Why GNOME Hackers Should Care About Usability
SethNickell, August 2001
Usable Us"a*ble, a.
Capable of being used.
For such a simplistic definition, this encapsulates the fundamental goal of usability very well. Usable software is software that people can use; whether to write Email, play games or develop the next killer application. GNOME is many different things, but certainely one of its significant aspects is to provide an environment for users - for people. What a disappointment it should be when a user's ability to access one of the features we have coded is impaired or altogether halted because they don't understand how to manipulate the interface.
A segmentation fault is not the only total failure mechanism.
What is it that makes a segmentation fault (or a thread lock, or a failure to activate a central component or any other manner of critical programming errors) a serious problem? Was the computer hurt? Probably not. Was the operating system destablized? Hopefully not, that's one of the primary motivations behind executing processes in seperate memory segments. Some of the characteristics that, I believe, capture why we consider segmentation faults to be serious are:
- The event is abrupt and unexpected (it is a surprise)
- The user is given no opportunity to recover their data
- The user's work flow has been completely interrupted
- Often, no explanation is given as to why the event occurred or how to work around it
Usability issues can generate all these problems, even with perfectly written code. And sadly, they often do in GNOME applications. There are many other classes of interface problems, as serious as a segmentation fault, that a segmentation fault does not produce (for example a confusing metaphor that leaves the user unsure how to proceed), but this list gives us some major ones to think about for now. These total failures are just as serious when they occur in an interface as when they occur in code.
Lets look at a seemingly innocuous UI problem that produces some of the same problems as a segmentation fault. There are a variety of different ways GNOME applications handle save and exit confirmation dialogues.
The "Standard" Save Dialogue
Given that a person were used to seeing the standard save dialogue above pop up when they exit a program without saving, how do you suppose a user in a hurry might react to the following dialogue?
Dialogue based (probably) on the XEmacs/Emacs saving conventions
Some users will inadvertantly click on the "Yes" button even though they want to save their document, and will hence lose work. Hopefully the user will figure out what happened and learn to be cautious of the program bringing up the latter dialogue and will not repeat this "mistake", but other than this would it have been much worse for the user if the application had simply segfaulted?
Even supposing that both dialogue conventions were equally good (the latter has some serious problems), it is dangerous to have a dialogue that defies the users expectations, even if the dialogue itself is clearly worded. People do not always read all text, they expect things to behave in certain ways. Interface is a language, and our users have built up a vocabulary and grammar.
Here's another question, what do you expect to happen if you click "Yes" in the second dialogue? OK, that seems clear. What about "No"? Will it offer to save, or does "No" mean the same thing as "Cancel" in the first dialogue? Then what about "Cancel" in the 2nd dialogue, what will it do? Perhaps we can figure out the behavior, but we shouldn't have to. The problem is that the "No" most directly says "No, do not quit without saving", but doesn't make it clear whether we are saying "No do not quit" or "Quit, but please save". This perhaps won't cause a serious problem for most users (since both "No" and "Cancel" will hopefully be non-destructive), but it is awkward and wastes people's time. This dialogue is not comfortable to use.
Just for fun here are two more "exit dialogues", interestly from the same application depending on whether you choose to close the toolbox window or the document window:
Dialogues are just one tiny example of how usability, or lack thereof, can affect people's abilities to use our software effectively. Often applications are rife with much more profound problems, but problems of any complexity cannot be easily presented in a short article like this one. Any time a user needs to ask for help to make progress, or commits an interface blunder due to a non-standard interface the application developer has failed to provide functionality. The program is buggy, as much as it would be buggy if it was filled with unitialized variables. You never know when or how the interface will cause a problem. The only garuntee is that eventually it will, maybe even frequently.
There's always a certain temptation to dismiss anyone who doesn't "show youthe code". It is true that often times people who fail to produce substantialcode (and, for example, offer to help with architecture ) are of little use to projects. Sometimes it seems that hackers perceive usability engineers in the same light as they might Lord Halifax in the following story:
The most important critic in our field of study is Lord Halifax. A most individual judge of poetry, he once invited Alexandar Pope round to give a public reading of his latest poem. Pope, the leading poet of his day, was greatly surprised when Lord Halifax stopped him four or five times and said, "I beg your pardon, Mr. Pope, but there is something in that passage that does not quite please me." Pope was rendered speechless, as this fine critic suggested sizable and unwise emendations to his latest masterpiece. "Be so good as to mark the place and consider at your leisure. I'm sure you can give it a better turn." After the reading, a good friend of Lord Halifax, a certain Dr. Garth, took the stunned Pope to one side. "There is no need to touch the lines," he said. "All you need do is leave them just as they are, call on Lord Halifax two or three months hence, thank him for his kind observation on those passages, and then read them to him as altered. I have known him much longer than you have, and will be answerable for the event." Pope took his advice, called on Lord Halifax and read the poem exactly as it was before. His unique critical faculties had lost none of their edge. "Ay", he commented, "now they are perfectly right. Nothing can be better."
