Linux on the desktop: GNOME Usability

Sun has a very nice study on GNOME Usability up at http://developer.gnome.org/projects/gup/ut1_report/

It demonstrates, rather nicely, that GNOME still has a very long way to go. Many of the issues are so glaring, I'm amazed they were allowed to slip through. I find myself constantly realizing that if I were designing the UI, I simply would have done the right thing without even giving it any thought. And I don't pretend to be an expert

Actually that can be a detriment. They take many preconceptions with them and use those preconceptions to color their usage. I have found that people that have never used a computer before find gnome and kde easier then those coming from a windows or mac background. Overall many of the fixes they suggested I disagree with.

Actually you see this in many parts of society. I have seen C programmers try to use the knowledge they know of optimizing stuff in C to optimize python programs and find some of those speed tricks in C will kill python's

As a amateur user you are actually the best UI designer there is. When you are a technical user is is difficult to cater sufficiently for non technical users as you are not aware of all the issues. This is exactly why Sun undertook this study and why Gnome is listening. It's great software design and adjustment at work right before your eyes and something to be celebrated rather than derided.

So, you expect GNOME to enter the world complete & perfect with no need of any further modification? I think that this study demonstrates that GNOME developers have enough confidence in their product to subject it to this kind of evaluation and then to publish the results. Can you really see MS engaging publicly in this kind of exercise?

There are some things I find difficult. Whether that is at the sort of level like driving a car where the indicator switch is on the other side of the steering wheel from what I am used to, I cannot yet tell.

My main issue with GNOME is the organisation of the menus... on the version I use everything appears in several places and it's difficult to remember when exploring how things whether you've been here before by another route.

Overall, I give it 8/10 and I am sure that given a few iterations they will make it better and better. I have faith!

[...]

: I haven't finished reading yet, but so far, it's quite a fun read, if : you're interested in UI design or the state of the open source desktop. : I especially recommend this document to those who disagree with my : assessment that Linux is a 'cobbled-together OS' (see especially the : section on customization tasks), or those who are constantly : overestimating the abilities of average users.

[...]

Kudos on posting from csma to csma and cola. A lot of people gripe about an OS 'from afar' without bringing in that OS's advocacy. I think bringing in cola shows an open mind.

I enjoyed the article too, though when you said 'cobbled-together OS' in csma earlier, I wasn't thinking of Gnome. I think a UNIX user tends to think of the OS as something that provides lower-level services and lower-level utilities. IMO Gnome is a ways up in the heirarchy, and is just one choice of UNIX GUI. It especially important to note that Sun may not actually hae been running Linux in this test. They are actually backing Gnone as a future desktop for _Solaris_. (Perhaps this study makes Solaris a 'cobbled-together OS' as well ...

Anyway, the GUI is pretty seperable from what I would call the OS. When I get concerned about this or that choice in Gnome, it is nice to know that other people are out there working in parallel. They may ultimately provide better GUI.

Getting to the results themselves, I found them very interesting. On one level they didn't tell me anything totally new (perhaps because I came up through the old Macintosh UI guidelines, etc.), but on the other they showed me how much 'UNIX-ish-ness' I've let creep into my worldview. In many of those tests, non-technical users are confused by the same clues that would be clear to me. It strikes me that Gnome (etc.) are trying to serve two consituencies: Currently a good fraction of UNIX/Linux users are looking for those UNIX-ish clues. On the other hand, visitors from Mac/Windows are going to be confused. To drop back to a less technical description would help the new users, but might (in some cases) confuse old UNIX users. I think the answer is to use 'user levels' to reveal more or less of the UNIX underpinnings as appropriate. More 'expert' users should see more 'UNIX-ish' prompts.

(There is much more of interest in this study, and this just hits the top of it ...)

Finally, I think it is great that this can take place in the public forum. You know, the new Mac OS X UI has some things (red/yellow/green buttons) that might not fare so well in the 'don't move your mouse and tell us what they do' test. Some prominent GUI designers, including some ex-Macintosh designers have expressed concern about these and other OS X UI elements.

Wouldn't it be nice if Apple, or one of its developement partners, were to do the same kind of open review of the UI ... just to make sure all the ducks are in a row ...

[gnome study]

: Do people prefer consistency or individuality? Have a look around at the : programs today and tell me the trend. Certain functions have to be : consistent, but there is a fine like between consistency and lack of : customisation.

This study is focused on walk-up usability, especially by non-UNIX users.

Walk-up usability is important (especially for a platform desiring users), but we meet a platform only once. Every time after that we are returning to it. There are different sets of conditions when you consider users who return occasionally, or very frequently. While a first response is interesting, it might also be interesting to know how well early UI learning is retained.

In my experience, the Mac had a real strength in its UI guidelines, and in the fierce(*) commitement users and developers had to those guidelines. The Mac was easy to learn, and easy to apply that learning to new situations.

Once you are past your early learning of a UI, the real benefit of consistency is a lowered learning curve for new applications. Again, in my experience, it is easier to feel 'at home' with a Mac application you've never seen before than with applications on other platforms.

