-better-thing again (which is probably bollocks anyway), but while KDE 4.x has way, way better technology than Gnome, I still switched. Simply because Gnome - or the polished up versions from Ubuntu and others - gets your work done more than KDE can at this point. Technology is all great, but only if it is finished and works a hundred percent - which unfortunately for me KDE 4.x never did, even in its OpenSUSE incarnations which I think are pretty polished. Some weird Akonadi-Strigi-w/e-popups all over the place (which even as a experienced technical guy I never, ever understood and probably never will unless I deepdive into that stuff) and the whole SC just seems to have too many rough edges still, which seriously hamper at least my productivity. At the end of the day noone cares what great technology platform I did my stuff on, they just want finished work for which I get paid. So I absolutely support the idea of less-technology-more-usability.

  3. > "So I think change needs to come from the outside."
    I think I misunderstood you here. In my view change could be perhaps activated from outside, but when it doesn't come (or at least is not supported) from inside it will fail.
    In all other points I totally agree with you. If we want to increase our market share we need more usability, simplicity (for regular users) and perfectionism.
    How could we do that?
    - Recruit more usability experts (we already have excellent once) and connect them with the developers. Most developers are open for such ideas!
    - Train developers. Might not be so good. They have other focus and skills.
    - Use collective intelligence and let many users improve the usability. Perhaps a new tool is needed. Simple, easy, quick. Maybe the forum or one of the x-bases could provide a perfect place.
    - Could that be a junior job for new contributors? Maybe we have the same "a developer is not automatically an usability expert" dilemma or missing usability expertize.
    - Company involvement. Sounds not so bad but what could be the business model for such an enterprise? Another distro? FluffPRO?
    - Or just work more efficient together with existing distros? Or coordinate and share different efforts between KDE and distro communities? (The distros should cooperate more anyway. We should not fight against other distros. That doesn't help anyone but MS and Apple)
    - ...
    - Finally support the plan by heave "marketing".

  4. No, please don't cut features. KDE4 is still missing features compared to KDE 3.5, however, it's slowly getting there again. We don't need a second Gnome with a dumbed-down interface. After all, Linux is about choice. The day KDE starts cutting features will be the day when I give up on the Linux desktop.

  5. >But we need to create a more usable and stable product if we want to grow beyond 1% of the market. So I think change needs to come from the outside. From a new start.

    As a non-expert, I agree with a previous comment, that change can be stimulated by an outside process but needs to be "owned" to be effective.

    A couple of ideas:

    1. A complaint that I've seen pop up a few times is that the "default" KDE settings don't work for people. A good and relatively easy solution would be a one-stop "Export/Import Settings" option, which would let users and distributions tweak and distribute settings more easily. Some of the things I find myself customizing are font sizes, font anti-aliasing, power profiles, theme/plasma options, dolphin file view defaults, etc.

    This tool could even develop into a "usability profile testing" mechanism - test different usability scenarios with ease, encourage users to play with different pre-packaged usability settings. Especially for wider users, changing a theme/wallpaper/font setting/etc. is a lot less intimidating than changing entire software packages.

    (Side note, I think KDE also needs a database export/import GUI solution for backup and transferability/portability)

    2. Kubuntu had a project (Timelord?) to eliminate annoying bugs. I know KDE has coding sprints, and I'm sure bug-solving is a part of these sprints, but maybe bug-solving needs to be its own project/event.

  6. Gnome is not even at half as usable as KDE is. It's missing many features and those which are present are not so rich. Its menu and control panel are messed up. Some apps like totem are jokes. I don't want hear a single word from the gnome fanboys (from the gnome camp especially) about usability, because they don't know the thing. Gnome looks so artificially. Does usability mean I've got to switch WM just to disable compositions? Does simple feature to turn it off doesn't fit to some small and stupid gnome devs world? Or maybe totem is an usable video player? KDE: Plasma, Amarok, Dolphin, Konqueror, Kopete, Akregator, Digikam, Gwenview, Koffice etc. Any competitive __GNOME__ equivalents? Not a single one, except Pidgin, but it was replaced by some crap like empathy which is useless. Every DE needs improvements, but Gnome is stagnating at it's years behind KDE in every aspect.

  7. Well well. I actually expected flames for KDE, not Gnome, here. Anyway, when it comes to usability: 90% of those features you mention which are in KDE apps but not their gnome equivalents are rarely needed. If I give my mother a digital camera and ask her to put the images on Flickr, I doubt she manages to do it with the KDE apps without guidance but she can probably do it with Gnome and certainly with Mac. And this is not the only use case I can think of. So there IS a difference, and it is something we must work on.

  8. I think you are on target. I can understand the potential is stuff like Akonadi and Nepomuk/Strigi, but mostly what they do today is get negative feedback from users. We still don't have an included network manager front end for KDE4 after how many years?

  9. IMO KDE should have a division which acts as a distro and provides what you called "the last mile": something like a meta-distro which provides patches, packages, default configs and cut off of "useless" features for a perfect out of the box experience.

    This should be led by the "few brilliant designers" that follow the vision of usability.

    Distros then should be able to choose which kde flavour prefer (vanilla or usability-enhanced)

  10. @last anonymous that sounds like a good plan, yes. But how do we get it? It needs a few usability specialists; and a bunch of developers implementing things. Might be boring work, money would help a lot but there is probably none to be made...

  11. You want to become KDE yet another desktop that thinks "cutting features" is somehow a feature? GNOME already does that just fine. What desktop shall *I* use? Xfce with less less features? Enlightenment, LXDE, Fluxbox with less less less features? MS Windows, maybe?!

    KDE 4 does not have too many features. It is still desperately lacking some features from KDE 3.5, especially regarding the K menu. How do you edit a file as root if you don't know what to type in KRunner? Fairly easy in KDE 3.5, there is a button to click and then you can enter another user - you can also right-click on programs in the K Menu and select the option, so you don't even need to know to press Alt+F2 and type kate. You can also edit menu entries by right-clicking on them. I accepted that all of this was not in KDE 4.0, but now we are at KDE 4.4 and IT'S STILL NOT THERE. A new user cannot find out anymore how to run something like a text editor or dolphin as root. The instructions given e.g. in a computer magazin that s/he needs to be root to install the Linux program on the accompanying CD is useless. (This is a real-life example from a friend who switched from Windows last year.)

