AskTog: Interaction Design Solutions for the Real World
Interaction Design Section   Living Section   About Bruce Tognazzini
NN/g Home > AskTog > Columns > OS X: A First Look Ask Tog, February, 2000

OS X: A First Look

Apple appears on course to release a great demo. Like all great demos, OS X has been stripped of all those extraneous details that detract from initial reaction, extraneous details like the names of the files you are working on. While I applaud the designers for what they have done to date--much of which is very, very good--their work, if OS X is to be more than a swell demo, is far from over. Right now, OS X, as a whole, is frighteningly reminiscent of New Coke, painstakingly developed to appeal to an upcoming younger generation, but destined to become the greatest marketing disaster of the Twentieth Century.

New Coke blew up in Coca-Cola’s face because they didn’t differentiate between the demo and the product. Pepsi had been killing Coke in county-fair taste tests because Pepsi was sweeter and less complex. Coca-cola whipped up a sweeter, less complex version of Coke, then tried it out in similar taste tests. What Coca-Cola failed to recognize was that people’s reactions to the taste of a tiny sip at a tasting booth did not translate into these same people’s experience when drinking 12 or 16 ounces of the same product. When Coca-Cola released their sweeter, less-complex product, people shied away. When these same people learned that Coca-Cola had ceased production of the old beverage they held so dear, they revolted. Coca-Cola was totally unprpared for the reaction of loyal Coca-Cola drinkers to the apparent desecration of a product their customers held so dear.

The Apple story is playing out a little differently, but, so far, not that much. OS X is not being taste-tested on a variety of users; it is being taste-tested on a single user: Steve Jobs. Like those county-fair tasters, however, he is not drinking deeply of the interface, but merely reacting to the initial fizz. Were it to be released today in its present form, Apple might well experience what Coke did as customers switched from the old Coke, but to Pepsi. Fortunately, we’re a year away from release, allowing much time to build upon the fine work already done.

The Fizz

OS X is beautiful. If you haven’t seen it yet, take a look. The new aqua appearance is cool and clean, in sharp contrast to the ponderously-heavy 3D chrome look that Microsoft visited on the world and Macintosh quickly emulated.

OS X features new, semitransparent dialog boxes, directly associated with the windows that generated them, ending the current confusion of untitled dialogs with ambiguous wording “Save changes?” suddenly appearing in the middle of the screen. The semi-transparency is slick and, while it does knock down readability, it should be OK for those dialogs stable content. Dialogs which impart unique information, such as “You have elected to save this file with the name FOOT8ALL; was that your intention?” could be problematic. Will users notice the substitution of an 8 for a B if it is happens to be set against a vertical stripe? I support transparency, but developers must be able to selectively turn it off, and Apple should provide them with clear, strong guidelines for when to do so. End users should also be able to block transparency altogether, using a preference. I also very much like the idea of the “gumdrop” window controls that, when moused-over, reveal symbols for close, dock, and zoom.

The new elevators, with their liquid blue color, not only are pleasing in appearance, they offer sharp contrast to the elevator shafts. This should help lessen the problem of the Illusion of Completeness, where users assume that the first windowfull of information is all that exists.

The new Single-Window mode, entered by clicking a control where the Close box is on Windows, might be useful for people who become confused by overlapping windows.

Addition of drop shadows to increase sense of depth. (However, Apple should also lower brightness and contrast of rearward windows to really "sell" the illusion.)

The new Quartz imaging, based on PDF, offers many capabilities, including improved WYSIWYG for printing and display-resolution independence. I have some reservations about this path, however, given that PDF has nver really blossomed even into the kind of limited tool that HTML + JavaScript has, and PDF has a propensity for turning 1K documents into 100K documents, making it ill-suited for the Internet.

OpenGL will offer first-class support of games, adding not only fizz, but sizzle.

All this new technology that directly affects the look and feel is coupled with protective memory and preemptive multitasking, ending, once and for all, the nightmare of daily—sometimes hourly—system crashes.

Mouseovers: Just-In-Time Information

OS X makes good use of mouseovers, one of the more popular features of weblications. Mouseovers are a form of just-in-time information delivery. They are useful for reducing screen clutter, while still supplying all necessary information. They must be used sparingly and well.

Principles for mouseovers

  1. Display control symbols unless those symbols are shown repeatedly, are memorable, and are redundantly coded in the constantly-visible objects they adorn.

