Archive for February, 2013

Agile Designers: A Collection of Resources for Web Designers


Agile Designers by Webalys is a platform for web designers, currently in beta. Up to now already around 1,900 designers have registered and put up 740 resources. This is massive, considering that the service just launched on the 7th of November of last year. Reason enough to take a closer look at what's in store.

Much Ado About the Main Element

» Much Ado About the Main Element

Jeremy Keith discusses the introduction of a main element in HTML, and laments that it can only be used once per document.

Much Ado About the Main Element

» Much Ado About the Main Element

Jeremy Keith discusses the introduction of a main element in HTML, and laments that it can only be used once per document.

A Definitive Guide To The Android Carousel Design Pattern


One of the best patterns for browsing a small collection of featured products is the carousel. Unfortunately, many mobile app implementations do not offer an engaging or satisfying carousel experience and are not effective at driving conversions.

In this article, we’ll use the analogy of a real-world amusement park carousel to explain what makes for an authentically mobile user experience, and we’ll give you the design, the complete source code and a downloadable mini-app, which you can use today to add an enjoyable and effective carousel to your own app on phones and tablets.

How It Works

The carousel is fairly simple. The customer is able to view several images of products across a row, and they can swipe horizontally across the row to navigate to the next set of products. An arrow indicating the direction of the carousel’s movement is usually provided as a clue to how to interact with it. Optionally, the next set of products in the queue may be partially hidden, creating what we call the “teaser,� indicating that more content will be visible by swiping.


One excellent example of this pattern is the Amazon app’s home screen.

The carousel on the home screen of Amazon’s app is excellent. Larger preview.

This implementation of the carousel uses the teaser method to hint at the required interaction, showing only a small tantalizing glimpse of the naked CAT5E Ethernet cable, which is (and I have this from most reliable sources) completely irresistible to Amazon’s more impressionable customers, who can’t help but swipe across to see more of the naked cable and get to more of the content.

When And Where To Use It

The carousel is a fantastic pattern to show off featured items and relevant new arrivals. For example, new items matching the customer’s last search in their local area would be a sure winner.

In fact, consider using a carousel any time you have a small set of products (8 to 20 items) that are easily recognizable from just their picture. Augmenting the mostly visual content with a small amount of overlaid text is also sometimes effective. For example, in the screenshot below of the Pulse app on a 7-inch Android tablet, the carousel’s visual content is augmented with a semi-transparent dark-gray overlay, which provides the date and name of the comic. Without this overlay, the thumbnails of the carousel, especially in the first row, could be easily misinterpreted as belonging to a single comic.

The visual carousel content with semi-transparent overlay helps with comprehension in the Pulse app. Larger preview.

I am a huge fan of semi-transparent layers in mobile design. This example is particularly effective because the overlay on the thumbnails is semi-transparent and thus does not interfere with viewing the thumbnails, all the while augmenting the carousel experience in a way that is both visually appealing and informative. (Well, it does interfere just a bit, but it keeps interference down to a minimum while making best use of the small screen space.) The example also demonstrates that a carousel is an exceptionally great pattern for the large swiping gestures that small and large tablets invite.

Why Use It?

The carousel is an attractive and still fairly underused control for presenting visual information. It takes full advantage of the multi-touch gesture of swiping available on a mobile device. The carousel is easy and intuitive to operate and takes full advantage of the compressed real estate on mobile devices, where few words are needed to support the content.

Other Applications

One of the best features of a carousel is that it works well for a wide variety of device sizes and screen resolutions. This includes the ever-tricky horizontal orientation, for which the carousel works even better than in the vertical orientation.

The carousel adjusts to various screen sizes, and it works even better in a horizontal orientation. Pictured here is Amazon’s app. Larger preview.

Whereas the usefulness of traditional search results is severely hindered by the lack of vertical space, a well-designed carousel shines by showing off even more inventory.

Also noteworthy is the presence of the “See all� link, which points to the featured products.

The “More like this� link in this carousel navigates to a list of search results. Larger preview.

