Peter TorpeyMedia Experience Artist

On July 22^nd, Microsoft announced that the next version of its Windows operating system would be named Windows Vista. Formerly known by its codename Longhorn, a beta 1 release of Vista was made available to the developer community. In the few days since then, a number of screenshots and reviews of the release have appeared on the internet.

I have not followed the evolution of Windows Longhorn closely— in part due to lack of time and in part due to apprehension. After learning about some of Microsoft's plans for distant Blackcomb (namely subscription services which I believe would drive consumers to embrace free open source alternatives), keeping up with the development of Windows made me a bit uneasy so I tried not to look for a while. Of course, with the release of Vista Beta 1, I couldn't help but to explore.

My reactions below are based upon studying literature, images, and other online resources demonstrating Windows Vista. Consequently, my concerns may in fact be unfounded. If this is the case, please feel free to allay my fears. Also, as the new OS is still in an early beta, much of what I will discuss will likely be addressed or change dramatically by Vista's final release in late 2006.

As reference to the reader, some excellent sources of information, as well as screen captures, are available on the web not the least of which are:

• Microsoft Windows Vista
• Microsoft Media Relations screenshots
• MSDN Windows Vista Developer Center
• Paul Thurrott's SuperSite for Windows
• Wikipedia article with concise info and links

A Clear Vista

Microsoft's goal for Windows Vista is to bring the user experience to a new level of ease, building on the success of Windows XP. As connoted by the name, one of the underlying goals of Vista is to provide clarity to users. To accomplish this, the Windows user interface has been redesigned and functionality added with the intent of improving the way users interact with their data, including searching and organizing files. However, even by attempting to decipher the myriad of proposed release editions of Vista, clarity may still be elusive.

Computer-human interaction and user interface design is a complex art form. From a usability perspective, it is difficult to balance ease of use with clarity of information-rich spaces. Whenever a software developer makes significant changes to the way the user interacts with a product, there will inevitably be a period of adjustment as users adapt to new methods of interaction and functionality. Extreme changes, however, can impede the ability of a person to use the software in familiar ways, negatively impacting productivity. Humans are very good at abstraction and classification. Consequently, if the user interface (UI) does not play into our expectations for categorization, oversimplification can undermine clarity.

Windows Vista is perhaps in such a precarious position. The visual metaphors of clarity throughout the new GUI can introduce context where it is not appropriate while other changes seem to occlude useful contextual information. In order to foster fast efficient interactions, the look and feel should provide means to carry out most tasks with as little user effort as possible, all the while offering unobtrusive feedback. In general, it seems that Vista requires more mouse clicks to accomplish simple goals. Also, favoring the mouse can hinder the use of the keyboard which is often a faster method of interacting with the computer. Keyboard shortcuts, accelerators, and mnemonics are extremely efficient mechanisms for navigating the interface and initiating commands. Though they are often employed by a so called ‘power user’, such an advanced user should not be slighted by concessions made for the accessibility of the program to novices who may prefer using the mouse.

The degree to which the computer's internal organization must be abstracted or simplified is debatable. Novice users may find familiar metaphors welcome only to discover that those elements of the interface do not behave exactly as they had expected due to an improper mapping of meaning. As a greater portion of the population becomes increasingly familiar with using computers, a time will come when all users will be at least somewhat familiar with the idiosyncrasies of computing, data storage, and the stagnant metaphors. At this point, extreme oversimplification will be unnecessary and perhaps hinder usability. Only a complete re-architecture of computer-human interaction would then constitute true progress.

The View from Where I Stand

I have been a user of almost every version of Windows since version 3.0. Prior to that, my experience with GUI shells included GEOS and Mac OS. Presently, I have Windows XP Pro on my primary system and it has proven to be highly stable and very usable. I enjoy working with Windows. Despite all of the bad press and ill will Microsoft seems to have garnered in some, I very much respect what the company has achieved. Certainly, there are many annoyances that crop up in any software. I personally have not had any serious problems with any version of Windows that I have used (yes, including Me). Could things be better, more stable, more secure, easier to use? Of course! Is Windows my favorite operating system? Well, probably not. The flexible design philosophy underlying various Unix-like operating systems (and related software for those platforms) tends to win out.

I enjoy studying user interfaces and have read and written on the subject in the past. With that in mind, the UI will constitute the bulk of the following discussion. As my primary OS will be receiving much of my daily attention, I am always wary of changes that will impact my workflow, affect usability, and become tiring to look at. Mustering the gumption to install Windows 95 over 3.11 wasn't easy for me, though I am now grateful I did.

I'm the type of user who immediately heads for the preferences when installing a new application and can spend days tweaking options and configuration files. However. with the limited ability to customize, it is imperative that Microsoft goes to great lengths to develop a look and feel that is aesthetically pleasing and usable, satisfying a large and diverse user base.

In recent releases, Windows has become increasingly rigid in its UI, to the point of allowing only so much configuration as three color scheme choices for Luna in XP (with visual styles enabled). This contrasts sharply with the Unix world where applications, window managers, and desktop environments are so readily and highly customizable. The possibilities for customization with XUL with CSS on the Mozilla platform are amazing. Cross-application consistency can suffer with custom application UIs. However, many developers have indulged in a unique UI for the application (customizable or not) for some time and users have coped just fine, not to mention the varying designs of web sites that users navigate with ease. So, cross-application consistency may not be as important as once thought. Avalon and XAML will present unique opportunities to Windows-based interface designers. I expect that we will see visually amazing interactive applications in the near future that will vary greatly any notion of cross-application consistency.

Microsoft has also long since eliminated any ready means to report bugs or submit feature requests online. Understandably, a company with such an impressive user base (not to mention those who just harbor ill spirits) would have tremendous difficulty fielding all of that input. Yet, such a matter of logistics has the unfortunate side effect of distancing users. Products are designed to satisfy certain target markets and the vast numbers of users that may be just slightly in the margins have no choice but to conform or jump ship. If Microsoft says this is the way I should find a document, then this is the way I must do it. This is precisely what is beginning to happen with Internet Explorer users migrating to Firefox, a product with desired features and the extensibility to handle variations in the user base.