    KDE's usability is mostly fine. You will not improve it by removing features. Removing features is an insult to users. Not have a checkbox for making the title bar blue, even at a time when there is no other indication whatsoever which window is active, not even the ugly stripes yet, not the blue glow, which you cannot see when you have two maximized windows on your two screens? BECAUSE IT'S SUPPOSEDLY UGLY? Because "we must not allow users to debeautify our desktop" (or something along the lines, a very sad day for Linux and KDE when that e-mail was written)? Be angry at the person who creates a fork with just that checkbox? Be angry at distributions who add this forked theme? WHAT WHERE THEY THINKING?!

    I don't think the Flickr example is right. I don't use Flickr, so no idea. But what I did try was starting a slideshow with photos on the GNOME that was preinstalled on my Netbook. I did not succeed in the one hour I tried showing them to my friends. We resorted to manually opening and closing them. (The solution is that you have to start the program, open a folder with it and then start the slideshow, opening a picture won't work, as I found out later.) It's easy in KDE. Though it could still be easier ... didn't it use to be possible to start the slideshow without having to maximize Gwenview?

    So yes, improve usability even more. But don't even think about removing features to achieve this. Because it doesn't.

    Enough flame? ;-)

  12. Jos, if I want to upload things to Flickr, I plug in my camera, click on the suggestion to open digikam, download the photos, then export -> Flickr. Not that hard.
    I don't agree with Gnome being more usable, just dumber, and because of that simpler.
    As a developer, one thing that does a lot for consistency is having the core libs implement the common features, and only have to implement those features which are particular to my app. That has done a lot for KMyMoney during the port to KDE4, which is really a step forward in consistency and usability, IMHO.
    Also, have some usability people go through apps and file bugs. Have those bugs discussed in the larger groups, like kde-core-devel so that common solutions can be found. It's not that the last mile is boring, just that we don't see it sometimes. If you don't have some user coming to you and pointing at the error, you just won't see it. So, start filing bugs and talking about it.
    Again, dumber might mean simpler, but not more usable. We DO NOT want the gnome way. The moment you start treating your users as stupids, you'll only get stupid users.

  13. Unfortunately, the way Gnome has approached usability has ruined most of you ;-)

    When I say more usable I don't mean we should remove features. It is about exposure and configuration. For example. If you currently start Dolpin on a default installation, it starts up very small in the top-left corner of the screen with two big dockers and barely enough room to show 3 file icons. WTF? Dolphin is a shining example of great usability within the KDE SC, yet this is still screwed up. Fixing such things can help a lot.

    So again, not removing features, but more sane defaults and maybe a few patches could make a huge difference. The average user doesn't need to be able to rotate plasmoids, so let's get rid of that button. Digikam has a hugely complicated interface, compare it to the Apple photo editor/manager. Probably almost as powerful yet it looks 10 times simpler. Developers should look at the competition and copy the best...

    Anyway, however form this takes, maybe it should be another SC or something, developed within KDE SVN/GIT and doing as much as possible in the normal SC... But because it is separate, the tough decisions can be made if needed.

  14. "The average user doesn't need to be able to rotate plasmoids, so let's get rid of that button."

    Sorry, first you say you don't want to remove features and then you're proposing exactly that. I *do* use that button.

  15. You're not an average user I'd bet, I can't really think of an usecase relevant to most people. But it might be my mistake. Either way, this can be done configurable I bet. And you still have the choice to use the default plasma, remember, this ain't for anybody reading this blog but for ppl who don't like computers... I'd probably not use it myself, I guess. I like trowing plasmoids on the panel. My dad wouldn't even understand it if I tried to explain what a widget WAS in the first place.

  16. He he. I thought for a long time that the rotate button was a reload button that didn't work. I cannot see a use case for it

  17. I use the rotate button to arrange photos, notes, etc. more naturally on my desktop. I think it's a feature that especially average users like a lot.

  18. I'm the anonimous that suggested "two" KDE..

    Maybe if the KDE e.v. stands up and organize something like this (most of) the commercial distros will contribute money or manpower, a project like this would increase the quality of the software they ship and to make them happier tools for simple differentiation could be provided and mantained.

  19. Maybe if Gnome "upgrades" to Qt we could see KDE and Gnome merging... the Gnome guys really have some good points when it comes to a consistent user experience.

  20. I know some things about usability, perhaps not enough, but enough to wonder "what are these usability problems that kde suffers from but not gnome?" The only concrete example I've seen is "too many options" (now, granted, I haven't gone out of may way to look for them either). Given that with sane defaults options won't matter for the majority of the users who don't bother to change the default settings. For those who need to change the default settings, you'd have to identify which options are problematic - and by then you're not discussing "options", it's a discussion about GUI design.

    What I'm trying to say is that saying "there are problems" isn't enough, if you want to discuss something, I think there is a need for specifics. Because there are other kinds of usability problems and not all of them can be addressed as "KDE" problems. For an example, that akonadi doesn't provide precise and constructive error messages can be considered a kde problem, at least for 4.5 but is it a kde problem that konq will render black text on a dark background when using a dark theme and visting a page without a set backgroundcoluor? (while on the subject, konq is a good example of where they improved the usabily from 3.5 which shows that configuration of applications doesn't have to be on a gnomelevel to be usable)

    I suspect that if you were to test "using KDE" with say, Neilsens heuristics, you'd find that most usability problems, and also the most severe, identified would be in non-core componets and not relate to the number of options (for an example, this was written in black on a dark background in rekonq since konq repeatedly died on this blog.

  21. "Many in our community hoped the distributions or other commercial parties would be able to pick up where we left, and finish that last mile."

    "Anyway, however form this takes, maybe it should be another SC or something, developed within KDE SVN/GIT and doing as much as possible in the normal SC... But because it is separate, the tough decisions can be made if needed."

