|
|||
Beige is Beautiful Bondi Blue is bad, but Beige is Beautiful. This be sacrilege in certain circles, but it epitomizes an important characteristic of tools and interfaces. A tool is a device which permits us to accomplish something we can't otherwise do, or to more efficiently accomplish something which would otherwise be more difficult. All tools have function and interface. The function is set of capabilities that the tool provides us, and the interface is the means by which we manipulate the tool. Both of these are important, but the interface is more important. A tool of limited capability with an excellent interface will get used. A powerful tool whose interface is horrible probably won't be, unless there's no choice. An interface has to be learned and then used. "Ease of learning" refers to the effort involved in becoming proficient with the tool to accomplish some end; "ease of use" refers to the effort involved in using the tool after you've learned to use it. Of the two, ease of use is more important. Again, a tool which is hard to learn but easy to use will find more favor than one which is easy to learn but hard to use. The single most important characteristic of ease-of-use is what I call invisibility or submergence. (If this has a formal name I've never heard of it.) It's non-obvious; indeed, by its nature it is unobtrusive when done well. It refers to the extent that the tool ceases to be external and becomes part of its user. This is not easy to directly describe, so I'll try to give examples instead. When someone is experienced driving a car, the car vanishes. When you are driving, the car becomes an extension of your body. You want to go forward, but you don't think "push on gas pedal". You want to go in a certain direction, but you don't think "turn steering wheel". When you're learning to drive, you think those things, but the process of learning to drive consists of transforming such conscious thoughts into something at a lower level, and once you are using the car, it's not really there anymore. Rather than driving a car, you transform yourself into a car. The nearly ideal tool is a pair of pants. It takes no time at all to learn, and use is equally easy: you just wear it. It conforms to your body, goes where you go, does what you do, and doesn't require you to operate it. The pockets hold objects and they are perfectly placed to be reached by our hands. The only time they get in the way is in the bathroom, and they are easily removed. 95% of the time you don't even think about the pants you're wearing (although you'd certainly be thinking about it if they weren't there). A carpenter wears a belt with tools hanging from it. One of those tools is his hammer. When he needs to pound nails, he performs a very rapid operation which converts his hand into a hammer; after which he pounds the nails, then performs an operation to convert the hammer back into a hand. The first operation is to reach down and get the hammer off the belt, the second to return it to the loop. But while a carpenter is using the hammer, he isn't thinking "hammer". He's thinking "pound nails". (Actually, he may be thinking "build house".) The characteristic of invisibility in a tool is the extent to which we are able to abstract it and forget that it exists. The better a tool is at this, the easier it will be to use. Which is why decoration on a tool has to be used very carefully. It's not directly bad except when it conflicts with invisibility. If the decorations on a tool have the effect of forcing the tool into the attention of its user, then it prevents the tool from becoming invisible. But carefully used decoration can have the effect of making the tool more esthetically pleasing without being intrusive. Like everything in engineering, form must always follow function. Where function doesn't constrain form, there is the opportunity for esthetics. But when esthetics and function conflict, a good engineer always favors function. (An engineer is someone who designs tools. That's what engineering is.) A computer is the most versatile tool ever invented, because it is a generalized tool for processing information, and nearly everything we do involves information. So because so many people spend so much time using computers, there has been extensive work and experimentation on their user interfaces, and much of that has been misguided. Here's a sin I've never seen committed, though it could happen: suppose that every icon on the desktop was animated and was constantly moving. This would be supremely annoying. That's because the one thing we are most sensitive to in peripheral vision is change. Any kind of alteration in peripheral vision makes us want to look at it. (It might be an attacking animal or some other kind of danger. It might be food. It might be anything, but the point is that it is a change and we need to know what has changed so that we can decide how to deal with it.) Animated icons on the desktop would constantly distract us, forcing us to pay attention to the UI itself instead of to solving the real problem for which the computer is our preferred tool. So the only time that animation should be used in a GUI is when it really does need to attract our attention. Animation for the sake of animation is bad. Too often, form dictates function. There are designers who forget that the goal of their design is to let their users get a job done, and instead get fascinated by the artistic vistas open to them. The result is a usually superb museum piece, but a mediocre tool. Among computer companies, the one which has traditionally committed this sin most consistently has been Apple, starting with the Macintosh. One of the reasons for this is that the engineers there have been optimizing ease of learning rather than ease of use, and as such have tried to minimize the number of controls and input mechanisms. (Another is a corporate cultural fascination with being "insanely great".) At a time when three buttons on a mouse was the industry standard, Apple insisted on only having one. At a time when every computer with a GUI also had a command line interface, the Mac did not, and in many cases even the tools didn't have places where text commands could be entered. Apple's engineers and designers have also seemed to be obsessed with art and "coolness" even when this conflicted with usability. The fundamental mistake is to confuse ease of learning with ease of use. Fewer mechanisms are easier to learn, but they are not necessarily easier to use. A keyboard would be a lot easier to learn if it had 8 buttons (one for each finger) instead of a hundred, but no-one designs keyboards that way. Equally, it's important to have a sufficiently rich set of controls to permit efficient use. I think that they have now made that same mistake with the Aqua interface of OS/X. The problem with it is that it is too beautiful. It looks stunning; it makes a great demo; it attracts the eye in a store -- and it will intrude, constantly disturbing that all-essential submergence which is the mark of a truly good use experience. I don't think that a user of Aqua will ever forget it's there -- and if that is true, then it fails. For example, generally speaking interactive transformation of the screen is bad when not in direct response to a human's request for a change. A GUI works because it interacts with the sections of our brain which normally deal with physical objects within arm's reach. But objects in our field of view don't transform simply because we're looking at them; so having that happen in a GUI is a disrupting influence and breaks that reverie which contributes to submergence. This is one of the reasons why "scrubbing the dock" is bad; it would be much better if the dock always presented enough information rather than forcing the user to seek it with the mouse pointer. Bondi blue is bad. For desktop computers the design of the case doesn't affect submergence, but "Bondi Blue" epitomizes interfaces which prize beauty and glitz over submergence. Beige is beautiful. A "beige" interface is unobtrusive and invisible. Submergence is to be prized in a user interface over all other things. The mark of all good engineering is restraint. This page has been viewed 4943 times since 20010726. |
|||