  2. Show all information necessary to reach a decision at once; do not expect the user to “scrub” large areas of the screen to gain sufficient context.

  3. Do not require the user to play “hide and seek.” You may hide secondary details of an object, but not primary information. Do not hide necessary, but “cluttery” information, such as titles, unless the icon is unique and memorable.

  4. When “lighting up” individual controls with mouseover, do not do so until the user is in position to click that control.

Some have criticized the new gumdrop window controls because the symbols x, +, and - for close, minimize, and maximize can only be seen when you mouse over the controls. Let’s look at how the gumdrops do:

Once you’ve used the gumdrops for more than a couple minutes, you will intuit the meanings of the colors, and the symbols will no longer be necessary. (Those who are severely color blind will have to memorize the buttons’ positions; since those positions are stable, a minor burden.)

When the mouse pointer moves over any one of the objects, the symbols appear in all three objects, offering full context.

It is not clear from the demo whether or not you can click as soon as the controls “light up.” Since the whole group lights up, rule 4 does not strictly apply. Nevertheless, in this case, I would ensure that the clickable region of the buttons matches the “light-up zone.”

Mouseovers increase speed and reduce error, but they can also teach. Apple has made the grayed-out gumdrops in the back windows light up when you mouse over them, enabling you to close or otherwise change a back window without having to first bring it forward. (You could do this in the old system I’m told, but few people, including myself, knew the secret means of doing it.) The roll-over action not only lets people know they are on an active control, it quickly teaches them about a new capability they might never have otherwise realized was possible.

Mouseovers also can harm. When the rules are not followed, usability can be badly compromised. More of this later.

Gumdrops, in depth

Notwithstanding all the praise above, there are problems with the gumdrops as currently embodied.

One objection I've seen raised, I would reject: While the gumdrops do use color coding, the coding is redundant: First controls differ in location and in tone. (The contrasting tones, in fact, were carefully selected just for the sake of the color blind.) Second, once you've seen the symbols and understand how they map to the actions, you will be unlikely to forget which control is connected to which symbol.

At the same time, the coding used is misleading and potentially hazardous. It would appear at first glance that painting the close box red is a good thing, since that warns the user about its potential danger. However, a study done at Apple almost ten years ago found that the user's mouse gravitates toward red objects almost as though they were possessed with magnetism. The study forced us to abandon the idea of making close boxes and the Shut Down option red.

Second, the symbols are apparently inconsistent with the actions. Specifically, we are offered - for dock and + for zoom. However, zoom on Macintosh works both directions. One click makes the window larger and a second click on the same object makes the window small. So, once you've enlarged the window, the symbol for zoom should become a - sign, in keeping with that same symbology on magnifying glasses. However, the minus sign has already found use in the move-to-dock button. So the designers have apparently decided to just ignore the problem and offer a symbol for shrink that, to most of the Western world, can only mean grow. That is tacky design at best and a severer learning and usability problem at worst.

The button sequence is also a problem, in that it clashes badly with both the current layout and with Windows. This will also be a problem.

Another criticism I've seen leveled that I would take exception to is the closeness of the controls. People with experience with Windows know how easy it is to hit the wrong control and accidentally close down the whole application when setting it aside was the goal. Apple has actually provided some pretty decent spacing in these controls, in addition to enlarging their size. I wouldn't expect the kind of error rates from this that people have been predicting, although the original scheme of having "dangerous" on one side of the title bar and "benign" on the other made a lot more sense.

Giant Icons, Small Screens

The first place we encounter the demo quality of OS X is in the use of giant icons in windows. This is purely the stuff of dreams. In the demo, they show one window with 12 wildly different data types, all clustered together. I don’t know about your system, but a given folder on my system rarely has more than one single data type inside it. If every single icon is identical, I don’t need icons the size of small cars. What I do need, in this Windows-dominated world, are long file names. Reader Chris Hanson suggests that OS X will finally have long file names: NextStep has always had them and OS9 supports them within applications, but not in the Finder (go figure). Give me long file names, and giant icons become even less relevant.

Few serious users use icon views at all except for their top level, volume view (if that). Screen real estate is just not available. Why? Because, contrary to Apple’s bald statement that today’s monitors have far higher resolution (“ For 15 years....resolution levels have dramatically increased.”), pixel resolution has remained virtually unchanged over the last twenty years.