    Using another git branch is exactly what I thought of in response to your idea of getting last mile improvements from downstream -- we can be our own downstream.

  22. A company could be set up to deliver a polished KDE experience. Work should be focused on stuff that organizations want: Kiosk functionality, seamless integration into their network, remote desktop, instant messaging, ...

    Remember how Ximian started. With software like Red Carpet (package installation) and Evolution.

    Delivering support is also a major point. Companies want to contact somebody when something goes wrong. Contacting an FOSS developer directly is not reliable enough, he might already left the project or could be on vacation. A company that can support them gives confidence.

    After the company is running well, you might want to shift more efforts into usability.

  23. Too many options are certainly something that might scare my mom. Remember how windows offer few customization for the workspace and how most people doesn't even bother to try them, or just ignore them. On the other hand, as the comments here show it, the fellow kde fans want to customize their workspace.

    The lock widget thing is a good thing imho, but could even be hidden further away. My mom will certainly make no use of it, im sure she doesn't want to know she can add a comic strip on her desktop. The 'activity' things are certainly not for her neither - what the hell is a folderview? i made a test once with putting a kde laptop in the hand of my father(which is the one who 'teached' me computing science and the one his friends call to fix their com) and that is the thing he couldn't understand... why can't i add icons on my desktop? what is a folderview? -, nor the desktop shortcuts. Remember that most user don't use shortcuts except ctrl-c ctrl-v. Alt-tab seems to be used more and more, but still slowly and by a minority.

    I think the idea of slowing down and fix the bugs and usability sounds great. It seems I hear this from a long time already though.

  24. I know the corporate involvement thing is not popular amongst many FOSS people, but one of the advantages I think this gives gnome is people who can work full-time on this sort of polishing. I think if KDE just had a handful of people, a few developers, a few usability experts, a few artists, who were paid to work full-time on the "last mile" part of KDE I think it would make a huge difference.

    I am not sure where the money would really come from. Maybe if someone could convince some companies that things like akonadi and koffice are the way to go and that they should enter a pool to hire these people it might work. But I really think unless there are people whose job it is to do the "boring" stuff full-time then we aren't going to see a benefit.

    So I guess I don't really think it is necessarily a strong top-down approach to decisions, too many features, or issues like that which differentiate KDE from Gnome. I think it is the fact that doing the last-mile bits are boring so people just don't put in as much effort in those.

  25. First of all, I'd like to sincerely commend you for this blog post. Pardon my french, but, it's about fscking time somebody "from the inside" brought these issues to bear. You rock!

    While we're being controversial, please forgive me. If we look at the history of humanity, human societies are *always* held together by leaders. Leaders, inherently, have significantly better understandings of how humans work. They are the ones who have the knowledge necessary to advance society, and they are the ones who are *responsible* for directing that society. The challenge, of course, is to do so such that the followers deeply believe *they* are the ones doing the directing (not think, not hope, but truly believe, because after all, it's human nature to believe that the self knows it all - everybody loves to give their opinion).

    Of course, if the followers were *actually* doing the directing, that would be anarchy, and everybody knows anarchy doesn't work. In other words, top-down is the *only* way. But doing so by force is not the answer... it works best when the bottom *truly believes* they influenced the top. If you need a model of how to do that, just read between the lines of any government who tells their people they are free.

    Good luck and god speed,

  26. What should be done is not removing options but putting least used features a few clicks away in things like docks or menus and most used features more visible in the ui.For example gwenview has export function for flickr but it is under plugins>export>flickr.
    This is a feature which is used a lot I think so it should be more accessible (A toolbar button for example), also features which are not so used should be more hidden in the gui or grouped into wider concepts for example, if crop,resize,etc were not used (I know they are just suppose they don't) they should be grouped under a operations button for example.
    I think the key to usability is not removing options or features but put the 10% of features that you use 95% of the time visible and clean, and the other 90% features hidden in accessible menus,or docks.(maybe hidden docks?)
    About who should do this work, it should be done just like the rest of the project, I don't see it boring, just as some people are code geeks,and will see this boring, there should be usability geeks out there willing to do this. But if not then companies,distros,donations,etc should help.
    PD:About plasma distros usually recieve you explaining what widgets are and other stuff, I guess users don't read it but maybe when first used desktop could show small tips about widgets how to add icons etc.

  27. I think the biggest problems in KDE 4 is lack of some good features, lack of good defaults, inconsistencies in some areas and good apps that has just been abandoned (or almost that).

    If some people are complaining about too many configuration options it is because they want to change something and have difficult to understand how to do it. If we have good defaults few people will have the need to change them. Having fewer options does not help per si, but having few very usefull options is better than having mostly unused features, that is what Ubuntu and Google do in their apps. I prefer KDE adding very usefull options than removing that ones we already have.

    Apps like Kopete, Amarok K3B, Konqueror, Quanta, Knm (K Network Manager) suffered a lot during 3.x to 4.x transition and are still lacking features from the older versions. Although I do not miss all the features from the older versions I miss some very usefull ones. I do not see a desktop today that does not need a network configurator, an instant messenger app, a music player, a web browser or a burner app (except for Netbooks, obvious). So I do not understand how we let that happen to our flagship apps. Amarok and K3b used to THE app for playing songs and burning CD/DVD, no other opensource app had their appealing. Now serveral people complaint about Amarok and K3b does not have a stable version after two and a half years since 4.0.0 release.

    I think we must add only usefull features to KDE software, specially if they are basic features. Basic features also means they will be used by most people. I am improving 3G connections in Plasma NM and I still do not understand why that have not been done before. That is a very usefull feature set and nobody stepped up to do it. More important than adding a feature I think is making it work properly, some KDE sofware have the same bugs for years, we should solve the more obvious ones and the ones that affect basic software functionality. One thing we should know is that if something does not work most people will not come to us and say: "somethings is not working, I fixed and here is the correction". Most people will just give up using that software and depending on how important that software is they will give up using KDE SC too.