A “More� link is an excellent idea if your carousel has a small subset of items (8 to 20) that fails to meet the customer’s desires but piques their interest. In this case, the entire carousel control serves as advertising of sorts, enticing the customer to explore the relevant area of Amazon’s massive inventory.


As with any pattern, many implementations of the carousel do not feel quite right. One instance is NewEgg’s “Shell shocker� carousel:

NewEgg’s carousel has a few issues. Larger preview.

Some recommendations based on the UX problems in NewEgg’s implementation might help.

Make the Scrolling Smooth

To begin with, NewEgg’s carousel is structured more like the iTunes cover flow feature on iOS, with a large central element and two partial elements on the periphery of either side.

The cover flow screen in iTunes on iOS emphasizes the central element. Larger preview.

Like Amazon’s carousel, NewEgg’s can be swiped faster to advance more quickly through the list of products. However, NewEgg’s carousel moves very jerkily because of the structure of the central element, making it hard to see the intermediate states while scrolling — a major disadvantage. Seeing the two peripheral elements change is particularly hard — things hop all over the place, instead of smoothly swimming by the way they do in Amazon’s app. Higher-end and traditional carousels accelerate and decelerate smoothly and provide a pleasant, mellow, smooth ride. A carousel should be a high-end visual viewing experience that induces calm in the user, not stress. All parts of the control, including transitions, should behave accordingly and work together smoothly.

Indicate the Scrolling Direction

NewEgg’s carousel appears to be scrollable in both the left and right directions, causing a confusion: Is this carousel meant to represent a circle? Have I seen everything already, or do I need to keep scrolling? Amazon uses Android 4.0’s standard “blue parallax� visual treatment to signal the end of the line, a much better approach.

Amazon uses Android 4.0’s standard “blue parallax� to signal the boundaries of the carousel ride. Larger preview.

Just as a real carousel has papier-mâché horses that point you unambiguously in the right direction (you wouldn’t sit backwards on a horse now, would you?), so must your own carousel show which way the ride goes.

End the Ride Quickly

Good carousel implementations indicate the end of the list with the same parallax treatment seen at the beginning, and they present only 8 to 20 items, after which the ride is over and the customer can get off the carousel. Most importantly, the customer needs to exit the ride with the feeling of still wanting more. By contrast, NewEgg’s carousel seems to go on forever, so the customer does not get off until they feel bored (or, more likely given the jerky transitions, weak in the stomach). A much better approach would be to accompany the last element in the carousel with an obvious built-in “More like this� link, as shown below.

Show a “More like this� tile at the end of the carousel. Larger preview.

A “More like this� link can be used to jump into search results, which are more efficient for scanning large quantities of data and which are more likely to be relevant because of the sheer volume of items. Think of the “More� link as a premium combo ticket that grants admission to all of the remaining delights at the amusement park, conveniently presented after the cheap introductory carousel ride is over. Kind of puts a different spin on the entire carousel pattern, doesn’t it?

Make Sure Your Horses Look Amazing

No matter how smooth the ride is or how far or how fast the carousel goes, the best carousels have the best-looking horses, period. Make sure your horses (i.e. thumbnails) tell the story you want your audience to be told. For example, Amazon’s thumbnails are much better looking than NewEgg’s, although sometimes even the e-commerce giant screws up the ride, dropping thumbnails entirely:

Amazon’s carousel sometimes misses thumbnails. Large preview.

It goes without saying that ghost horses make for a terrible ride, even on Halloween (unless you’re the Headless Horseman). Which brings up my next point: some products are just not that visual, making them poor candidates for inclusion in a carousel. A classic example from my first book, Designing Search: UX Strategies for Ecommerce Success (Wiley 2011), is coffee — Peet’s Coffee to be exact. Even though Berkeley, California-based Peet’s makes some of the world’s most divine premium coffee, all of the thumbnails are identical pictures of a generic coffee bag and some spilled beans.

Some thumbnails, like these coffee bags from Peet’s, are not good candidates for the carousel pattern. Large preview.