Perhaps I don't use a computer the way most do. I would consider myself a member of the Power Users group, with all the keyboard shortcuts and privileges thereof. All the while, I understand the need for usability and simplicity for users that are not quite as experienced. Task-centric interaction and context sensitive UIs are a boon for helping people be productive. However, when the OS starts making too many decisions for me and hiding functionality, I begin to feel stifled. I have experienced this most frequently on the Macintosh since the Mac Plus days. Windows tends to follow suit, though to a lesser extent. This may be the reason I feel most comfortable in Linux where I know, if I want to, I can be in full control. In Windows, so far, I have been able to find what I am looking for and make changes if need be. I don't give a second thought to popping open RegEdit to fix an annoyance in a program or the OS (remember, it is recommended that you backup your system registry before editing… can't say that I have). I can usually navigate quickly throughout the system using the keyboard, reserving the mouse (or stylus in my case) for painting and drawing. In many cases keyboard macros, accelerators, meta keys, Tab, and the arrow keys are much faster than delving into mouse driven menus, performing cumbersome dragging operations and clicking to drill down through over-categorized options and wizards. These are often not the default or recommend methods for interaction for novices, but there should be a thoroughly rich facility for such accelerated interaction in the system, even if this means enabling it through options or the installation of extensions (like the TweakUI shell for certain registry options).

There’s Something in the Aero

The first thing one notices when sitting down before a computer screen is the look of the interface. It should be engaging, soothing to look at for hours on end, and unobtrusive. The meaning of the components and controls should be clear and readily learnable, if not intuitive, and the interface as a whole must provide affordances as to how user can interact with it. It's the eye candy that draws users in, but it is the usability that keeps them there.

Since Windows 95, Microsoft's operating systems have shared a fairly similar user interface. The controls were familiar; the layout reasonably consistent. Minor changes to color and appearance were made throughout the years. The default grays became warm neutral tones. Solid title bars (window captions) gave way to gradients. Toolbar buttons ceased to reveal their tactile 3D look until mouse over.

When Windows XP came along, the familiar controls got a new skin thanks to Visual Styles. This raised the eye candy factor a bit to compete with the brand new visually rich look of Mac OS X. The XP themes didn't quite wow users with the Mac's fluid animations and shiny transparency. Of course, Apple prides itself on visual design, both on screen and in the physical form of their hardware. While far from perfect, OS X's GUI is easy on the eye, reasonably consistent, and attractive. Indeed, minor changes to the look and feel have been introduced in the point releases of OS X as a result of user feedback. In Vista, Microsoft takes strides to bring the Windows UI up to par with Apple's look and feel.

In a brilliant blend of visual cuing and branding, Microsoft chose to take the four colors from the Windows logo (blue, green, yellow, and orange-red) and applied them to the Windows XP user experience where the colors' typical associations reflect the behavior of a command. For example, closing an application window or shutting down the OS is a drastic or final operation, potentially resulting in the loss of data, and thus would be colored red. Green is used to initiate an action, such as the Go button in IE or the Start button. Luna's Bliss blue window caption and taskbar contrasted nicely with the complementary warm neutral toolbars and controls. Nevertheless, the bright blue could be distracting at times, drawing the eye toward the top and bottom of the screen. This colorful scheme contrasts with Apple's approach in OS X. There, the UI is essentially monochromatic color scheme with a subtle blue accent color to denote default actions or manipulators.

Known by the moniker Slate, the visual theme of Vista's UI is pleasing, refreshing, and elegant. Taking a cue from Apple, Slate is rather monochromatic, toning down the vivid colors from XP by introducing color subtly as feedback to the user to aid in navigating the interface. Some of the XP color scheme survives in certain buttons and icons in Vista, however this will likely change as development continues. Standard window frames and widgets are predominantly black on white while the taskbar and start menu are a beautiful glossy black with crisp white text and icons. The neutral palette is not distracting and all the while classy. A light shade of blue is used to indicate focus and selection. While bright and attention grabbing, as I suppose a highlight color should be, it is not at all distasteful. I am grateful that windows are not overly colored and distracting as in Office 2003's default blue theme. Borrowing gradient effects from Windows Media Player 10 (though thankfully not quite as shiny) and the Windows Media Center Edition theme Royale (with better contouring of the terminator between light and shadow), Slate offers the eye candy appeal of OS X's ‘gelcap’ Aqua buttons, though not transparent (and arguably less distracting). It is a far cry from what has been dubbed the ‘bubble gum’ or ‘Fischer Price’ look of Luna.

The window captions, borders, address and search bars have a unified white translucent appearance that thanks to Aero Glass and advanced display technologies reveal a diffused view of whatever lies beneath the window. This effect makes one “Ooh” and “Aah” and then ask, “Why?”

It seems that Microsoft's UI designers took the notion of clarity a bit too literally. Presumably, having translucent windows is intended to offer visual cues as to the context of the window. In practice, despite the fact that the diffusion mitigates the influence of the underlying image, the effect is simply distracting. I don't need to be able to see a blurry version of my wallpaper or several other illegible title bars beneath the active window. It is not necessary to ascertain the context (stacking order and other open windows) in this manner, especially in an application with sovereign posture. It is likely that the pared down versions of Aero, intended for lesser graphics hardware, will prove useful in disabling the translucency.

If used at all, transparent or translucent windows should be reserved for use in a special few cases. Active sovereign windows should not be transparent as they are the primary focus of the user's attention. It may be justifiable to make inactive windows become translucent, allowing an image through just the chrome or even the entire window. Early versions of Aqua in OS X (10.1 and 10.2) used the former approach making the title bar partially transparent on inactive windows. However, this effect was removed from future versions to eliminate clutter.

Transient and secondary windows are candidates for transparency effects. Most likely this would be limited to modeless dialogs (especially if inactive) and menus where the attention is still on the parent window. Though not widely used, transparent windows were available to developers in Windows 2000 and XP. One particularly useful application of transparency (in fact the only one I can think of) is in the Helios Software's TextPad. In that editor, the non-modal Find and Replace dialogs become transparent when the document beneath has focus. This allows the user to see text and even the found selection region that would otherwise be occluded by the dialog. How I wish Find in Internet Explorer were transparent as invariably the word for which I was searching is scrolled into view but hidden beneath the dialog. (Other practical ways of handling searches of a document in IE and other applications would be to move the dialog automatically so that it doesn't obscure the word, as is done in Word, or incorporate find into an unobtrusive toolbar, as in recent versions of Firefox.)

