Conference Review: UIE Web App Summit 2007: Part III
Published: March 6, 2007
Content and Presenters: Highlights
Presenter: Sean Kane
In “Building a Great User Interface, the Netflix Way,” Sean Kane (Figure 39) spoke about the important role the Netflix culture plays in “delivering one of the most … elegant user experiences on the Web.” Through continuous iteration, the development team completes 2–3-week release cycles 21–24 times a year. “If we put out half the releases, we’d only be able to learn at half the speed,” said Sean. Their goals are to make Netflix user experience
- easy—Usability studies give us “metrics to understand what we’ve done and where we want to go. We do as much as we can for users” to allow them to accomplish their goals “in an automated way. Advanced features are there, but out of the way.”
- personalized—“Your experience is going to be different from your neighbors. We want to make sure you enjoy every movie you get from us.”
- innovative—“Innovation is just a fancy way of saying we’re making it up as we go along.”
- cinematic—“How can we make the site enjoyable for everybody?”
Figure 39—Sean Kane
Sean said, “Intuition is a great place for an idea to start, but at the end of the day, only a percentage of those ideas will fly. We fail a lot. We define metrics—criteria for success. What is going to make a feature successful? For any feature, it’s important to find out what makes it successful. Qualitative and quantitative testing give us a complete picture of what our members are doing. Getting people outside the company to look at this stuff is very important. Dissenting opinions can lead to conflict, so we’ll test all the ideas. If something doesn’t work, we know not to do that again. Always do what’s right, what’s going to move the business forward. The data tell us what users are doing, but we won’t know why. The qualitative testing tells us why. Paper prototyping can only get you so far. You need to actually see users interact with your product.” Figure 40 describes the development approach at Netflix.
Figure 40—Beyond intuition
The lifecycle shown in Figure 41 “takes longer than two weeks. Paper prototyping takes two weeks. Engineering takes two weeks. Designers have more lead time than design engineers.”
Figure 41—Development lifecycle
“When we first launched, … people didn’t necessarily understand clicking wouldn’t take them to another page.” On the page shown in Figure 42, “how do I get back to the starting point? People tended to get lost. So we tried adding a layer on the page,” as shown in Figure 43, “which took away the ability to get lost. … People struggled with the Back button a bit, but learned quickly.”
Figure 42—A successful page
Figure 43—New interaction
Figure 44 shows the consequences of not doing usability testing. “Without testing, you can’t have the proof that something works. Historical data may not be as valid as you thought it would be,” said Sean.
Figure 44—Consequences of not testing
Presenter: Thomas Vander Wal
Thomas Vander Wal (Figure 45) gave an interesting talk about tagging (Figures 46) and folksonomies (Figures 47 and 48), “Tagging in Your Web World,” which provided a comprehensive view on the topic. He spoke about the business value of tagging on the Internet, where tagging provides “an improved understanding of customers and market segmentations,” and on intranets, where tagging gives improved refindability and is a “cost effective means of building taxonomy.” Figure 49 shows people’s personal reasons for tagging. Thomas described various tag venues and user interactions with tags and showed many examples of sites that support tagging.
Figure 45—Thomas Vander Wal
Figure 47—Folksonomy defined
Figure 48—The value of folksonomy
Figure 49—Reasons people tag
Thomas told us, “A TagCloud is most often used for navigation, and it’s a really poor way of doing navigation. It starts breaking when people get distracted. … Tagging is most useful to people building collections, shopping, on intranets. On intranets, I see insane positive value. Different terminology is often a problem. Often there’s no taxonomy, so folksonomy helps. Tagging has a 1.8% adoption rate on the Internet; 60–70% on intranets. … If people focus on tagging for themselves, their tagging will have value for others. If tagging for others, they might guess.”