Screen sizes have grown. The average Mac screen now has around four or five times as much real estate as the original. Meanwhile, the Average Mac user has gone from having a few dozen documents to several thousand documents. (Just do a document count of your system folder some time if you want to see something scary.)

I applaud the move to 24 bit, dynamic icons and thumbnails. The giant size, however, is, for most users and most applications, naive. It does, however, make for a great demo and, fortunately, the size is apparently a user preference.

The NextStep Browser

Apple has eliminated the Finder. Yes, I know that they are using the Finder label for the reworked NextStep browser, but the Finder we have known and loved for so many years appears dead.

Not all of what they have done is necessarily bad, even if it is new to traditional Mac users. For example, navigating through several layers of nested folders is ugly on the Mac, leaving behind each and every parent window in your wake with no apparent way to close them without closing every other window on the desktop in the process. (Several readers point out that if you Option-click a folder as you traverse down, the clicked folder opens and the parent folder closes behind it. This is a trick I had not learned even after 16 years of working on and using the Mac. Yet another trick involves a click-and-a-half of the mouse, but that only works if you have folder-popping turned on. Folder-popping, having your folders suddenly jump open at you, frightens me. Mine is turned off.)

The new "column view" solves that problem nicely, and should work out well if navigation is what you typically do with the Finder.

Alas, a lot of us who really use our computers are as likely to be manipulating large groups of documents as we are to be searching for a single one. Today’s Finder facilitates that by allowing us to have several Finder windows open simultaneously, with simple, spatially-oriented drag and drop among them.

These windows are not detritus left over from our path through the information space, but windows with contents that we are actively using. If you have a 640 X 480 screen, having mechanisms that enable you to shift from one window to the next may be nice; when you have a large monitor (or three), removing the ability to have several Finder windows open simultaneously is a serious blow.

Of course, there may be ways around this that have not yet been revealed. Steve did talk of a way you could have multiple windows open at once. Depending on how this is accomplished, it could be a good thing or bad.

What the Finder brought to the world was spatially-related storage. Aspects of this spatial orientation include not only the nesting of folders within folders, but being able to recognize a window based on its size, shape, and the layout and coloring of the objects within. If all of these are thrown away, it won't matter how many Finder windows you can have open.

What people need is an easy, visually-apparent way to have windows automatically close behind them as they dive down toward a window of interest, with a way to have several windows of interest open simultaneously. If we are given nothing but a choice of a single window or today's clutter, it will be a Hobson's choice indeed.

A One-Time Opportunity

What Apple has so-far done is to dip the 10-year-old NextStep interface into the 20-year-old Lisa-Macintosh interface, then cover the lot with a thin, candy shell. What is absent is a great deal of true innovation. As proof, one wag whipped up a Windows "skin" that made Windows appear, for all practical purposes, just like Aqua, released on the web less than one week after aqua was first revealed (and pulled less than a week after that).

Apple has a unique opportunity to make serious, effectual changes to the interface for the first time in sixteen years. It has the opportunity to push the user-productivity of the Macintosh far beyond where Windows now lies. Such an effort should logically begin with Fitts’s Law.

Fitts’s Law

Fitts’s Law, by way of review, dictates that the larger the target, the faster the acquisition, distance being equal. As discussed in my column, "A Quiz Designed to Give You Fitts," the four corners of the display have the “magical” property of being infinitely deep, since the mouse pointer is locked into them no matter how fast it is “thrown.”

In addition, the four sides of the screen are infinitely deep. That is why, for example, Mac’s menu bar is so much faster than Windows’ and why the Windows Task Bar, which is anchored to the screen edge, is so popular.
Apple, for the first time in sixteen years, had the opportunity to grab the corners back and make more effective use of screen edges. I would recommend the following uses, clockwise from the top left:

Location Action
Top Left Corner Apple menu
Top Menu Bar
Top Right Corner Trigger for The Dock
Right Edge Window Scroll Bars
Bottom Right Corner  Trigger for Trash
Bottom Improved high-density Pop-up Menu mechanism
Bottom Left Corner  Trigger for 2nd Dock?
Left Edge Tool Bars

Note that clicking in the top left corner will reveal the Apple menu, whereas simpy mousing into the other three corners will immediately trigger their actions.

Some details:

The Top: The Menu Bar

The Menu Bar, in the OS X prototype, remains at the top where it belongs. However, things have been shifted around. The apple logo stuck in the middle of the menu bar is just plain silly. We know it’s a Mac. The Apple logo belongs on the left side of the menu, where it has resided happily for the last 16 years. Kill off that hideous pseudo-Picasso smiley face and return it to where it belongs.