  28. I should probably qualify what I just said about having developers paid for the "last mile". I think these developers can't just be generic developers paid to help with the last mile. I think they each need to have specific jobs that they need to do. For instance there could be someone whose job is just to go down the list at bugs.kde.org and fix bugs, starting at the "worst" and moving down. Another person could have the job of doing the same with wishlists. There could be someone whose job it is just to implement usability improvements highlighted by the usability team. Another person might have the job of bringing KDE up to feature and ease-of-use parity with other desktop environments (like how the OP said KDE could get idea from other operating systems). Another person might just work on speed improvements. Another person might be a corporate liazon who only works on fixing things companies would need. Another might just work on making things look and behave consistently.

    These are just examples, but these are the sorts of jobs they might do. These are the sorts of boring grunt-work that are unpopular with volunteers, but are essential if KDE is going to succeed. Developers do these things, but some more than others and the efforts aren't really coordinated.

    These people, of course, would not be working alone. They would be getting help and giving suggestions and coordinating their activities with developers. They would not be bosses telling developers what to do (although they could certainly give suggestions), instead they would be just doing their specific job, but since that job spans all of KDE they would have a better global picture of what needs to be done so their efforts would not be as fragmented. I don't think it is so much strong top-down control that KDE needs, volunteers being able to do what they like is critical to KDE's success, rather it is people whose sole job is to look at KDE as a whole and work on KDE as a whole that I think KDE lacks but really needs. Developers can do their own thing as long as at the end of the day there are a few people who are able to do the little things that tie everything together.

  29. "What should be done is not removing options but putting least used features a few clicks away in things like docks or menus and most used features more visible in the ui.For example gwenview has export function for flickr but it is under plugins>export>flickr.
    This is a feature which is used a lot I think so it should be more accessible (A toolbar button for example), also features which are not so used should be more hidden in the gui or grouped into wider concepts for example, if crop,resize,etc were not used (I know they are just suppose they don't) they should be grouped under a operations button for example."

    Great in theory, but how do we assess what features are used often and which aren't? For instance I never export to flickr, but I resize, crop, and rotate all the time. So which if us is right? Who decides which features should be easily accessible and which aren't?

  30. I would like to note a few things...
    1) I use KDE precisely because it lets me do things that no other desktop that I have used lets me do, or does it better.

    2) The default stable of applications in KDE is really targeted towards advanced users. Look at Sound Juicer in GNOME it's a great single task oriented program that anybody should be able to pick up. The equivalent functionality in KDE is usually covered by K3B which is awesome but overkill for most people.

    3) The bits of KDE that suck the most generally are the bits the distros do (especially with Kubuntu). End users don't know the difference and assume it is all KDE. It great that KDE now has a built in networking component and that pacman is being ported.

    4) Distros like Ubuntu put a lot of polish into the Gnome environment - AFAIK there is no distro with the same amount of resources trying for a similar kde based distro. This does make a difference.

    5)The default look is OK but not that great. I am very particular about how my desktop is set up so this doesn't bother me. I always have a beautiful desktop :).

    6)It's hard to make everybody happy at the same time.
    If you put more space in the UI some people will complain that it wastes space, take it away and others will complain that it is now too cramped. Oxygen does need some attention to detail with spacing.

  31. 7) A lot of KDE's best features aren't really that discoverable. I think that might be the secret

  32. "I think we must add only usefull features to KDE software, specially if they are basic features. Basic features also means they will be used by most people. I am improving 3G connections in Plasma NM and I still do not understand why that have not been done before. "

    Once again, there is a lot of disagreement amongst users as to which features are "basic" and which are "advanced". I, for one, would consider support for back and forward buttons on mice "basic", since basically every other modern desktop environment has them. Other people, however, do not care. Other people, however, consider quickly changing keyboard layouts to be basic, while for me it is useless. Some people think audio support in bluetooth to be the most important feature, other an obex kio slave. It is easy enough to say that certain features should be implemented and certain are a waste of time, the problem is that nobody seems to agree on which is which.

    In short, it is easy to say that we should be focusing on only the most important features or the ones that are the most commonly-used, but such comments are obvious to everyone and don't really help. I doubt anyone would disagree with them. The problem is figuring out WHICH features are basic, which features are needed most by people and which aren't. I think without a specific, practical proposal on how we could actually determine which features would be most useful and most-used such comments are not the least bit constructive.

    Especially not helpful are comments like "this feature is basic because I use it a lot". That completely misses the whole problem with the subject: no one can agree on which features are the most important.

    The same is true for setting "sane" defaults. Everyone agrees that good default values are needed, but no one can agree on what is a good default and it seems no matter what is chosen some segment of the user base will scream their heads off because it isn't how they want it. For a specific example, for a vertical panel should the task bar text be rotated or not, and how many row should it take up? If it is rotated, it is hard to read. If not, it takes too much horizontal space so the panel has to be too large. If it is on multiple rows it takes too much vertical space, if it is on one it takes too much horizontal. When it is one way, lots of people complained. When plasma devs changed it, lots of people complained about the change. How do we decide who is right? Pretty much every decision will upset someone, and those people are the ones who scream the loudest. It is easy to suggest that we need sane defaults, but without specific proposals on how to choose them then it doesn't really help anything.

  33. "Look at Sound Juicer in GNOME it's a great single task oriented program that anybody should be able to pick up. The equivalent functionality in KDE is usually covered by K3B which is awesome but overkill for most people. "

    KDE has that: the audiocd kio slave. KDE's basic CD ripper is built right into the file manager. Which gets back to your point #7 ;)

  34. I really appreciate your post, Jos, and I think that it is a great idea to start a conversation about these issues and gather opinions about possible options. However, if this is to get anywhere, in the end it will require somebody to take charge of it as a project, identify exactly what the problems are, what possible avenues can be taken to solve them, and to start a process to implement the solutions. I disagree with the previous comment that says that this needs to be top-down. That someone needs to take the lead does not necessarily mean that it has to be done in an authoritarian manner. Each step could (I would say 'should') involve openness to the community, deliberation, and a clear process for decision-making (which does not necessarily need to be an open vote).

