With the recent launch of Windows 8 and the Surface tablets, Microsoft has reversed its user interface strategy. From a traditional Gates-driven GUI style that emphasized powerful commands to the point of featuritis, Microsoft has gone soft and now smothers usability with big colorful tiles while hiding needed features.
The new design is obviously optimized for touchscreen use (where big targets are helpful), but Microsoft is also imposing this style on its traditional PC users because all of Windows 8 is permeated by the tablet sensibility.
How well does this work for real users performing real tasks? To find out, we invited 12 experienced PC users to test Windows 8 on both regular computers and Microsoft's new Surface RT tablets.
Double Desktop = Cognitive Overhead and Added Memory Load
The Roman god Janus; Dr. Jekyll and Mr. Hyde; even Batman's arch-foe Two-Face — human culture is fascinated by duality. We can now add Windows 8 to this list. The product shows two faces to the user: a tablet-oriented Start screen and a PC-oriented desktop screen.
Unfortunately, having two environments on a single device is a prescription for usability problems for several reasons:
Users have to learn and remember where to go for which features.
When running web browsers in both device areas, users will only see (and be reminded of) a subset of their open web pages at any given time.
Switching between environments increases the interaction cost of using multiple features.
The two environments work differently, making for an inconsistent user experience.
Lack of Multiple Windows = Memory Overload for Complex Tasks
One of the worst aspects of Windows 8 for power users is that the product's very name has become a misnomer. "Windows" no longer supports multiple windows on the screen. Win8 does have an option to temporarily show a second area in a small part of the screen, but none of our test users were able to make this work. Also, the main UI restricts users to a single window, so the product ought to be renamed "Microsoft Window."
The single-window strategy works well on tablets and is required on a small phone screen. But with a big monitor and dozens of applications and websites running simultaneously, a high-end PC user definitely benefits from the ability to see multiple windows at the same time. Indeed, the most important web use cases involve collecting, comparing, and choosing among several web pages, and such tasks are much easier with several windows when you have the screen space to see many things at once.
When users can't view several windows simultaneously, they must keep information from one window in short-term memory while they activate another window. This is problematic for two reasons. First, human short-term memory is notoriously weak, and second, the very task of having to manipulate a window—instead of simply glancing at one that's already open—further taxes the user's cognitive resources.
Flat Style Reduces Discoverability
The Windows 8 UI is completely flat in what used to be called the "Metro" style and is now called the "Modern UI." There's no pseudo-3D or lighting model to cast subtle shadows that indicate what's clickable (because it looks raised above the rest) or where you can type (because it looks indented below the page surface).
I do think Metro/Modern has more elegant typography than past UI styles and that the brightly colored tiles feel fresh.
But the new look sacrifices usability on the altar of looking different than traditional GUIs. There's a reason GUI designers used to make objects look more detailed and actionable than they do in the Metro design. As an example, look at this settings menu:
The bottom of the Windows 8 settings menu on Surface RT.
Where can you click? Everything looks flat, and in fact "Change PC settings" looks more like the label for the icon group than a clickable command. As a result, many users in our testing didn't click this command when they were trying to access one of the features it hides.
(In that task, we asked users to change the start screen background color. As a further problem, the very command label had misleading information scent for some users; they thought of the Surface as a tablet, not a "PC.")
We also saw problems with users overlooking or misinterpreting tabbed GUI components because of the low distinctiveness of the tab selection and the poor perceived affordance of the very concept of clickable tabs.
Icons are flat, monochromatic, and coarsely simplified. This is no doubt a retort to Apple's overly tangible, colorful, and extremely detailed "skeuomorphic" design style in iOS prior to iOS 7. For once, I think a compromise would be better than either extreme. In this case, we often saw users either not relating to the icons or simply not understanding them. (Update: unfortunately Apple repeated many of Microsoft's mistakes in iOS 7.)
Icons are supposed to (a) help users interpret the system, and (b) attract clicks. Not the Win8 icons.
Low Information Density
The available advice on designing for the "modern UI style" seems to guide designers to create applications with extraordinarily low information density. See, for example, the following screenshots:
Start screens from the Bing Finance (top) and Los Angeles Times (bottom) apps for the Surface tablet.