Carousels are way more fun when the horses are all different; likewise, these generic bean-bag thumbnails are not good candidates for inclusion in a carousel pattern. A much more interesting horse would have been a close-up photo of a bean variety, or a map of a coffee-growing region, or a Tufte-style diagram of coffee taste attributes such as acidity, earthiness and body.

Even for products that have good thumbnails, presenting enough information can be a challenge. Including the right text in the individual tile is especially important for information scent; this could include key specifications such as processor speed, hard drive capacity and so on for the kind of technical gadgets sold by NewEgg. Picking an image resolution that can handle the amount of detail you are trying to show is also key. For example, can you tell what the item below is by looking at the small thumbnail and truncated description?

A small thumbnail and overly short description lead to “pogo-sticking.� Large preview.

Only by drilling down into the item do we see that it’s actually a screwdriver set. A larger thumbnail and better description would have helped a great deal and reduced the “pogo-sticking� (i.e. the frustrating navigation back and forth between carousel control and details page), which ruins the entire ride.

Having to pogo-stick between the carousel and details page ruins the ride. Large preview.

If the picture tells only half the story and you have to include a great deal of text, then you might find yourself having to increase the size of the individual element to the point that the carousel is no longer the best choice of presentation. For such items, think twice about whether you even really need a carousel, and whether a simple vertical list (more akin to a themed roller coaster) would make for a better experience.


Carousel code can be fairly straightforward. One way to implement an ultra-simple demo using Java is shown below. Plugging cute pictures of puppies into the carousel, you should end up with a mini-app that looks something like this:

Mini-app with a puppy carousel. Large preview.

First, we define how many items to show, and compute the width of the carousel item based on the screen’s width. (You may need to use more sophisticated code that computes a dynamic INTIAL_ITEMS_COUNT if you’d like to accommodate longer carousels for tablet devices.)

* Define the number of items visible when the carousel is first shown.
private static final float INITIAL_ITEMS_COUNT = 3.5F;

// Compute the width of a carousel item based on the screen width and number of initial items.
final DisplayMetrics displayMetrics = new DisplayMetrics();
final int imageWidth = (int) (displayMetrics.widthPixels / INITIAL_ITEMS_COUNT);

Next, for the purposes of this demo, we will create a static array of pictures. In a real app, this list would come from a database, of course.

// Get the array of puppy resources
final TypedArray puppyResourcesTypedArray = getResources().obtainTypedArray(R.array.puppies_array);

Then, we simply populate the carousel with items in the array and display it by overriding the onPostCreate() function. While onPostCreate() is mostly intended for framework use (according to the documentation), we can use it for this simple demo to simplify things a bit.

// Populate the carousel with items
ImageView imageItem;
for (int i = 0 ; i < puppyResourcesTypedArray.length() ; ++i) {
// Create new ImageView
imageItem = new ImageView(this);

// Set the shadow background

// Set the image view resource
imageItem.setImageResource(puppyResourcesTypedArray.getResourceId(i, -1));

// Set the size of the image view to the previously computed value
imageItem.setLayoutParams(new LinearLayout.LayoutParams(imageWidth, imageWidth));

/// Add image view to the carousel container

This code is provided free of charge and distributed under the GNU General Public License v3. See the README_LICENSE file for details. Download the complete source code for the app.

Try It Out

If you want to see how the carousel app runs, the completed puppy carousel mini-app is available for you to download and try out. To install it, I recommend using an app installer (I use the one made by FunTrigger, which you can get free from the Play market).

Here’s how it works. Connect your Android device to your computer. You should see the Android file-transfer window open automatically. Download the APK source file (download the entire package), and place it in the “App Installer� directory (you may have to create it).

Place the APK file in Android’s file-transfer window. Large preview.

Now you will be able to launch the app installer on your device, navigate to the right directory, and tap the icon for the APK file that you want to install.

Use the app launcher to install the app. Large preview.

After a few customary Android disclaimers, the app will be installed in the normal apps area on your device, and you can launch it from there.