Like OS X, windows under Aero have drop shadows, which are a welcome form of visual enhancement. In XP and 2000, drop shadows were only available on some menu drop-downs, tooltips, and mouse cursors. Window shadows do help to convey context information, namely helping the user recognize the z-order of windows. We see shadows everywhere in the physical world, so our mind's eye has become quite adept at interpreting shadows and lighting while inferring information from such phenomenon. This is often done without conscious awareness, thus such a subtle visual cue as shadows do not detract from areas of focus.

The window control buttons that appear in the title bar in Vista are nicely integrated into the continuous look of the window chrome. While some images show a window occupying the entire client area, I don't believe I have come across any screenshots of a maximized window as I have not seen the restore icon on the middle button. Furthermore, a maximized window should not have borders, offering as many pixels as possible to an application assuming a sovereign posture. Keeping Fitts's Law in mind, the title bar buttons are a bit smaller in size, reducing the clickable area. However, they still seem quite usable. The Close button is slightly larger than the Minimize and Maximize/Restore buttons. As Close is used most frequently, this is a reasonable design decision. However, it also marginally increases the likelihood of a user accidentally initiating a destructive action. For a maximized window, it is important to be able to throw the mouse cursor up to the top right corner of the screen and be able to click to close the window without much fine movement. Without having used a maximized window in Vista, it is unclear whether the clickable region for the close button extends to the edges of the screen in order to facilitate this interaction. In all previous versions of windows, the hotspot was cheated beyond the visual boundary of the button. It appears that, if the borders are removed when the window is maximized, the close button will be sufficiently close to the screen edge to accomplish this.

One suggestion would be to change the transparent nature of the window control buttons. When not hovered, they have a transparent colorless appearance with slightly inset white icons. However, in several screen captures, the buttons looked particularly nice when over a dark portion of the desktop wallpaper. It may be pleasing to make the buttons an opaque glossy black much like the Start Button unifying the appearance of such controls.

Start Menu du Jour

Usability of the All Programs submenu in the Start Menu has been improved somewhat by trading sometimes-difficult mouse movements for extra clicks. In previous versions, the nested menu-like structure of the Start Menu expanded across the screen requiring the user to keep his mouse over the menu to prevent it from closing or switching to a sibling menu. This mouse movement has been replaced by clicking to drill down using a scrolling list of items. When a parent item is clicked, and a submenu would have opened, the entire half of the Start Menu is replaced with the child menu. A Back button is provided to navigate in reverse, up the menu hierarchy. While this introduces more clicks, making launching an application a bit more involved, it does alleviate the fine motor control required of mouse users in previous versions. Perhaps, after viewing All Programs, the Back button could be re-titled “Back to recently used programs”, or from Accessories it could read “Back to All Programs”, as just “Back” is a bit vague in this context. Due to the shallow nature and limited complexity of the All Programs menu, Microsoft's favored linear navigation is suitable here, as it is comparable to flyout menus and introduces little additional excise.

A search box is nestled in to the lower portion of the Start Menu. This is a splendid place to keep a search box, as it is readily available, accessible within one click (or press of the Winÿ key) and does not detract from the interface. Though it would be difficult to determine the intent of the user's input (perhaps an option), it would be particularly utilitarian if this box could also function much the same way as the Run command or address bar. For a while, I had kept an address bar in my taskbar for this purpose, but it consumed too much space. Now, I just press Winÿ+R for convenience. Similarly, in Linux, I always keep a console window at the bottom of my desktop to quickly launch programs and execute commands. A user could be able to type a fully qualified path to open an Explorer window to the folder, a URI to launch the browser, or name of an application to execute it. Unfortunately, I have read that the search box receives focus immediately when the Start Menu is open, making the use of mnemonic keys to access other items in the Start Menu impossible. I hope that the arrow keys will still function, not being captured by the text box, allowing the user to navigate the menus with only a keyboard. If this is the case, using the arrow keys to blur focus from the search box could allow subsequent key presses to be interpreted as mnemonics for the Start Menu items.

I typically don't search my system much, so I didn't mind pressing Winÿ+F every now and again. Having a search box tucked conveniently in the Start Menu is not a bad idea either. For one who searches frequently, the search box at the top of each shell window may be practical. However, I would hope that its presence along side of the address bar is optional so that I may reclaim some space for more of the address bar or other toolbar items.

Icon Hardly Believe My Eyes

With every version of Windows comes a new set of icons that are more elaborate and detailed than what came before. In the past, improvements came from greater color depth, followed by 3D perspective drawing, transparency, and drop shadows. In Vista, the new Windows Presentation Foundation layer (Avalon) offers the ability to create scalable vector graphics that may be used for icons.

Multiresolution icons have long been available in traditional 16 and 32 pixel sizes. XP introduced additional icon sizes, which were rarely used. OS X also features scalable icons. However, despite impressive vector technologies, large icons aren't necessarily a good thing. Indeed, what icons have been created for Vista so far are nicely designed, with a shininess and translucent glow reminiscent of icons in OS X. However, beyond potential concerns for low vision users, icons that can consume nearly one quarter of the window (as I have seen in some screen shots) are impractical. Hopefully, preferences for icon display sizes persist across folders, regardless of the selected view.

Graphical icons in the computer world were intended to encapsulate the meaning of an object or command. Typically, they are very small in size and are designed to convey such information in far less space than the equivalent words. Yet, with the excessive size of icons in some parts of Vista, their original purpose is lost, as text could concisely tell the story in a far more space-efficient manner than these oversized icons.

In some parts of Explorer, massive icons just serve as a grotesque waste of space. Simply because an icon can be made larger and there is space to do so, does not mean that it should be drawn as large as possible. While the user can choose a reasonable size for the icons in most views, the colored horizontal task panes necessarily consume a significant portion of the Window. The bars feature some useful textual information. To view more text, the task pane must be resized larger. What better way to fill the remainder of the wasted space than with a gigantic icon? The icons may be beautiful at that large size; I imagine that they would just become tiresome over time.