Despite running on a huge 10.6-inch tablet, Bing Finance shows only a single story (plus 3 stock market quotes) on the initial screen. The Los Angeles Times is not much better: this newspaper app's initial screen is limited to 3 headlines and an advertisement. In fact, they don't even show the lead story's full headline and the summary has room for only 7 words. Come on, this tiny amount of news is all you can fit into 1366 × 768 pixels?
www.latimes.com in the tablet-mode browser.
Visiting the newspaper's website in Internet Explorer gives you much more information, though it's unfortunate that the site doesn't exploit the real estate offered by the widescreen aspect ratio on the Surface (and many full-sized computers). The website shows 9 stories (and 3 ads) in the same space as the 3 stories offered by the Metro app. Plus we get full summaries of the top articles.
Yes, big photos are nice. Yes, spacious layouts are nice. But you don't have to be a fanatic follower of Edward Tufte to want a bit more "data ink" on the screen.
As a result of the Surface's incredibly low information density, users are relegated to incessant scrolling to get even a modest overview of the available information.
As it turns out, users didn't mind horizontal scrolling on the Surface, which is interesting given that horizontal scrolling is a usability disaster for websites on desktop computers. Still, there's such a thing as too much scrolling, and users won't spend the time to move through large masses of low-density information.
Overly Live Tiles Backfire
Live tiles are one of the UI advances in Windows 8. Instead of always representing an app with the same static icon, a live tile summarizes current information from within the app. This works well when used judiciously. Good examples include:
Weather app showing current (or predicted) temperature and precipitation
Email app showing the subject line of the latest incoming message
Calendar app showing your next appointment
Stock market app showing the current market level
Unfortunately, application designers immediately went overboard and went from live tiles to hyper-energized ones. To illustrate …
Quick, without reading the caption, which apps do the following 4 tiles represent?
Live tiles for (clockwise from upper left): Urbanspoon, Los Angeles Times , Newegg, and Epicurious.
Newegg is the only app that includes its full name in the tile. When we asked participants to use the other apps, they couldn't find them. This on a new tablet with only a few applications installed. We know from our user testing of other tablets and mobile devices that users quickly accumulate numerous applications, most of which they rarely use and can barely recognize—even with static icons that never change. (Which is why we spend time on launch icon design in the course on Visual Design for Mobile and Tablet.)
The theory, no doubt, is to attract users by constantly previewing new photos and other interesting content within the tiles. But the result makes the Surface start screen into an incessantly blinking, unruly environment that feels like dozens of carnival barkers yelling at you simultaneously.
Charms Are Hidden Generic Commands
One of the most promising design ideas in Windows 8 is the enhanced use of generic commands in the form of the so-called "charms." The charms are a panel of icons that slide in from the screen's right side after a flicking gesture from its right edge (on a tablet) or after pointing the mouse to the screen's upper-right corner (on a computer).
The charms panel includes features like Search , Share (including email), and Settings that apply to whatever content the user is currently viewing. In principle, it's great to have these commands universally available in a single, uniform design that's always accessed the same way.
In practice, the charms work poorly — at least for new users. The old saying, out of sight, out of mind, turned out to be accurate. Because the charms are hidden, our users often forgot to summon them, even when they needed them. In applications such as Epicurious, which included a visible reminder of the search feature, users turned to search much more frequently.
Hiding commands and other GUI chrome makes sense on small mobile phones. It makes less sense on bigger tablet screens. And it makes no sense at all on huge PC screens.
Furthermore, the charms don't actually work universally because they're not true generic commands. In our test, users often clicked Search only to be told, "This application cannot be searched." Enough disappointments and users will stop trying a feature. (Also, of course, it violates basic usability guidelines; that is, you shouldn't tease users by offering a feature that isn't actually available.)
Finally, not all users understood that the commands are context dependent and do different things on different pages.
Many other features are initially hidden and are revealed only when users perform specific and often convoluted gestures. For example, all of our users had great difficulty with an extraordinarily basic task: changing the city in the weather app. Obvious gestures, such as clicking the name of the current city to change locations, didn't work. Users' difficulties were exacerbated by the fact that the "Modern" GUI style doesn't indicate which words and fields are active and/or can be changed.
What's the long-term usability of the hidden features in Windows 8? We might expect users to grow accustomed to the need to reveal the charms and other non-visible commands, even though this imposes additional cognitive overhead on using the system. That is, people must think to do something, rather than being reminded to do something, and thus users will sometimes neglect useful Win8 features.