    But from what I read above, it is necessary to agree on what we are talking about. Some people talk about 'rough edges' around the SC, which seems to be more about bugs and things that do not work properly than the more restricted usability issues that you seem to be talking about. Other people are talking about the amount of configuration options available, which is one way of dealing with usability issues, but by no means the only one. So should our focus be on how to deal with usability in the restricted sense (maybe with a few guidelines, a usability team that developers could consult with)? Or are we talking about making sure that KDE works for the average user out of the box, which could even mean, as suggested above, shaping one stripped down version of the KDE SC in each release? The latter option would mean making specific usability decisions as well as keeping the releases free of buggy programs and of technologies that are not ready for the masses yet, and this would have serious practical implications such as duplication of resources, a lot of extra work, a need for a new type of developer contributions, possible conflicts with developers who are used to working in a different way, less exposure for new technologies which would thus take longer to mature, etc.

    In the end, I believe that figuring out what to do requires a more detailed diagnostic process than what we can gather from the responses to a blog post (even if I think that this is a very useful first step). So I would suggest finding someone who is legitimized by the board to start a process of figuring out what the usability issues really are and to present alternative courses of action that the community can discuss.

  35. bla bla bla. this was interesting about four years ago...

  36. >like with the + hover button on folders).

    I would like to see more buttons like this. What about another button for "add to playlist" at the other corner? What about a "delete" button? I think more of this would improve the usability enormously. Also this feature should be more consistently used alond the SC, Doplhin and Gwenview is not enough, IMO.

    I always compare KDE to Apple Interfaces, eg Mac OS X and the iOS4. You can find many good usability examples on the ipad. Take pages on iPad as an example for a nice UI. Very polished and unconventional, compared to standart office UIs. MAybe we should be more unconventional.

    I would support that LAst-mile-Distro. I could imagine to test some certain layouts for desktops and icon themes, or some new menus and programs. As an average user there is just too few to test, because devs venture too little. Plasma Desktop Layout hasn't changed since its first days and is even the same like the early KDE2/3 days.

    I like how Gnome gets polished in Ubuntu. Take the Me-Menu, which is a nice and slick feature. Or the easy backup mechanisms and the UbuntuOne integration. I like that they work on the whole system and care about just the desktop or the themes. I admit that this is easier as a distro since as such one can care about consistency beginning from the boot to the certain programs.

  37. "I know the corporate involvement thing is not popular amongst many FOSS people, but one of the advantages I think this gives gnome is people who can work full-time on this sort of polishing."

    There's nearly no polishing in Gnome. I'd fire almost every guy who's doing some usability job (not only for Gnome, but for Windows too), because it's just a fake job!

    @to the guy talking about consistency

    There's no more consistent DE then KDE. Gnome is not consistent at all: mono, java, gtk, python, theme engine messed up, etc. etc. mix compared to C++/Qt and very, very little percent of python and java.

  38. "Great in theory, but how do we assess what features are used often and which aren't? For instance I never export to flickr, but I resize, crop, and rotate all the time. So which if us is right? Who decides which features should be easily accessible and which aren't?"

    Ordering things in the gui is the key, it is not that much about deciding if this or that option is really used.
    The key is making features more easy to use and more discoverable for example grouping all export options of gwenview under 1 "download/upload" button in toolbar which opens 1 window which prompts for service,etc. Or putting all batch prossesing options in one button in operations dock.

  39. Some suggestions:


    What about setting up a dedicated, more wiki-like tracking system for usability improvement ideas?

    Using bugs.kde.org for usability discussions between developers, community members and users tends to get kind of frustrating for all involved parties (see e.g. https://bugs.kde.org/show_bug.cgi?id=157284 for a sad example).

    Instead of the developers of each respective application, a combined KDE usability team would lead and moderate this new tracker/wiki. Together with the whole community which would be able to openly contribute, they would try to "develop" submitted usability ideas further and formulate polished solutions, which then the application developers could implement (or not, if they don't want to).


    What about setting up a "KDE usability - going the last mile" account into which people can donate money, with KDE ev. additionally transferring a certain amount of money there each month (as much as it can spare)?

    Out of this account, KDE developers could be paid for investing a certain amount of time for working on specific usability goals set out by the KDE usability team.

    Kind of like a mini-GSoC, except for shorter periods of times, with existing KDE hackers as the target group (mainly), and managed by the KDE usability team instead of Google...

    If it's really the case that "going the last mile" regarding usability work is just too boring for KDE developers to do it in their free time, this might be the solution...

    For me as a user, this would actually provide a real incentive to donate some money..

  40. Complete the HIG. Look at the tabs in Dolphin, Cantor, Blogilo and Konsole - all 4 are different (behaviour of tab-ordering and style of close-tab-buttons). Such things have to be described in the HIG. And it needs to be enforced (no new features / applications in KDE SC that violate the HIG).

  41. I read an usability study about the iPad and it explains how bad it was its usability.
    I don't actually have tried one, just read ;)

  42. OMG! It's almost embarrasing when people show ubuntu next to latest osx or even seven. It looks terrible! Something from 2001 desperately tried to make look "modern". Like ms would now try to sell windows xp with a new skin.

    At least you can make KDE4 look like cool something from this decade, so yeah better defaults would be nice and maybe some love for bespin. Some of the themes made for it looks very cool.


  43. "Matthias complained Gnome builds a great user experience on inferior technology, KDE creates super technology while not doing enough on usability. He's not wrong."

    He's not only not wrong, he's absolutely right! I just recently switched to Ubuntu 10.04 with Gnome on my laptop and couldn't be happier. Where KDE didn't even provide me a proper NetworkManager that can adequately list the >30 WiFi signals in range at my appartment, the one I have under Ubuntu Gnome just does it's job and get's out of the way. Until KDE achieves this kind of polished UI, it will remain a tiny niche product.

    And just to reiterate the obvious, this is not about KDE's technology but about what is (not) being done with it that really matters for end users.

    Really really good idea!

  45. Jos, this is a great post. Lots of insightful comments too. This all got me thinking up some crazy ideas... It's a bit long, sorry about that... I tried my best to keep it as short as possible.