When designing a multi-resolution icon, it is difficult to create an image that is appealing at a large scale and maintains its visual integrity when reduced in size. Some of the icons appear strange at smaller scales as anti-aliasing bleeds detail into sharp edges. The translucent arrow box overlaid on the icons for shortcuts is a nice touch. However, like Aero Glass, the blurred see-through areas create more visual clutter, which can be somewhat distracting. The more different types of effects, surfaces, and styles that the mind's eye perceives, the more unnecessary attention each element warrants.

One of the most visible icons in any desktop interface is that of a folder. The continued usefulness of a folder to denote a directory node in a hierarchical file system is debatable. For now, let us resign ourselves to the persistence of this aging metaphor and focus on the representation itself.

It is the folder icon that set the tone for my initial reaction to my first glimpse of Vista. Upon seeing the representation of a manila folder on its end, I experienced a physiological reaction. I felt very unsettled and insecure. Perhaps it was a sensation that the folder may collapse from the precarious balance or that its (in some cases, already protruding) contents would spill out rather than be safely contained while peeking out from the top. I would not attempt such a feat with my real-world manila folders on my desk as I have a difficult enough time keeping the papers inside.

In that initial moment, seeing the interface, the folder icons weren't even readily recognizable as such, perhaps due the familiarity with the traditionally-oriented image of a folder. Though it still makes me a bit uneasy, I have begun to adapt to the upright folders in their larger views and perhaps, with time, it won't be an issue. However, the vertical orientation of the folder representation does not translate at all well to the 16 × 16 icons. These smaller icons hardly resemble folders. The horizontal small folder icons in prior versions were identifiable (though they looked rather fat in XP). I have seen many folder icons for themes for other desktop environments that are more readily recognized, even if radically different from the classic manila folder, in different positions with contents sticking out, that aren't unsettling. Some additional redesign for these icons may be justified, prior to Vista's release.

Vista promises a thumbnail view that is more comprehensive than in previous versions and is similar to other file managers such as in KDE's Konqueror. The image previews on the items within the folders are reasonably useful and clever. Having the photograph protrude from the folder is more consistent with the metaphor than displaying sample photos tiled across the surface of an irregularly proportioned folder as in XP. However, multiple sample images are obscured, and may not be identifiable, thus it is somewhat less useful as a preview of the folder's contents.

This leads to some inconsistencies in the way a folder's contents are rendered, though I expect that these would be ironed out by release. Traditionally, a folder appeared empty regardless of whether it contained any documents. Only the Recycle Bin icon had a notion of being empty or containing items. However, since the folder and Virtual Folder icons in Vista are slightly open and some special folders, such as those containing images, display their contents, all folders that contain items should indicate that fact for consistency's sake. It appears that some of the small folder icons (which aren't open) and the scalable icons in the task pane indicate that there are items inside. It is unclear whether this representation is unique to folders that do actually contain objects. As it is possible to have thumbnails of office documents or videos, they could appear to be inside folders just as selected images emerge from the folders that contain them. At the very least, a generic document (a piece of paper or two) should be seen in non-empty folders.

Now, for one of my icon pet peeves. The beautiful icons for hard drives are some of the most photorealistic representations in Windows XP, Mac OS X, and other environments. Yet, from a usability perspective, this is absolutely a disgrace. While the folder metaphor for a directory is questionable (an object that ‘contains’ documents for the purpose of organization), the representation of a disk drive bears no metaphorical or logical relationship to its function as perceived by the user. A hard drive should not look like the physical internal drive. It is likely that many users have never seen their internal drives and have no concept of what the physical hardware looks like. If they can recognize the device, it is likely from association with the icon, not firsthand experience. This is the inverse relationship of that upon which icons are intended to function.

Icons that represent devices and peripherals are common and their images depict that device. For display settings, an image of a monitor is reasonable, as the user knows what a monitor looks like. The same holds true for a mouse. Network adapters, on the other hand, often appear as circuit boards, again something the average user is unlikely to have seen. An image of a compact disc is acceptable for CD and DVD drives, as the user knows what the media look like and realizes that there is information stored on them. However, a picture of the actual medium or device, in the case of removable media and hard disks, doesn't really impart meaning: the notion that these items contain data as a file system full of documents. In keeping with the folder metaphor, perhaps a filing cabinet might be a more appropriate representation (though, users tend to have negative associations with filing cabinets). Some may recall the filing cabinet icon used for File Manager in Windows 3.x versions. Most of the icons in the system employ a metaphor. However, these drive and device icons do not, creating a confusion that presupposes a certain amount of knowledge by the user of how the computer works.

In Vista, internal drives now exhibit a faceplate and housing of some sort, toning down the complexities of the icon in previous versions. While this might make the icon more intelligible, it is less representative of the drive's actual appearance without introducing a metaphor. Furthermore, the icon resembles a clunky external disk circa 1988 that has been augmented with a blue LED. The proportions of the drive depicted do not seem to match that of a 5¼″ bay, as would be expected. While this might be a better representation of a disk volume than an image of an internal drive, it still does not look like any drive that I have or that users are likely to see.

Desktop Clutter

Since the early screenshots of Longhorn, there seems to be a preoccupation with consuming valuable screen real estate. I am pleased to see that the rather large “Sidebar” with analog clock, calendar, slideshow, et cetera does not appear in recent builds. The Sidebar is a vestige of ActiveDesktop components, channels, and other once-promising technologies that never bore fruit. Though the base metaphor for the computing experience is the Desktop, the screen should not be subject to such clutter as accumulates on my physical desktop. This is especially so in a user experience that prides itself on clarity. Having every conceivable feature one click away consumes a considerable amount of real estate and the resulting visual clutter can make finding the one thing you are looking for more difficult. It is often easier to find data in a well-organized, though preferably shallow, hierarchy rather than in a homogeneous pile. The UI should, and often does, take advantage of the organization and grouping that a user might expect.

The new Google Desktop 2 seems to have supplanted the role of Sidebar for those who want it. In a similar form, Google has incorporated many of the features of the Longhorn Sidebar and then some.