Also, the familiarity bred by long-term use might be counteracted by the fact that well-designed websites have trained users to expect important features to be shown directly in the context in which they're needed. You simply can't design a website with hidden features and expect it to be used: website features are usually ephemeral, meaning that they must be explicitly represented if they're to gather any use.
Thus, people's experience with the web excerts a powerful pull in the direction of expecting visible features. It remains to be seen whether the Surface tablet's physical presence creates enough of an opposing pull to remind people to look for hidden features when they're using Surface apps.
The tablet version of Windows 8 introduces a bunch of complicated gestures that are easy to get wrong and thus dramatically reduce the UI's learnability. If something doesn't work, users don't know whether they did the gesture wrong, the gesture doesn't work in the current context, or they need to do a different gesture entirely. This makes it hard to learn and remember the gestures. And it makes actual use highly error-prone and more time-consuming than necessary.
The worst gesture might be the one to reveal the list of currently running applications: you need to first swipe from the screen's left edge, and then immediately reverse direction and do a small swipe the other way, and finally make a 90-degree turn to move your finger to a thumbnail of the desired application. The slightest mistake in any of these steps gives you a different result.
The UI is littered with swipe ambiguity, where similar (or identical) gestures have different outcomes depending on subtle details in how they're activated or executed. For example, start swiping from the right to the left and you will either scroll the screen horizontally or reveal the charm bar, depending on exactly where your finger first touched the screen. This was very confusing to the users in our study.
Windows 8 UX: Weak on Tablets, Terrible for PCs
As mentioned in the introduction, Windows 8 encompasses two UI styles within one product. Windows 8 on mobile devices and tablets is akin to Dr. Jekyll: a tortured soul hoping for redemption. On a regular PC, Windows 8 is Mr. Hyde: a monster that terrorizes poor office workers and strangles their productivity.
Although Win8 has usability issues on tablets, there's nothing that a modest redesign can't fix. In fact, usability could be substantially improved by revising the application guidelines to emphasize restrained use of active tiles, higher information density, better visibility of key features, and many other usability guidelines we've already discovered in testing other tablets.
(I was stunned to see the Architectural Digest app for Surface replicate a host of well-documented usability bloopers, such as not making the cover headlines clickable. Swipe ambiguity ran rampant, and users were often lost in this app's confusing combination of vertical and horizontal scrolling. All of this could have been avoided by reading reports we have published for free . I can just barely understand companies that ruin their user experience because they don't want to pay $298 to find out what the usability research says. But to create a bad app to save no money seems a puzzle.)
I have great hopes for Windows 9 on mobile and tablets. Just as Windows 7 was "Vista Done Right," it's quite likely that the touchscreen version of Windows 9 will be "Windows 8 Done Right."
The situation is much worse on regular PCs, particularly for knowledge workers doing productivity tasks in the office. This used to be Microsoft's core audience, and it has now thrown the old customer base under the bus by designing an operating system that removes a powerful PC's benefits in order to work better on smaller devices.
The underlying problem is the idea of recycling a single software UI for two very different classes of hardware devices. It would have been much better to have two different designs: one for mobile and tablets, and one for the PC.
I understand why Microsoft likes the marketing message of "One Windows, Everywhere." But this strategy is wrong for users.
I Don't Hate Microsoft
Because this column is very critical of Microsoft's main product, some people will no doubt accuse me of being an Apple fanboy or a Microsoft hater. I'm neither. I switched from Macintosh to Windows many years ago and have been very pleased with Windows 7.
I am a great fan of the dramatic "ribbon" redesign of Office (we later gave several awards to other applications that adapted this UI innovation), and I proclaimed the Kinect an "exciting advance in UI technology." I have many friends who work at Microsoft and know that it has many very talented usability researchers and UI designers on staff.
I have nothing against Microsoft. I happen to think that Windows 7 is a good product and that Windows 8 is a misguided one. I derived these conclusions from first principles of human–computer interaction theory and from watching users in our new research. One doesn't have to hate or love a company in order to analyze its UI designs.
I'll stay with Win7 the next few years and hope for better times with Windows 9. One great thing about Microsoft is that they do have a history of correcting their mistakes.