    Regarding "target users", "smart defaults", and the extent of features:
    As a student of design and usability, I've was taught to design for a specific audience with specific needs. GNOME, IMO at least, accomplishes this by designing for the "base case" and assuming more advanced users will still be able to use the final product. But they also cut "advanced" features left and right, which personally, I don't think this jives with "the KDE vision." I don't think these must go hand in hand if the product is still to be easily usable and discoverable for 'random person X at the grocery store.' However, if in the same interface we are to simultaneously stay true to our configuration-centric heritage, this obviously presents a significant challenge if we are to significantly expand our user base.

    And, that's not to even mention the problem that we're trying to target KDE at devices ranging from smart phones to kiosks to notebooks to quad-monitor workstations. And that there are many different input technologies, sometimes but not always mutually-exclusive available for each of these device classes. Furthermore, each different ability-level user has different needs and requirements specific to each of these devices. [I can be a PC power user but a smart-phone newbie.]

    Quite literally, the needs, desires, and technological understandings of "grocery store users" versus "power users" are about as close to exact opposites as one can get. Cars provide a good analogy. My brother Nick loves to work on cars, he mods the engine, the electronics, the suspension and so on. He's always changing things, swapping in and out new components with the latest-and-greatest. On the other hand, my grandma wants a safe, slow, super-smooth-riding car. The fact that it even requires an oil-change is a huge PITA. These extremes are roughly equatable to the levels of customization expected by advanced power users, as contrasted with the need to do routine security updates.

    [Continued below...]

  46. [Continued from above...]
    Instead of bickering over a mashed up middle-ground compromise, what it we could instead embrace KDE and FOSS's inherent advantages - I have universal integration and superior technology in mind - to design and build a system where the software can cater itself to the needs of individual users? I'm not talking about some kind of complex self-adapting observe-and-learn system. Just a first-run Welcome-to-KDE widget that asks the user what kind of computer/device they're using, and how they would rank their abilities respective to that device. This maybe sets a global flag which any application must read on start-up, and can automatically set appropriate UI/interaction defaults for the user's ability and the specific hardware this instance of KDE is running on.

    It seems to me that the vast majority of needs and understandings of "grocery store users" are really just further abstractions and combinations of the granularities expected by developers and advanced power users. Therefore, developers can continue to write their code just as they always have; I am of the opinion that KDE is already extremely good at figuring out *what* should be configurable. The how and when parts are a mess, but that's what we're trying to tackle, right? This is just dreaming, but say somebody were to embrace QtQuick with KDE-ized widgets and application layout/navigational structures which were capable of being scripted/pluginable/whatever to adapt themselves to the global usability flags. All of these would build upon the base of developer-oriented-granularity (as set by usability designers in a process maybe similar to what Oriol or Sam S mentioned). If anybody remembers the IMO insulting and embarrassing 2010 April Fools article posted on the dot, that's somewheres around like what I would envision as various final outcomes.
    Link: http://dot.kde.org/2010/04/01/announcing-upcoming-release-new-customized-kde-software-compilations

  47. I started using Linux with KDE (Kubuntu 8.04) so I also suffered from the switch 3.x -> 4.x
    All in all, I like KDE and I am getting more and more used to it. Anyway, some bugs or random weirdness still affect my "user experience".
    Why doesn't the KDE team just stop for a while adding features and tries to focus hard on optimization? After all, KDE already has many features and eye-candy as it is now. I think users can survive 1 more month without the n-th option to set if they can see some basic functions cleaned up.
    By the way, great work! I can't live without wobbly windows!!!

  48. I guess we need some persona's?

    Becky, 24 years old, likes to play tennis, no knowledge of computers (etc etc)

  49. @Anonymous, at thursday, 17 june, 2010


  50. All in all I very like the way KDE is doing its config.
    Neverthelesse there are a lot of papercuts, where small changes would greatly improve the experience.
    I think a new forum or wiki for this is a great idea.

    A small example for the lack of consistency: Look at the screencast of the different behaviour of Kmix, Klipper, KNM, Nepomuk, Battery and Amarok in the system tray when hovering, left-clicking and right-clicking: Not one program feels exactly like the other: Different notification, popping-up, sliding out or fading in... Of course, not all of them are fully comparable, but anyhow:

    http://www.cyrilbrosch.net/screencast.ogv (sorry, my first screencast, for the low framerate you donÄt see all details)

  51. More features doesn't mean less usable.

    The key is to not overwhelm the user straight away. For example Firefox can be used by an average person quite easily. Yet if you are willing to dig through the menus there are many options. 'Grocery store users' will never go there so that's not a problem. At the same time power users still have all their options available.

    As for what we need for sensible defaults, I think it would be best if we aim for the mainstream audience. Again these are casual users that are not specifically interested in computers. Although this may initially upset some power users(like myself), we have to remember that power users can still configure the program to what they want(whereas casual users can't/won't). Also it will benefit them both in the long run as more mainstream users means more market share, which means better vendor/hardware support etc etc.

    I believe it is possible to accomodate for both types of users. Again look at firefox. It is the main browser for a large number of casual and power users.

  52. if KDE goes the gnome way I'll completely switch to windows 7.

    And I think I will not be the only one. I was already tempted when kde 4.0 came out, since the lack of features compared to kde 3.5, but now things are slowly going back to normal, fortunately.

    KDE sure has defects, inconsistencies, etc. Like any other very large piece of code. But gnome sucks greatly, so please stop the useless flaming and please take a good look at what gnome really is.

  53. if KDE goes the gnome way I'll completely switch to windows 7.

    And I think I will not be the only one. I was already tempted when kde 4.0 came out, since the lack of features compared to kde 3.5, but now things are slowly going back to normal, fortunately.

    KDE sure has defects, inconsistencies, etc. Like any other very large piece of code. But gnome sucks greatly, so please stop the useless flaming and please take a good look at what gnome really is.

  54. you can see some comments here: http://forums.whirlpool.net.au/forum-replies.cfm?t=1468799&p=-1#bottom

  55. Gnome is only popular, because of stupid people propaganda: Canonical, Novell. Novel and Red Hat aren't even interested in desktops and they're using gtk, because Qt wasn't free at the begining and Novell is ms dog, so they don't want Linux to be popular on desktops (novell!=OpenSuse!!!). Canonical chose gnome, because it reminds os x a little and they compete with it. However, there are many gnome fanboys in Canonical, so this can be another reason of their choice. Using gnome is a real pain, so don't care, but keep working on the best DE - the K, DE.