My Desktop rarely sees the light of day. Only when rebooting every few weeks do I see my pretty wallpaper. Since I'm trying to keep my current systems neat and clean, few documents and shortcuts have found their way to my Desktop. Indeed, I'm a fan of applications that take an immersive sovereign stance. Nearly all of my staples (eg. Office, Outlook, IE, TextPad, After Effects, Photoshop) are forever maximized. I find it very annoying when a program (IE) decides to stop opening its windows in a maximized state for no apparent reason. I have no need to visit my desktop. Thus, any widgets and dashboard components and Sidebars are of little use to me. The small form factor of the taskbar and the system tray (status notification area) is the welcome location for any controls that should be visible at all times.

Internet Explorer 7

The killer feature of many new browsers, such as Firefox, is none other than tabbed browsing. Whenever someone incited one to switch to Firefox, his or her number-one reason (possibly aside from security concerns) is that it has tabbed browsing. So important is this UI feature that it had been implemented in Internet Explorer 7 beta 1, superseding much-needed updates to the browser's CSS support.

As I see it, I have always had tabbed browsing. My tabs, however, are just the instances of IE in the taskbar. Now, before you try and convert me with an argument favoring true tabs: I do use Firefox on my Linux machine where I open and close tabs plenty. I have known true tabbed browsing and yet I don't miss it when I am using IE on Windows. Certainly, the taskbar can get a bit cluttered and then, “whee,” all of thus IE instances get grouped, which is not all that bad. My tab bar in Firefox can get pretty messy, too. There isn't any grouping or rearranging of tabs (without extensions) to help there, either.

With an option in IE, each instance of the application can be a separate process. This is useful when the browser crashes as only one process will terminate. I usually have many IE windows open and, although crashes are infrequent, they typically occur at inopportune times. I hate losing the other windows. With tabbed browsing, if the browser crashes, all open tabs are likely to be lost without some mechanism to save the browser's state.

IE 7 does promise improved CSS support, satisfying all of CSS 1 and a good deal of CSS 2.1. As of beta 1, it is still disappointing, though the IE team on their blog promises some additional much-needed fixes by the application's release. Yet, it will not be comparable to Gecko and Safari. The other browsers will continue to plug on, adding support, looking toward the horizon when CSS 3 is ratified. IE will still be lagging, despite its early commitment to web standards (i.e. taking its proprietary features and making them standard such as JScript contributing to ECMAScript). Considering the time when IE 8 (or whatever future version number) catches up in CSS support and has a sufficient install base, replacing earlier versions of the browser, the web designer's dream of writing standard code without hacks and tests for compatibility is still a long way off. Of course, I am ever so grateful for the progress that has been made since the Browser Wars of the late 1990s.

Of course, web designers have become so accustomed to hacking around Internet Explorer's quirks and coding hacks or multiple versions of site files to suit IE, that proper CSS support isn't as high of a priority as it should be. Microsoft is less concerned with appeasing lowly web designers than users who likely will jump ship for features such as tabbed browsing. Hence, it is necessary for Microsoft to focus their attention on tabbed browsing and RSS and security to prevent IE's slipping market share from slipping away altogether.

As for security, I have never really had much of a problem with IE or Windows for that matter, keeping in mind I still have yet to upgrade to SP2. I make a few minor tweaks to IE's security settings on a new install just in case I stumble upon an ActiveX control that wants to install itself. Otherwise, I have managed to avoid having my home page hijacked, accumulating spyware and malware, and other such annoyances.

Me, My Shell, and I

As is to be expected, a number of changes have been introduced in Vista's shell. In the quest for clarity, Microsoft has modified the shell UI in order to integrate new technologies, such as the search and organizing facilities of Virtual Folders, as well as changes to common components with hopes of improving the user experience.

For a while now, versions of Windows have been trying to get the user to abandon hierarchical navigation (such as through the file tree) in favor of some sort of linear navigation. The metaphor is borrowed in part from browsing the web. When Windows began integrating web functionality into the bulk of the operating system (with Active Desktop, Web Folders, and the like) two little buttons appeared in Explorer: Back and Forward.

We were familiar with these buttons from our web browser. When we traverse the World Wide Web, clicking on hyperlinks from one web page to the next, we create a story. It is a linear narrative. Like walking down a path, we can turn around and go back to where we have been, or proceed forward choosing a fork when we come to it. Going back and choosing a different fork does indeed create a hierarchical tree or even a general graph (should we find our way back to a page we had already visited), but we cannot jump from one branch to another without retracing out steps. Such a linear interpretation of our web journey seems to serve us well.

A file system, on the other hand, is indeed a well-defined hierarchy. However, when Microsoft sought to blur the line between the web and the local system (which is not necessarily a bad thing, but I am strictly concerned with methods of interaction at this point), they chose to bring that same mostly-linear way of navigating an information space to the local computer. To my mind, this metaphor just does not belong here. I am much more comfortable navigating a tree view than trying to remember where I have been and figuring out where I want to go without any displayed context more than one arrow pointing left and another arrow pointing right.

Through the years, however, those two little Back and Forward buttons have gained prominence in the Explorer UI. More vivid coloration has given way to larger buttons. The not-so-little-anymore buttons now dominate the Vista UI, and I believe undeservingly. This linear paradigm is the way Microsoft wants you to navigate all of your information spaces. To me, at least, this seems like the adoption of an inefficient metaphor that provides a confusing oversimplification rather than useful context and efficient interaction. With one click on any node in a tree and I can be there. Walking back along a path and then forward again constitutes tedious excise.

Adjacent to the back and forward buttons in Vista, Explorer windows now feature a redesigned breadcrumb-style address bar with drop-downs for each tier of the file system hierarchy revealed by the displayed path. While this is an interesting method of interaction, it is essential that a full path or URI can be typed into the address bar, as well. It is necessary to retain the ability to navigate by pressing Alt+D to focus and select the current address and begin typing a path followed by Enter. From what I have seen, this doesn't appear to be possible or at least easy. Clicking on a drop-down menu to alter the path means that the current directory can only be changed one tier at a time with a refresh of the main Explorer pane occurring each time. This tedious mouse-based approach is not an efficient means to navigate from one directory to another unless the target directory is a sibling of the current one. The keyboard is the most powerful and quick method by which users can interact with the system, thus the ability to use the keyboard and directly enter text where appropriate is always more efficient than mouse clicking.