Presenter: Joshua Porter
Joshua Porter’s talk (Figure 50), “Learning from Social Web Applications,” explored the common characteristics of social Web applications like MySpace, del.icio.us, Digg™, Netflix, and Amazon.com—which “has 11 social features on each and every product page.” As shown in Figures 51 and 52, Josh has broken down social applications into their constituent elements and established some principles of social design. The primary element of social design is messaging; its primary principle, motivation. “Identify users’ primary motivation … and make satisfying that motivation the easiest thing to do.”
Figure 50—Joshua Porter
Figure 51—Elements of social design
Figure 52—Principles of social design
Independence (Figure 53) is a very important principle of social design. “When people act independently of one another, you can learn some really interesting things,” said Josh. “If independence is achieved, popularity is valuable.” Figure 54 describes the gaming of social applications.
Presenter: Rashmi Sinha
In her presentation on “Design Strategies for Web-Based Recommender Systems,” Rashmi Sinha (Figure 55) defined recommender systems as “systems that help people find information that will interest them, by facilitating social/conceptual connections or other means.” Rashmi provided a fascinating analysis of the attributes of this evolving class of Web applications and provided some useful design principles.
Figure 55—Rashmi Sinha
Rashmi surveyed the challenges of recommender system design, as shown in Figure 56. “Why should people tell you what they like? … Present recommendations to users at the right time. Users must trust the system, or they won’t buy your recommendations,” said Rashmi. Another challenge is establishing trust, as described in Figure 57, which also necessitates transparent system logic, described in Figure 58. “When you start, you have nothing. How do you get a thousand ratings? A search experience is much easier to design than a recommender system.”
Figure 56—Challenges of recommender systems
Figure 58—Transparent system logic
Rashmi elicited some design principles from early recommender systems, as follows:
- “Motivate participation.—We have to think of some creative way to get people to provide input. Offer an easy and engaging process that keeps the users from getting bored or frustrated [and provide] continuous feedback.”
- “Give users control.—Offer filter-like controls over genres or topics.”
- “Provide detailed information about recommended items.”—Shown in Figure 59.
Figure 59—Detailed information
Some recommender systems—like Amazon.com and Netflix—have become very popular, but “overall, recommender systems lost steam. They’re nowhere near as popular as search,” Rashmi told us. Rashmi explored some examples of existing recommender systems and determined their common attributes, as shown in Figure 60.
Figure 60—Common attributes
Rashmi observed that the user experience of recommender systems moves at a slow pace and is exploratory. “The way to the information is also rich.” She concluded her presentation with eight design principles:
- Make the system personally useful, even before receiving any recommendations. “Get people to express themselves” rather than thinking “What can I get out of the user?” Figure 61 describes this principle in more detail.
- Make the system participatory. “Think about how to make the size of individual contributions extremely small.”
- Make the participatory process social. “Just allowing people to upload pictures is not enough. Let people feel the presence of other people. Real-time updating makes it feel more like a conversation. Otherwise, it’s an extremely lonely experience.”
- Create instant gratification. “Provide personalized recommendations as soon as a user provides some input.”
- Cultivate user independence. “Gather input from a lot of people.”
- Provide access to the long tail of content and keep content fast moving. “Think in terms of helping people find things.”
- Expose metadata and make it linkable. Encourage people to “pass information along.”
- Provide balance between public and private. “People can be willing to share a lot if they get the right returns.”
Proceedings and Other Things
UIE thoughtfully provided proceedings both on a CD-ROM and in print. The CD-ROM included PDF versions of all the presentations given at the Summit—with the exception of those for a very few sessions that involved last-minute changes to the program. The proceedings on CD-ROM comprised
- the books of slides for all four of the all-day tutorials
- Book 1: “Deconstructing Web Applications”
- Book 2: “Designing Web Applications Using RIAs and Ajax”
- Book 3: “Usage-Centered Application Design”
- Book 4: “Measure Twice, Cut Once: Product Strategy and Planning Tools for Web Applications”
- the big books for each of the two days of conference sessions
- Book 5: “Web App Foundations”
- Book 6: “The Latest Perspectives in Web Apps”
Choosing which tutorial to attend wasn’t easy. So I was pleased to see I’d be able to sample the tutorials I didn’t attend by reading their slides on the CD-ROM.