  56. There WAS a civilized discussion here, let's keep it. I've deleted two silly comments with just name calling.

    Oh and yes, Gnome is a niche product. But so is the KDE software - compared to win and even to mac we're nowhere to be seen. This can only change if two things change: the business and the product.

    The business = the tricks MS uses to keep users away from FOSS. We can't do much about that, the FSF(E) might.
    The product = our software. It's now perfect for geeks, and with some configuration reasonable for advanced users. It's bad for beginners.

    We should provide that configuration and start working on the beginners stuff.

  57. "This can only change if two things change: the business and the product."

    No, product doesn't need to change. My young sister loves KDE and she prefers to use it instead of Windows and Linux+Gnome.

    Win and Mac are advertised everywhere. Even on the streets. KDE is supperior to what Win or OS X offers. However, there are things where Linux needs to catch up like video acceleration and sound support in some cases, but the most important things are games! If there were games on Linux, Linux+KDE would just rule. It's very easy to show Gnome is loosing compared to KDE, Win or OS X and that's why I said Linux+KDE would rule. Don't judge DE in this unfair race, please. Without games we just don't have to feel worse, because it's the most important thing for desktops.

    P.S. sorry for silly comments, but I couldn't resist ;)

  58. I think you're forgetting that a lot of KDE devs are paid for their work on KDE. I'm getting a bit tired of the 'heroic unpaid volunteer dev' myth in FOSS, when in fact many devs are paid. I'm also tired of the attitude that 'unpaid volunteer' means 'it is OK to crank out shoddy work and leave a mess for someone else to clean up.' All professions have their share of volunteer contributors: astronomy, paleontology, medicine, all kinds of community services, and so on, and overall they are expected to behave professionally and competently. FOSS should not be any different.

    You want ideas on how to make KDE4 better? Quit dismissing the many KDE3 users who are unhappy with KDE4 and listen to them. KDE3 gave a wonderful user experience, and KDE4 threw it all away, and then the KDE4 devs act all puzzled at all the ensuing unhappiness. Duh! Sheesh, all the information in the world is out there, if you would quit trying so hard to resist it.

  59. For all these kde 3 users, what is better with 3 that its soo good compared to 4?

  60. read this, then you see what gnome users think about gnome http://www.omgubuntu.co.uk/2010/06/i-know-im-in-no-position-to-criticise.html

  61. obviously that guy is using the wrong DE, perhaps we could trade him for some akonadi/nepomuk haters? :P

  62. Just recovered from a crash of the konqueror embedded in news in kontact .... (Text is lost..)

    Just my 2cents

    The ONLY aim of the development of the KDE-destkop has to be the improvement of the user experience and NOTTHING ELSE.
    I do not care about the underlying technology, because I am a user

    (1) Despite of this statement I am heavily disappointed about the PIM software development.

    PIM software was the main thing for me using KDE and now I got kaddressbook in a very dirty state in 4.4. So I got unwanted
    - an additional server
    - an additional database
    - a very poor user interface
    - a large backstep in usuability

    AND I am much more afraid for kmail in 4.5

    (2) Arkonadi and nepomuk left my desktop on my old computer in an unusable state so I had to switch it off.
    With my old hardware set up I was unable even to use my old TV-card without distorsions.

    Of corse this is a rant. But as a linux user since 0.99.13p3 I followed the development carefully and
    have to mention that if the Joe Doe is not more the main aim, something is wrong.
    So better to get things ready when they are ready then to believe that technology jumps are interessting for the end user, THEY ARE NOT!!!

    I know that many talented developers are working in their spare time on KDE. BUT the project takes the wrong road when the END USER EXPERIENCE is not in the center of interest.

    Greetings Dieter

  63. This is obviously not a new observation, but the last few comments seem to point at the inherent difficulties of Linux's production model. KDE produces software, and it releases it early in the development process so that the feedback and maturing stage is sped up. However, it is the distributions who are in charge of packaging and delivering KDE's software to the final user. Despite KDE's warnings that some software might not be ready for average users, and sometimes because of the lack of such warnings, many final users end up with systems that give them more problems than they are willing to endure.

    In its main webside, KDE says that "It is our hope and continued ambition that the KDE team will bring open, reliable, stable and monopoly-free computing to the everyday user." So we say that we want our software to be reliable and stable, yet the way things are done now KDE often has no control over what the final user experience is going to be, and the only way to achieve it - withholding the release of software until everything is worked out fully - would go against all the principles of free and open source software production. What is to be done, then?

    I think that we need to do one of two things. First, we could re-think our goals. We could acknowledge the inherent difficulties that our development process has for mass average end-user adoption, and be ok with targeting a smaller audience of relatively computer-savvy users. Second, we could re-think our approach to delivering software to users, particularly the steps that go between our products and the end user. There would be a range of options here:

    1) Promote/sponsor/collaborate closely with one or more KDE-centered distributions that are capable and willing to do the filtering and to put in the extra work that is needed to ensure a (mostly) problem-free and consistent use of KDE for average users. This implies important challenges in resource mobilization, but it is definitely an option that could be explored (e.g. would it be possible to convince Nokia or any other large corporation to devote funds to such type of distribution?).

    2) Exercise more control over what is being released officially. There could be clearer and better enforced guidelines, and a team that worked as gatekeeper in close collaboration with the promo team and with the distributions. It is not clear, however, how successful this would be given the way FOSS development works.

    3) Put more attention to usability features, maybe even having a team that can work as a consultant to developers. This would be less of an 'enforcer' team as that of option 2, but would be closer to the developer side to encourage usability-friendly consistent habits among developers before releases.

    4) Other ideas?

    I am not sure that this is possible, and if it is it would require community-wide input and support, but from what other people have been saying it does look like we have to either re-calibrate our goals about who we can reach or we have to re-think the way we get our products to the final user.