Stephen Pile, "The Book of Heroic Failures"
A lot of hackers and "advanced users" seem to have the misperception that usability is about dumbing interfaces down, and is hence diametrically opposed to their use of the software. This is completely untrue. It is true that sometimes a decision will need to be made the favours either advanced users, or novice users, but this is the exception not the rule. Most usability improvements are helpful to everyone. We are all novice users the first time we use a new application. You need to remember that as the developer of an application or interface you are the most expert user of it, not the average. Every new user of the program goes through a learning curve from the time their first launch the program until they are proficient with it. Look at the following screenshot of "GNOME DiskFree" and its entry in the Panel "Foot" Menu.
What's wrong with this picture? The tooltip and the name of the application itself suggests that GDiskFree will be presenting the free space available to the user. The interface of GDiskFree itself has no labels or captions to contradict this, and prominently displays a dial and a percentage value. Based on a lack of contrary data, many (perhaps most) users immediately conjugate the available information and mentally view the value as the "percentage free space". Users with some knowledge of their disk usage may quickly realize this is wrong, and perhaps even realize that the application is displaying the used space rather than the free space. Ironically, because of this single simple failing GDiskFree's command-line kindred "df" is more usable (df's column titles save it). This is in spite of df displaying the information in a less human readable format and including more technical information.
Also troubling in GDiskFree is the lack of details as to what the various other information mean. It isn't too hard for an experienced user to infer that "19.1 Gb" probably refers to the total size of the partition, "/" refers to its mount point and "/dev/hda1" refers to the device... but it would be much better to caption these. The GDiskFree tooltip also makes mention of an extraneous implementation detail, but that is of less significance. It is amusing that when I had an experienced GNOME user and developer review this article, they commented "Oh, I didn't know that's what GDiskFree was doing, I knew the values were wrong and I thought it was broken". The entire purpose of GDiskFree has been subverted by a trivial to fix usability bug. Once again, I'd like to emphasize that GNOME is full of brokenness like this, I'm not trying to pick on any particular application.
Usability also extends far past "first time experience". Often usability is about streamlining programs, removing unnecessary keystrokes and mouse clicks, providing a more accessible interface to a commonly used feature, perhaps burying a feature that gets in most people's way. Yes, this will often cause pain to some user of the program, and at some point you will be in the subset that is afflicted by a decision made to optimize for the majority. But the same thing is true of API development. To keep interfaces clean and usable by the majority of programmers we have to pare them down to a certain size. That means sometimes optimizing for the majority at the expense of some subset of API users. We can try to minimize the problems it causes, but most people accept that this tension is inevitable. As we make your software useful for a greater number of people, there will sometimes be a price to pay.
More positively though, I am convinced that GNOME is not the epitomy of a perfect advanced user's / hacker's desktop environment. We like the flexibility, but there are many times that the GNOME environment is awkward, and even some ways in which it is extremely inflexible. The Usability Project is trying to concentrate on issues that will make it a better engineer's desktop, because we believe that in most cases improvements made for the hacker community will benefit everyone. Everyone saves when a needlessly cluttered interface is straightened out, or a much used feature is made more accessible. Everyone saves when the menu system is navigable, or icons are clearly recognizable. Everyone saves when they can expect pressing the "Enter" key to have a consistent effect in all dialogues (what effect that will be is still in contention, but its more important that there be a convention than that a particular convention is chosen).
Other than as an academic or exploratory excercise, software needs to be measured by the benefits it brings to the people who choose to access it. No matter how potentially useful a program is, only when the user succeeds in accessing this functionality does it have merit. In fact, in the case that a user fails to discover or employ basic functionality...as far as the user is concerned the code may as well have not been written. That is why you should care about usability. Usability is the key that unlocks the abilities of your program to users.
User Us"er, n.
One who uses.
Most of us probably want people to operate our programs, we desire "users". Why this is desirable is out of the scope of this document, but to suffice it to say that most hackers find gaining a userbase to be a motivating factor in developing applications. Certainly we hack some things out just for ourselves, but with most projects into which we invest a significant amount of time, we envision the library or program being used by a number of end-users or other developers.
How do we attract users? Users will be drawn to software that helps them do what they want to do, software that does not confound but assists, whether that be for entertainment, creativity, or office work. Development needs to be driven by users, driven by usability. What do people want out of your application? How can you make it more convenient for them to extract this functionality?
Usability people are a resource that help you to attract more users by making your carefully constructed features accessible to a wider audience, and more convenient for your existing audience. We would be very happy to work with you to make your applications better! If youare already an experience GNOME developer we'd be very happy to help design an interface for your projects from their very inception. Some usability issues, particularly interaction problems, are much easier to fix when they are considered before coding. It is the difference between choosing a good architecture and just poking around broken code to make it half-heartedly function. Try visiting #usability on irc.gnome.org or sending a message to email@example.com.