The carousel pattern is a microcosm of mobile design: deceptively simple. It is also a somewhat new Android pattern, and pitfalls abound. Nevertheless, if you take the time to get it right, the carousel is a fantastic pattern for showing off featured items and relevant new arrivals, as well any items that are highly visual. It dovetails perfectly with the local context by showing new items that match the customer’s last search in their local area. Finally, with the right implementation of the carousel, you will be supporting touch gestures on today’s Android devices to the fullest extent, fitting more products on the screen and making them accessible with a natural swiping gesture.

When designing your carousel control, think of it as you would its real-world namesake, the carnival ride. Make sure your horses look amazing; indicate the scrolling direction; make the scrolling smooth; and end the ride quickly, providing an exit to more of your inventory. We’ve provided you with the Java code and a demo implementation of the carousel pattern, so now you have no excuse not to try it!


© Greg Nudelman for Smashing Magazine, 2013.

Cennydd Bowles on UX & Design: Looking Beyond User-Centered Design

User-centered design has served the digital community well. So well, in fact, that I’m worried its dominance may actually be limiting our field.

The terms “user experience designâ€� (UX) and “user-centered designâ€� (UCD) are often used interchangeably. But there’s an important distinction.

UX design is the discipline: what we do. Precise definition is elusive, but most attempts focus on experience as an explicit design objective.

User-centered design is a process: how we do it. Again the specifics vary, but usually form shades of the same hue:

  • Research. Immerse yourself in your users’ worlds to understand what they do and why they do it.
  • Sketch ideas that address these learned needs.
  • Prototype the most promising ideas to evaluate them more accurately.
  • Iterate through testing, repeating steps as required.

Other design processes

UCD is the dominant design approach within UX, so pervasive that some UX designers behold it as the Platonic ideal of design. Deviation from the UCD faith is even met with derision. A naive recruiter whose job specs aren’t explicit about direct user contact soon learns not to reoffend.

But other design processes are available. Jared Spool’s article 5 Design Decision Styles explores alternatives to UCD, including:

  • Self design, aka “scratching your own itch.â€� The designer acts as a surrogate for the audience. It’s convenient and quick, but clearly only reliable in narrow circumstances.
  • Genius design. Genius design has no first-hand research phase. To anticipate user behavior, the designer draws upon stockpiled experience, imaginative analogy, and psychological fundamentals.
  • Activity-focused design. Here, the designer addresses users’ primary tasks rather than any underlying needs. Tasks are derived a priori, from a logical interpretation of the domain, rather than from research.1

It seems arbitrary to regard these alternative design processes as inferior substitutes. Surely other modes can fulfill the broad UX mandate of creating experiences?

Weaknesses of UCD

UCD’s ascendancy deserves historical context. Its success came largely as an antidote to what preceded: the Wild West of the early web, dominated by Hey-This-Looks-Cool hackery. UCD offered rigor (or at least the perception of rigor; see Scientism below) that helped the immature web refocus on its audience. But that phase is long past, and the more experience I earn, the more flaws I see in UCD’s finery.


UCD simply takes longer than genius or self design. Clients typically identify research as the culprit, meaning the research phase is usually targeted when time is short. The UX industry has countered this variously through client education, seeking shortcuts, or by slipping research in without formal consent. But—whisper it quietly—some design research is wasted effort. For research to be valuable, it must:

  • be free from sampling or cognitive biases;
  • address issues that are central to the product;
  • offer genuinely new insight; and
  • be used to forge new ideas, not to validate predetermined decisions.

In these circumstances UCD is unparalleled, enabling breakthroughs other modes can’t. But I think UCD advocates overstate how often these planets align. I argue that genius design and iteration will often achieve better results in the same time.

Someone with experience as not only a designer but also as an attentive user has built up an unconscious repertoire of patterns and approaches that suit various contexts. As this library grows, it frees the designer from the need to research every problem.