    My 2 cents anyway.

  64. Good analysis, Oriol. 1 has been tried, distro's couldn't manage it from a business point of view. 2 would hurt our community and 3 has been tried (the appeal project). so we need a 4: a separate team doing a second SC, for example. That might just work... In time they might get some funding etc too.

  65. - Single click madness.
    KDE is the only desktop I'm aware of that has "single click" that does not select the file but "runs" it instead.
    I hate it, but this is not the point. The point is that one of the basic and most used interaction has simply an unexpected behaviour. If you complain, probably developers will tell you that you can change setting (as I did). But the default is deeply WRONG (big damage to users with no gain. Also many people has to use M$crap daily, so this makes things even harder for them).
    - beauty instead of usable. With my Oxygen theme (default) I've hard times very often in quickly find the menu of the active window. Just having the title border with a colour different than grey would help... old style? Yes, but effective
    - Pager: just a red frame around the miniature of the active desktop would help a lot in understanding what is the current active one

    Just some examples, and many will tell me that I'm wrong, but with common sense I'm sure I'm damn right.
    I also would suggest developers to read Alan Cooper books, and learn about "goad directed design". Usability is not UI fancy, is understand your user's needs and mental model.
    Last but not least... amarok volume control!!!! How can you sanely manipulate a rotating small gizmo like that with the mouse? How on earth can such a thing be considered by someone "good design"?
    Programmers don't have to try to be interaction design expert, as Cooper says.
    Anyway I thank a lot KDE developers for their efforts, I do love KDE and I'm upset only because could be much better just refocusing on better design.

  66. I guess since Gnome lacks as many, or more, options than M$ Windows, then it must be usable like Windows. I've not found one thing in Gnome that is more usable than its counterpart in KDE. KDE has a number of problems that need to be addressed, Akonadi errors that give the user no idea about what they mean for one, but that doesn't make KDE less usable. Just because KDE has some things that don't work correctly, doesn't mean its less usable than software that doesn't even offer to do it in the first place. How does "it only does this" make something more usable? Gnome's foundation is The GIMP and I don't think there has ever been a worse program for usability made than The GIMP. The developers of Gnome remind me of the developers of M$ Windows and The GIMP with their attitude of "We don't need to give the user any options because we know exactly what they want to do." No thanks. Please keep KDE the way it is.

  67. Someone should hire Joseph Cheek, who would be ideal to tackle these usability issues !

    Back in the early Redmond Linux days Joe created a whole swag of KDE2.x patches that standardised the look and feel across the various applications.

  68. ssssshhhh... another KDE vs GNOME flame war. Use the one you like and stop bitching about the other.

    Systemsettings --> Keyboard & Mouse --> Mouse -- Icon
    On that panel click either single or double mouse action.
    And, learn to use a desktop before you start bitching about it or spreading disinformation.

    In stating that Linux has only a 1% share of the desktop market you repeat FUD created by NetApplications, a windows executable rebrander, and sock puppet for MS. IF you want to quote MS sources about Linux desktop market share then quote Steve Ballmer who, in a speech last year, said that Linux was a greater threat than Apple and displayed a graphic showing Linux with about 12% market share, slightly greater than Apple's.

  69. Aldo "xoen" GiambellucaSunday, 20 June, 2010

    Last day I've seen this document (findings from usability testing on empathy) and it's impressive : http://design.canonical.com/the-research

    I think we need more of this things.

  70. @GreyGeek77
    You've completely missed my point, nor read carefully what I wrote (I wrote that I did change the default behaviour: "If you complain, probably developers will tell you that you can change setting (as I did)").
    So funny that you wanted to contradict me confirming what I complained about, lol

  71. @markit, regarding single-click:

    Nowadays many desktop users spend just as much, or even more, time on the web as they spend with actual desktop apps. And on the web, pretty much everything is single-click.

    So people *are* actually familiar with both double-click and single-click schemes.

    So, why shouldn't KDE be allowed to chose whatever scheme it considers most appropriate?
    Why should it's file manager stick to the old-fashioned double-click behavior, just because other desktop environments do it?

    New users switching from other environments will maybe accidentally open one or two files by accident while first getting familiar with KDE, so what? They'll learn it quickly, and in the long run they'll either recognize the superiority of single-click or they'll change it in the settings.

    Me personally, after having used single-click KDE for some time now, I actually find it very unpleasant whenever I have to work on a system like Ms Windows that forces you to do tedious double-clicks all the time. There *is* a big difference, and in my opinion it's strongly in favor of KDE's default.

    Single-click-to-open just feels more natural, easier and more modern, doesn't need any extra mental focus (think trying to hit the right double-click speed), and of course it also prevents RSI[1].

    [1] http://en.wikipedia.org/wiki/Repetitive_strain_injury

  72. force the developers to watch users using their programs

  73. KDE should be more modular, so you could install only features you need.

  74. Well,

    KDE4 is a revolution. As any radical change, it has its costs.
    If I was in KDE's devs team, I would look first for performance and stability issues.
    Any DE has its target. I have used MacOSX and I was very happy with its interface. Nothing gets in the way, all comes naturally, there is not much to setup, it learns. It is good for users not related to IT. It is built for productivity.
    Gnome is also simple and stable. It gives the solid rock feeling.
    KDE is for geeks. For people loving customization. They need know-how. It's like Linux itself. Pluggable and dynamic. To me it is like Microsoft's products. If the developers of gadgets, for example, follow 'the rules', anything goes fine. Otherwise... A quality audit process should be done first.
    I would take the good features from others and implement them. I would set polls for knowing the needs.
    Sometime, IMHO, KDE4 goes too far.
    To have shortcuts and applets implementing shortcuts, for example, looks crazy to me. Better to have it than missing it, yet...
    Anyway, strongly customized or not, KDE4 should be more intuitive and user friendly. How? Just stay and think, try to feel what's wrong and act accordingly. Sometimes, when you know it is wrong something, don't need much to get into the awareness state and know what's wrong.
    For any design - no matter of taste - harmony, consistency, gives it the 'I like it' feeling.


Say something smart and be polite please!