Then, place the application name next to it, just before the File menu. In today’s interface, the current application’s name is way over on the right side. That would be OK, except usually the reason you are trying to read the application name is to figure out whose menus you are about to use, yet the name is as far from the menus as it could be.

Under the new scheme, of course, the application name might not have anything to do anymore, since the Dock is the location of open applications, as well as open documents. (I’ve been using that system under Windows for four years and I’m still confused. Maybe I just need another five or six years to get used to it.)

The application name could find two uses:

It could replace the word, File. The File menu is the top level of the application, and using the word, “File,” in each and every application on the entire computer is just a bit redundant.

It could replace the Apple. The first couple of items on the Apple menu already pertain to the application anyway, so one could make an argument for such a shift. The Apple logo could then slide all the way across, into the absolute top, right corner, and become the triggering device for revealing the Dock, for those who prefer the Dock to normally be hidden. (A lot more about this later.)

Some have objected that having a variable-length application name would cause the menus to shift around, reducing the predictability of the File and Edit menu locations. First, this is not necessarily a bad thing. The weakness of having the menu bar in a fixed location has always been that people sometimes become confused as to what application they are in. Any changes in the menu bar upon application-switch will tend to reduce that confusion. Second, the name could be right-justified, within a fixed-size box.

The Bottom

The bottom edge, today, has been put to excellent use as a location for pop-up menus displaying folder contents. They act as high-speed, high-density access to perhaps 250 to 5000 items. OS X, as demo’ed, appears to do away with this mechanism entirely. Were the new system to offer a fair substitute, such a change would be OK. However, I find no evidence of that.

The pop-up menus, if retained, wil need work: They tend to migrate around the screen at odd times quite on their own, they often disappear back into their folder forms at odd times, and they are likely to semipermanently foreshorten if you dare boot up from a CD ROM to, say, upgrade your system software. When these unexpected changes take place, they are a bother to get back in line. All in all, they are a great idea not so perfectly executed. What is being offered in their place, however, is a low density device of dubious merit: the Dock.

The Dock

OS X, as it is being presented, is filled with lots of pretty pictures, enabling the well-rehearsed Demo Dolly to dance around the interface to the delight of the audience. The deleterious effect of this emphasis can best be seen in the Dock, a graphical equivalent to Microsoft’s Task Bar.

The Dock, with its meaningful and attractive graphics, is, at first glance, far cooler than Microsoft’s text-heavy device. It shows a row of brightly-colored, meaningful, and differentiated icons. The demo sells the belief that it will be easy to pick out exactly what you want among these many unique-looking documents. But consider what will happen in “real life.” Let’s say you’ve been working on your mp3 collection for a while. Guess what? Every single document icon in the dock will look identical. "Well, that's alright," you say, "After all, all you have to do is look a the titles." Except there aren't any. The only way you can tell one document from another is to laboriously scrub the mouse over them, momentarily revealing one title at a time. Since the dock is capable of holding more than 20 documents, this could take some time. The Dock violates Rule 3 by requiring a vastly extended game of hide-and-seek.

The dock also appears to pile everything into one place: running apps, favorte apps, folders, etc. It probably makes more sense to have a second dock, perhaps triggered by the bottom right corner, that would contain just working apps. Or perhaps the Dock Manager could enable people to define their own docks.

A Word is Worth a Thousand Pictures

The Dock:

  1. Should have a default position along the left side of the screen, taking advantage of Fitts’s Law.

  2. Should display icons only when few enough items are docked to provide room.

  3. Should be stable, not jumping in and out, as in the current demo.

  4. Should offer the same auto-hide capability as the Microsoft Task Bar, but done correctly, with the user triggering it by throwing the mouse into the extreme top right corner, so the Dock doesn’t jump out at them every time they try to hit the scroll bar, as does Microsoft’s.

  5. Should not hold the trash can. The trash can should be on the desktop, where it belongs, with a mouse-movement into the extreme lower left corner bringing the traskcan to the front, with the mouse poised over it.

  6. Most importantly, the dock should always display document titles.

Such a Dock can show more than fifty files on a 17 inch screen, each with a unique identifier (their title). The great thing is that it can be made just as flashy as the one in the demo! As long as you have few enough documents, those great big flashy icons will appear. The only differences are that it builds down from the top, rather than shifting around from the middle (weird), and, rather than reducing down to an unrecognizable, unlabelled icon as the density increases, it would turn into a pure text label, always recognizable.