As with any significant change in functionality, an option should be provided in order to revert to the previous method of implementation. So, somewhere there could be a checkbox to disable the breadcrumb address bar in favor of the familiar editable combobox. Microsoft has been pretty good about such options in the past: re-enabling Personalized Menus, disabling the Control Panel category view in XP, disabling visual styles to return to the traditional line-drawn look, and others.

A tree view is important, too. While the breadcrumb address bar provides access to the file system hierarchy, it does not immediately provide context as does a folder tree. The siblings of a node only appear as a drop-down when the node is clicked. Clarity and understanding are often derived from context. In an UI that strives for such lofty goals, it is disappointing to see that some of that context has been removed. Visual context was part of the rationalization for the transparent frame of Aero Glass windows, yet it is missing from Explorer. Vista should reintroduce the information context and forego the visual context, which is really just clutter and eye candy.

The breadcrumb address bar is similar to the Mac OS X Column View seen in Open and Save dialogs. As if all elements of the breadcrumb path had their respective drop-downs visible, the OS X view provides more context at once than the Vista address bar by showing multiple items at each tier of tier hierarchy. Essentially, it is a left-to-right flowing tree, but consumes considerably more space and is not quite as easy to navigate as a traditional tree, with only one path visible at a time. I was initially confused by that view in OS X when I first encountered it. Even after time, I found it to be a tedious and inefficient means to navigate the file system. A folder tree view consumes less screen real estate and can be navigated more quickly. It requires only a few clicks allowing branches to be expanded and collapsed at multiple nodes.

In Windows Vista, the tree view offered by the Folders Explorer Bar seems to have been replaced by a dynamic list of shortcuts that accompanies the display of some Virtual Folders. This is similar to the “Other Places” section of the XP task pane. This has a certain appeal, but I don't like it when the OS decides what is relevant to me in an effort to streamline usability. It usually doesn't get it right. If the user finds the OS's “help” inadequate, he should be able to take control with more advanced views and interaction (such as switching from Simple to Advanced File Sharing in XP… I never could completely wrap my head around the Simple File Sharing). This reminds me of Personalized Menus introduced in Windows 98. This feature was generally abhorred and still exists, but is no longer the default. Let us applaud Microsoft for choosing to maintain functionality, even though it is no longer the preferred or default method of interaction.

I usually keep an Explorer window open with certain branches of my file system revealed in the tree view at all times. Throughout the day, I am always jumping among these branches, which are accessible within one click and in easy view. With the dynamic list of related files and the breadcrumb address bar, it seems like this sort of interaction will be more than one click: routing through breadcrumb menus or backing up until Explorer decides to show me the folder I want to access in the list of relevant choices or the main view pane. Again, the representation of context and location in the file system is weakened, revealing only one pseudo-linear path and a collection of items that are not necessarily related to my current location.

In keeping with their emphasis on linear navigation, it appears that Microsoft no longer believes that the tree view is a useful control. It is not enabled by default in XP in order to reveal the task pane. Additionally, for some time, the now optional lines connecting the nodes in the tree view have been disabled by default. While not essential to determining the level of the hierarchy and an object's siblings, they do clearly convey a bit of context that I find speeds up my interpretation of the tree slightly. Thus, I'm quick to turn that feature on. As evidence of their deprecation, the lines do not even render properly in XP. I suspect Vista will continue to steer users away from using the traditional folder tree as it is not visible in most screenshots that I have encountered. Drawing the lines may not even be an option especially considering the dynamic nature of the convoluted new tree, which doesn't necessarily represent a hierarchy.

Nearly all shell windows sport a new layout featuring those large Back and Forward buttons in line with the address bar at the very top of the window. Below are the usual menus, then controls, and the display pane or document itself. This rearrangement of window components is a bit jarring at first as the menus are usually just below the title bar.

The layout of a window's controls should maintain a visual relationship, by proximity and containment, with the document. Components specific to the document should be positioned closer to the document. Controls that relate to the application should be closer to the window borders. The outermost part of the window contains the title bar, window controls, and system menu. This is where the application interfaces with the operating system that contains it. Just inside the window border should be the application's menus since they contain global options that affect the application as a whole, including operations on multiple or individual documents. The layer within that would be comprised of toolbars and palettes that contain controls having a direct effect on the active document. In an MDI, controls for manipulating open documents, namely windowing controls and scroll bars, are just inside the toolbars, which seems appropriate for visual consistency. However, when a document is maximized in an MDI, its window controls understandably appear inline with the menu bar. Last, we find the document itself. The proximity of the control to the document viewport demonstrates in degrees how the functions relate to the active document.

By placing the navigation buttons and address bar above the menu bar in Vista, this relationship is destroyed. The address bar, Back and Forward buttons control the document being displayed and thus should appear in close proximity to the document itself, as was the case in the past. The menus that control the behavior of the application are now close to the document. The problem becomes more apparent when looking at IE 7 with its new tabbed browsing feature. The tab bar appears between the address bar and the file menu with the web page below that. The order should most definitely be reversed: Title Bar, File Menu, Address Bar, Tabs, and then the document. This is the layout found in most other browsers. It may even be reasonable to have the tabs precede the address bar, so that the currently displayed URL is visually tied to the active tab pane. The controls would then be arranged in order of specificity to the active web page. As it stands, the layout is confusing at best.

Fitts's Law is also compromised by the new layout. Since the majority of the user's interaction will be in the document or display viewport, especially in applications that support rich direct manipulation, controls used frequently should be closer to the document area requiring less travel by the mouse cursor in order to access them.

I understand that since the navigation buttons and address bar will be present on most shell windows, they are positioned to maintain consistency, even when the order violates intuition. Consistency of layout versus logical layout is a difficult matter to weigh. The ideal interface would both be consistent and logical without presenting disabled controls when not needed and other similar clutter. However, within a relatively small area from the top of the window to the top of the contained document, users are more attuned to the appearance and function of a control than its location. Thus, in this instance, relaxing the constraints of consistency in favor of logic may have been the wiser choice.