The UX industry appears to acknowledge the relevance of genius design by its adoption of the expert review—a tool that epitomizes the approach—but often feels it has to prop this review up with user validation. It’s hard to escape the thought that the primary function of this redundancy is to retain the appearance of neutrality.

Negation of style

Among the UX community’s favorite quotes of late:

“[Good design] dissolves in behavior.� —Naoto Fukosawa
“The best interface is no interface.� —various
“Great design is invisible.� —various

At first glance, these are elegant statements of aesthetic intent: the user should never notice the designer’s influence. This “disappearing designer� motif holds self-sacrificial appeal, and for many interactions it’s great advice. I don’t want my tax forms to bear any trademark flourishes. However, when we extend this line of thought to its logical conclusions, these quotes start to look like mere slogans.

By negating the idea of a designer’s influence, we also negate the idea of style within the UX discipline. We’re saying that, done properly, it should be impossible to tell one UX designer’s work from another’s. There should be no signature elements, no philosophical movements, no overarching tenets except that of transparency.

The commoditization of designers that this idea suggests is troubling. Moreover, style is crucial for a creative discipline’s evolution. The best writers and architects—whose work, just like UX design, has function and engenders experience—have unmistakable styles. Throughout history, daring work from iconoclasts has sparked entire movements, and thus transformed creative practice. The transition between the Neoclassical and Modernist architectural eras, for example, wasn’t simply a case of replacing Doric columns with perpendicular glass. It was a total reframing of architecture and its values. Modernity usurped antiquity.

Is our form of functional art any different? In a system that deprecates style, is there room for a designer to pioneer entirely new approaches?2 If not, are we happy with the resultant ideological homogeneity?

Of course our designs must put users first. But there is never just a single way to meet user goals. Instead of trying to deprecate style, we should embrace it as a way to drive our practice forward and lend personality to the things we make. In a marketplace of bewildering clutter, products with a damn opinion are by far the most interesting.


Given its academic influences, it’s not surprising that UCD has been sold as a science. Empiricism runs through its discourse, to the unfortunate extent that the UX industry often oversells the certainty it can offer.

Scientism—akin to Colbert’s “truthinessâ€�—is the veneer of science where little scientific validity exists. While UCD is methodical, it is manifestly not scientific. There can never be a universal truth to design. Solutions applied in one context may fail magnificently in another, and the few governing principles (Fitts’s Law, the Gestalt principles, and affordance, say) give at best partial guidance. Some supposed laws, such as the “magical number 7±2â€� persist in ill-informed fringes of UX, despite being largely rebutted.

While researchers and designers can learn plenty from the scientific method, design simply does not yield to scientific analysis in the way its scientistic proponents wish.

To treat design as a science is to retreat to the illusory safety of numbers, where designers are mostly seen as agents of skewing the odds in your favor. This can start a race to the bottom as design gets less and less leeway. Weak leaders overtest in lieu of trusting designers to make decisions: it’s just a small step from there to the infamous forty-one shades of blue.


Finally, I’m concerned about the mindset that UCD can instill in its practitioners.

It’s unsurprising that a user-centered process can skew inexperienced designers’ loyalties away from business priorities. Some claim that this serves as counterweight to the business-first leanings of other employees. The argument strikes me as infantile. When a designer adopts simplistic, reductive arguments that ignore business reality, it undermines him. It limits his potential influence. Only the well-rounded designer who can fight for what’s right while accommodating business reality will be seen as a true leader.

What next?

I don’t expect UCD’s pre-eminence to change. Nor do I think it necessarily should. But a design community is most healthy when it shares a respectful variety of opinions. I don’t see that in the UX industry today, and I hope we can begin to appreciate the value of alternative design approaches.

The designers who will stand out in future will be those who are familiar with many modes of design. These designers may have a favorite, of course—and UCD is an excellent candidate—but they also have the versatility to draw on other processes when suitable. Perhaps they will even pioneer new approaches to add to our toolkits.

  • Copyright © 1996-2010 BlogmyQuery - BMQ. All rights reserved.
    iDream theme by Templates Next | Powered by WordPress