Attendees received a separate printed book for each day of the conference. So, we didn’t have to carry around any more than we needed. For Day 1, attendees received only the book for the tutorial for which they had registered. The books for subsequent days contained the slides for all presentations—again with only a few exceptions for late changes to the program. So, if you wanted to be in two places at once, at least you had the slides for the presentation you missed.
Each volume of the proceedings was a thick Wire-O® bound book with sturdy, glossy covers, so I could easily open the books flat to take notes or fold them open to a specific page. The books for Days 2 and 3 opened with the day’s schedule. A blue title page at the beginning of each presentation provided an introduction and the presenters’ bios and set the presentations off from one another. Tabs with session titles would have been a nice addition. Each presentation page contained two slides and, on most pages, there was ample room for taking notes. However, on a few pages, slides with black backgrounds were larger and consumed almost all of the available space, leaving almost no space for notes.
Every attendee also received the UIE publication The Designer’s Guide to Web Applications | Part I: Structure and Flows, by Hagan Rivers.
All of this came in a nice canvas tote bag, with the conference logo on a convenient outside pocket, which also contained a welcome letter, a single-page agenda, an event newsletter, a list of nearby restaurants, and a map of the surrounding area. Our conference badges were in high-quality plastic badge holders, with securely attached lanyards, so there was no danger of anyone’s losing a badge when wearing it.
Monterey is an idyllic setting for a conference. Surrounded by rolling hills, this charming seaside community on the Monterey Peninsula offers plenty of good restaurants and sights to see. The Marriott is well situated in the heart of Old Town Monterey, just a few blocks from the waterfront and the famed Monterey Bay Aquarium. The hotel’s meeting rooms provided comfortable and spacious accommodation for the conference sessions. Though, on Day 1, there were tutorials on both the 2nd and top floors. The hotel elevators were slow and too few in number, so whenever we had to move en masse to or from the top floor, there was a serious traffic jam. The audio/visual team did a superb job, but the lighting on the speakers could have been better. UIE provided computer stations and a bulletin board for attendees’ use, and Wi-Fi was available for those with their own computers.
Hospitality and Special Events
The conference staff and volunteers were all very helpful and cordial, and the entire event was efficiently run. The organizers of UIE kept everyone well refreshed throughout each day. For the continental breakfasts and at morning and afternoon breaks, there were ample spreads of fruits, pastries, cheese, veggies, coffee, and bottled water (Figure 62).
On Sunday, January 21st, the first night of the conference, I organized a dinner for members and friends of the Interaction Design Association (IxDA). About 45 people gathered for stimulating conversation, made new friends, and had a delicious dinner at Montrio Bistro, which offers a fusion of European and American cuisine. After dinner, a number of us went on to the Crown & Anchor, a British pub with a nautical theme.
On Monday, UIE served a lunch of rosemary chicken in the Ferrantes Bay View Room on the top floor of the Marriott (Figure 63). From the windows, we could see a broad expanse of shimmering sea below (Figure 64). Monday evening, from 5:30 to 7:30 pm , the Summit Reception (Figure 65) took place in the Ferrantes Bay View Room, where guests enjoyed drinks and pasta and a convivial crowd.
Figure 63—Lunch in the Ferrantes Bay View Room
Figure 64—The view from the Ferrantes Bay View Room
Figure 65—The Summit Reception
Josh Porter’s wife, Alana, and their baby Tessa, with David Malouf and his baby Caleb at the Reception (Figure 66). We saw Tessa take her first steps on the last day of the conference.
Figure 66—The youngest attendees
The UIE Web App Summit was a truly great conference. If UIE decides to make this an annual event, I highly recommend it to anyone working in Web application design. My hope and expectation is that the conference will evolve along with Web application design practice, making it the Web application UX conference for novices and experts alike.
Photographs by Ron Yoder, for User Interface Engineering, and Pabini Gabriel-Petit.