It is interesting to note that since time in memoriam, Mac OS has placed the menu bar outside of the application window at the top of the desktop. Completely distancing the menus from the application, this is quite the opposite of Microsoft's new approach. The items specific to the application are intermingled with menus, indicators, and icons that have system scope making it difficult to differentiate amongst them in a glance. This scheme also hinders usability, as an application must first have focus before its menus can be seen and activated. Presumably, placing the menus in the same location for all applications was done for consistency. However, the visual link between the menus and the application is lost. It seems almost as if multitasking is an illusion as the active application dominates a portion of the system's visual interface in a modal way. Then again, in Mac OS, some programs do not have strict application windows to contain their components and thus do not have any appropriate location for a menu bar. An application's document windows, palettes and dialogs all exist within the same visual space as the desktop and other running applications' windows. This creates a cluttered and confusing workspace.

As an aside, Microsoft seems to be becoming fond of putting the application menus in a little button somewhere on the toolbar (as seen in Windows Media Player 10 and in some early Longhorn screenshots), rather than in a menu bar. Indeed, the menu bar as a control is growing a bit weary as the functionality it exposes is often replicated in toolbars and other controls. It is necessary from time to time to access the menu and trying to identify the little button under which it lies can be tedious. I have spent a fair amount of time trying to find the menus in WMP 10. Which little arrow button is it? Furthermore, the horizontal flow of the menu orientation requires more nuanced motor function, a degree of difficulty that had previously been reserved for less essential sub-menus and a technique that the Windows User Experience Guidelines frown upon.

Icons and navigation buttons are not the only things that have gotten larger in the new version. The file Open and Save dialogs in Vista are almost complete Explorer windows themselves and consume much of the screen. Quite a bit of potentially useful information is presented, though the poor layout and large sizes are inappropriate for a transient dialog. This approach is quite different from the relatively small proportional sizing of such dialogs as recommended in the User Experience Guidelines.

A Pane in My Sidebar

Vista makes some modifications to the way Explorer presents metadata and task-related commands. Colored bars appear above and below the content pane of Explorer windows revealing information previously available in the task pane and tooltips as seen in Windows XP. Horizontal layouts such as this are increasingly popular in design. The influence is borrowed from the appeal of cinematic images and acceptance of widescreen formats for HDTV and other displays. The letterboxed look is used in many designs, such as television commercials destined for broadcast at less extreme aspect ratios, wide screen DVD releases, websites, and has even found its way to the Windows XP Welcome screen. The strong aesthetic appeal of horizontal layouts has its place in many contexts. Though the elongated bands can contribute to the design through color or graphical elements, they are rarely used to contain important information.

To the contrary, Vista's horizontal task pane and command toolbar are used to display important data and interface controls. The UI designers are sacrificing optimal layout in favor of style. The metadata displayed is composed of short lines of text. Such lists of data flow downward and it is much easier to move one's gaze (and head) down continuously rather than from side to side. It is for this reason that newspaper copy is laid out in narrow columns, a technique that is also being adopted by many websites. Such columns consume vertical space, leaving plenty of horizontal space to spare. To view any significant amount of the metadata displayed in Vista's horizontal task pane, the pane must be resized upward (since the text is flowing down vertically) stealing considerable screen real estate from the main view pane (which also flows vertically, competing for limited vertical space). The main view has the primary attention of the user and should be given as much space as possible in order to avoid excessive scrolling.

Since most common monitors have an aspect ratio that is wider than tall, and considering the vertical flow of data in the panes, it makes sense to maintain the longstanding vertical orientation of the task pane. The horizontal layout may have aesthetic value, but it simply is not a practical or efficient use of space. Thankfully, it seems that Vista's task panes can be vertically oriented off to one side of the window.

The hue of the color gradients appear to have some relationship to the content being displayed, serving as a visual cue to the type of folder, be it a Virtual Folder, search, or directory containing media. From looking at the screenshots, I cannot seem to discern the associations of the Explorer task pane colors to the current directory. I thought Virtual Folders appeared in green and directories containing images were orange, but then I'll find an Explorer window colored blue containing one of the aforementioned types which serves as a counter-example to the rule.

The non-neutral color palette employed by the task panes could be distracting and bothersome to users, outweighing its usefulness as a visual cue. Icon designers concerned with aesthetics must be wary of the potential array of vibrant colors upon which their icon may appear. The User Experience Guidelines wisely warn that users may find certain colors objectionable. Personally, I do not particularly like orange. However, due to the nature of my work, I spend a considerable amount of time browsing folders that contain images. This leads me to wonder if the colors for each association are user-configurable. Classic color versions of Mac OS and more recent versions of OS X offer Labels, which are a set of user-definable colors with particular associations that could be applied to icons for documents and folders. Adobe After Effects allows the user to customize the set of colors used to denote types of footage throughout the interface. (Both features are similar to the color labels available for calendar items Outlook 2002, though the color palette is immutable.) Perhaps such a mechanism can be provided for customization of the Explorer task panes.

I rarely use the task pane in Windows XP. The task pane contains the sections “File and Folder tasks”, “Other Places”, and “Details”. Often the tasks presented are readily accomplished in other ways. What is more is that I prefer to have the tree view offered by the Folders Explorer Bar open. Unlike the task pane offered by Web View folders in Windows 98, Me, and 2000 which appeared to the right of the folder tree, the Explorer Bar in XP covers task pane. Thus, to use both, the Explorer Bar must be closed to access the task pane and reopened to navigate the tree. Even more frustrating is that there is no keyboard shortcut to toggle folder tree. Shortcuts for other Explorer Bars do exist.

During the brief period in which I was running Windows Me, I had highly customized Folder.htt for the task panes of the Web View in order to improve my workflow. I placed commands useful to me in the task panes for certain types of folders, enhanced media and metadata previews, and hid the low bit-depth logos at the top of each bar. I was extremely disappointed that the task panes in XP could not be customized as such. All of the convenience and customization offered by Web View, to which I had become so familiar, had been deprecated in favor of XP's task pane.

This is yet another instance where Microsoft removed functionality that was present in an earlier version. A technology or function could be improved upon or deprecated (not enabled by default). However, functionality should not be eliminated in later versions of software as it prevents users from interacting with a system in a way that is familiar. The concept of Web Folders continues to be viable and has not descended into obsolescence. It offers many opportunities for tailoring workflow to a user's needs and branding in corporate environments. Indeed, the notion of a task pane persisted and gained higher precedence than the folder tree. Choices that could be made by the user were taken away. Fortunately, as mentioned above, Microsoft is careful to maintain past functionality, providing, in this instance, an option in the registry to restore the Web View folders.