The dock should also go beyond what Microsoft has already supplied. If you have many processes, applications, and documents open at once, it can be hard to find your way around a task bar. People need ways of quickly organizing the bar, or perhaps have several bars showing, for example, only processes or only documents. One way to achieve this would be to enable people to use different corners to trigger different docks. Perhaps I might have the top right trigger an application bar, while the bottom right triggers a document bar, and the bottom left reveals my favorite folders. Food for thought.

The Left Edge: Tool Bars

The right edge of the screen has never been used for anything, and that tradition appears to continue in OS X. The left edge should be reserved for tool bars. It has a far higher access time than the foolish below-the-menus location Windows typically uses and could double the access speed for tools on the Mac. Of course, it will be important that developers can set the clickable area of their tools so that they bleed right into the right edge, even if a pixel or two is given over to offering a nice frame. Otherwise, the entire Fitts’s advantage will be lost.

Fitts’s At Large

Apple’s attempt to remove the clutter of all those extra windows reveals their lack of understanding of Fitts’s Law: Those nice, big windows are the easiest of targets to hit. Windows act as great big buttons, letting users swiftly click among documents and applications. It seems inconsistent to insist on getting rid of windows, as clutter, while introducing icons that border on the size of windows.

"Fittsized" controls should replace slower controls wherever possible. Contextual menus should be round or have reverse-zone-plate spacing (so lines farther away are fatter). Buttons and icons should be as large as practical and huge edges when possible.

This may seem incongruous with my criticisms of the Dock. My problem with the Dock and the giant icons is that massive amounts of screen real estate are being expended on just a few items. The interface must be balanced, as many different objects vie for their place in the sun.

Toning it down

Overall, the interface is too obtrusive for professionals. While I hope that is not intended, it is a possibility. It would fit in with Apple's decision to no longer make professional towers, instead supplying either all-in-one machines or low-capability minitowers with limited slot space, abbreviated keyboards, and round mice.

It is difficult to work on your own graphics when Apple's graphics are constantly demanding your attention. If Apple is interested in regaining the professional market, it will need to supply a softer, more neutral "skin" than the aqua we've seen.

Compatibility with the Monopoly

Many, if not most, of us who remain faithful to the Macintosh have had to take on a second partner, at least during office hours. It is a ratty experience to bounce back and forth between Mac and Windows. Apple recognized this problem in the storage and networking area years ago and provided solutions that let us easily move data between machines. The Apple interface, however, has been a different matter. Apple could argue, and few would deny it, that Apple was first and Microsoft is the one who made things difficult by failing to accurately copy the Mac interface. Indeed, Microsoft is the one at fault, but Microsoft is also the one who won the war. Apple can no longer afford to increase the difficulty of living in a Microsoft-dominated world.

OS X offers a way out. The user could be given a preference item that would map Command keyboard menu shortcuts to Control to conform to Windows. (Control is also in a better location, causing less twisting of the hand.) This preference, however, would initially be set to Command, rather than Control. Should the user remap the keys, the menu bar and OS X compatible electronic documentation would reflect the changed keys.

The default action button in dialogs (the SAVE button, etc.) could be moved to the left, where they don’t belong, but where 90% of the people in the world have been trained to find them. Again, this could be made a preference item.

Apple should add full support for a two-button mouse and cease forcing two-button mice manufacturers from doing mappings to the Control key. (Apple should also put a two-button mouse in the box, but that is for another discussion.)

Compatibility with the Mac

The lesson of the New Coke debacle is that change for change’s sake is not well received by loyal customers. Coke has customers that display a loyalty few products command. Macintosh has customers at least as loyal and perhaps more fanatical. OS X represents an opportunity for positive change and growth, but it is not an opportunity to arm-twist people into working in a new way just because it seems like a good idea.

A lot of really good work has gone into OS X already, and I anticipate more. At the same time, the Finder issue seems to me to be critical. If, as implied by the verbiage on the Apple site, people are to be offered a mere echo of the power of Apple’s traditional Finder as some kind of sop, Apple could find itself losing critical market share to Windows.

Don't miss the next action-packed column!
Receive a brief notice when new columns are posted by sending a blank email to

return to top

Contact Us:  Bruce Tognazzini
Copyright Bruce Tognazzini.  All Rights Reserved