Once you get out a few years on a platform, when you've learned the basic UI, and the UIs of your most-used applications, I think it is a new game again. Consistency is most important in the beginning, but I think customization becomes more important with time. You *know* what you want to do, but you may be frustrated at the number of steps it takes to do it.

Mac and Windows have systems allowed fairly agressive changes to the UI, through drag-and-drop, or through more in-depth theming. I think Apple and Microsoft have both become more resistant to this, for two reasons: First, there are usability issues. A new user may customize his system faster than he learns it, creating more trouble and tech-support calls than the vendor would like. Second, the basic 'look' of the UI has become a big part of the branding for Mac or Windows. They are very up-front about protecting this UI as 'trademark' or 'trade dress'. A trademark has a purpose. It is a unique look or image that is used to sell a product. I think they would like a Mac or Windows to look like a Mac or Windows. They want each screen to help sell the product.

I hope the open GUIs will guide new users toward a consistent experience, while allowing experienced users more customizability. I hope branding and trade dress won't become too strong in the open source desktop!

* - maybe sometime too fierce, and resistant to change.

Well, I'm not an amateur user. I'm an amateur UI critic. It looks like the GNOME people know even less about the subject. Apparently, being an amateur with UI design doesn't help <g>.

I didn't come to the same conclusion as you. If you read further into the document, the participants had quite a bit of praise for GNOME. In addition, the complaints uncovered did not at all doom the entire UI. I think if any UI underwent this sort of scrutiny, they would all flop to some degree. It's just that we don't see manufacturer's publish their useability studies for the public's eyes. Frankly, I don't think they have the guts (or maybe the stupidity), because they aren't well designed enough.

Michael Adams

I don't agree that we meet a platform only once. A platform of sufficient complexity, such as Linux will involve a long learning curve over months if not years to learn all the aspects of it. That learning curve is diminished if most things work similarly, instead of having to figure out a totally different way to learn each individual piece.

Yes, and then when we add new things (say we upgrade the GUI) things change yet again.

There is such a thing as muscle memory, which is when your fingers or arm or wrist automatically do something without even thinking about it. That's where consistency comes in long after you've become acustomed to things.

Microsoft has taken on the steps to allow more customization in Windows XP, however you must sign an agreement with them to be given the information to do this. The reason, is that they want to keep the UI consistent, even while allowing it to be consmetically modified.

: > : > Walk-up usability is important (especially for a platform desiring : > users), but we meet a platform only once. Every time after that we : > are returning to it. There are different sets of conditions when you : > consider users who return occasionally, or very frequently. While a : > first response is interesting, it might also be interesting to know : > how well early UI learning is retained.

: I don't agree that we meet a platform only once. A platform of : sufficient complexity, such as Linux will involve a long learning curve : over months if not years to learn all the aspects of it. That learning : curve is diminished if most things work similarly, instead of having to : figure out a totally different way to learn each individual piece.

I agree with what you are saying about discovering new elements.

I'm just trying to distinguish between the first exposure to a specific element and repeat visits. Sure, you might not decide to change your background until some months in. I'd call that a first-time experience.

As an example of the other thing:

There is a lot of discussion about how intuitive common controls on windows should look (minimize, grow, go-away, etc.). These are things we really do see on the first day we use a UI and (without theme changes) see ever day from there on out. Some people think that very stylized controls (like Apple's red, yellow, and green buttons) work because: 1) once you learn the representation, it will stick with you, and 2) many controls are identified by position. I've heard other people argue that such stylized buttons will slow people down even after learning ... as the mouse closes on the position it will tke more thought to confirm that 'yellow' really does what you want.

This is the kind of thing a study a study of 'second week' or 'second year' efficiency can help resolve.

I've tried using some stylized icons on my Gnome Panel, but I've noticed that even with just three (foot, apple, bug) and very clear spacing, I sometimes slow down to make sure that the apple means what I think it does. FWIW, I've switched back to icons with more correlation to what they represent.

Perhaps Gnome is NOT intended for the MS user that knows Little of NO knowage of computers? Funny how comments about ONE of MANY desktops available for Linux becomes a blanket statement about 'Linux on the desktop'.

NO, it is the state of GNOME. NOT the state of KDE or ANY OTHER desktop available for Linux. Thus it is NOT about the state of Linux as a whole.

Again, this is about GNOME. Gnome IS NOT the say all of desktops.

Keep in mind this is a study on Gnome which is only ONE of many desktops available for Linux. As far as I can tell, Gnome is not even the most used Desktop. It is totally unreasonable to studies on Gnome and pally them to ALL of the Linux desktops much less apply a single study on Gnome to the state of Linux as a whole.

Besides, how would Sun's desktops stand up to the same study? CDE? That would be a good laugh! Openwindows? Right that will stand up to Gnomes results!

Another point to keep in mind: Not all Linux desktops need to be aimed at computer illiterate. I know what I am doing and do NOT want to be having to open a bunch of advance options boxes. If I open an IP interface configuration box, show me ALL the possible options so I don't have to open yet another box. I do NOT need to have higher functions isolated into advanced sections and do NOT want that 'functionality'. ]

Rather than making ALL Linux desktops as dumb as Windows, a better solution would be to have a couple tailored to cater to the computer illiterate while others are tailored for advanced users even some that are tailored for servers!