Functionality should never be removed, only augmented. Occasionally, keyboard shortcuts may have to change, but the ability to access that item via a keyboard shortcut should not. Alternatively, providing means for a user to create a custom keyboard shortcut map is a great way to allow users to assign shortcuts to the items they know well while still taking advantage of new functionality. This approach is used in many Unix programs and is increasingly being adopted by Windows and Mac applications such as Adobe Photoshop. Nuance functionality of the UI should be preserved as well. An example of this might be clicking twice (not double-clicking) on a document or folder icon to rename it. As users become more experienced, they abandon menus and task panes in favor of such time saving shortcuts. Removing them would seriously impact usability for seasoned users.

Virtual Folders

In Vista, there is quite a bit of emphasis on searching with search boxes readily accessible in windows and the Start Menu and Virtual Folders to save searches. The concept behind this plays along with the goal of clarity by making it easier for users to find their documents. There is another side to this, too. If searching is so necessary to find documents, doesn't that imply that there is a lack of organization in the underlying file system? True, we have more and more documents with digital photos and music. In the case of the former, these images, especially when taken directly from a digital camera, rarely have descriptive file names, thus searching for a specific photo would require the user to input extensive metadata. The search facility also must be able to find items by their metadata, not just a file name. It is unclear that this is possible using the ubiquitous simple search boxes throughout the UI. Thus, the complex search window would be necessary after all, even though there is that tempting little search box sitting up there in the toolbar. Also, with the ability to so easily find documents, the need to file documents in an orderly manner in the file system's hierarchy is less important and may regrettably give users the freedom to become sloppy and disorganized… not exactly fostering clarity. Why be tidy when I (supposedly) can find a needle in a haystack at the drop of a hat?

I do my best to keep my documents and project files in rather tidy directory structures so that I can easily find a file within a few clicks. Not being a big searcher, I keep the Indexing Service in XP disabled. In the event that I do need to search for a file by its contents, I am not inconvenienced by waiting twelve seconds to search several hundred gigabytes of data instead of one second. I suspect, due to the nature of all of this searching and Virtual Folders, that disabling indexing will not be a practical option in Vista.

Searches may be saved as Virtual Folders, which are really just XML encapsulations of the query parameters. However, Virtual Folders appear to the user as containers in the file system with dynamic contents as the search is executed on each view. A natural language interpretation of the query for a Virtual Folder appears in the address bar of an Explorer window when viewing the results of the search. However, the address bar may not be large enough to display complex search descriptions or to reveal the full path to the Virtual Folder. There is also a bit of confusion that arises when the search summary is so long that it consumes the entire address bar. It appears that the control serves dual purposes: displaying queries or displaying paths.

As noted in other online articles, Virtual Folders can lead to confusion in that documents can be retrieved from them but not saved to them. A set of default Virtual Folders will replace the shell folders (magical folders with additional functionality provided by the OS) from older versions of Windows. This may add to the confusion as Windows is telling the user that the Virtual Folder is the place to find a certain type of item. Yet, the user cannot save similar items to that imaginary location. That is a lot of overhead to keep track of: the real folder hierarchy and a parallel overlaid Virtual Folder heterarchy. In some cases, it would helpful to be able to save back to a Virtual Folder, having the real file saved to the directory containing the subset of items represented in by that Virtual Folder. The problem here is that Virtual Folders do not necessarily subset a single real folder and may contain items from various locations. Perhaps a better solution would be to have all Virtual Folders be user-defined and not favor the use Virtual Folders in place of shell folders meaning that the default location used by the OS is always a real (or at least constant) folder.

The prepackaged Virtual Folders, which present icons grouping the folder's contents by metadata, are indeed a good idea. However, these should really be presented as parameters for a search query or as view options. For example, having such metadata searches as the default view of the Documents folder requires additional clicks for the user to descend into virtual tiers of the hierarchy before reaching the document for which they were looking. The initial view of any folder should be the actual contents of that folder. If the user then has difficulty finding an item, a change in the view or a search can be executed. Lost documents should entail more work to be found. However, if the user does know the location of a document, none of that extra effort should be required.

I foresee certain instances where it may be useful to save a search query. Since it is essentially a saved search query that the shell represents as an object in the file system hierarchy, it should be termed as such. Virtual Folders are not the best way to introduce a heterarchy on top of a hierarchical file system. Symlinks are better at accomplishing that task. Since it is reported that true symlinks will be a feature of WinFS and possibly the initial release of Vista, complex relationships can be created as such.

People tend to adopt metaphors of physical spaces in order to interpret information spaces. The lexicon surrounding web browsing uses metaphors that tie the experience to how we move around physical spaces. This is one of the reasons that the desktop metaphor scheme proved to be weak. Terminology originally intended to describe information spaces has been supplanted in common user usage by spatial metaphors. There is no real-world analog to a Virtual Folder so it may be difficult for users to understand the odd directional relationship that they have, when presented as folders (i.e. containers) in the file system, to real documents on the system. Shortcuts, such as symlinks, do exist in physical spaces, to a certain degree, and are more readily understood.

My understanding is that Vista's desktop search is based on MSN Search, which is mildly concerning. I'm not a user of MSN Search and when Microsoft announced their new search algorithm several months ago, I gave it another try. Sadly, for most of my searches, the top few results weren't even related to the topic on which I was searching. These were usually followed by several hits that were on topic but not the sort of resources for which I was hoping. If this is any indication of how Vista's desktop search facility will function, the situation doesn't seem promising. Of course, if I'm searching my own computer, the searchable set is considerably smaller and I would be reasonably certain that somewhere on my computer is the information for which I am looking, so maybe it won't be that bad.

If I were to use the search feature extensively, which I may need to do if only to navigate the file system in light of Vista's other changes, I would be sure to take a few moments to enter some metadata. Metadata is a beautiful construct. I just imagine that most users will not take the time (or because they are not used to doing so) to avail themselves of the resources it can provide.