Facets are fundamental: Rethinking information architecture frameworks
Reviewed by: Fred Beecher
This talk was very interesting. Abe’s argument was that information architects treat faceted classification as supplemental to topic-based organization and that we ignore or minimize non-topical methods of organizing information.
Abe cited two studies published in the Journal of the American Society for Information Science and Technology (JASIST) to support his point. A study by A. Tombros, et. al., shows that users use different elements of a page’s structure to determine relevance. Another study by C.L. Barry suggested several other non-topical criteria that users use to judge relevance, such as depth, novelty, credibility, and more.
From here, Abe began discussing what a facet is and what it is not—something that really interested the audience that spent much time discussing. The key confusion here is between attributes and facets. Abe said that an attribute is something that is inherent to the item and has a particular value. For example an attribute of a RAM chip for a computer is how much RAM it carries. This is a value that describes something inherent to the chip. Facets, however, are different. Abe says that facets tend to be more loosely defined and that they tend to represent human attempts to make sense out of the world. “Genre” would be a facet of literature, for example.
Abe went on to describe the structure of a facet, taking pains to point out that faceted classification does not preclude hierarchical or topical organization. He said that facets are composed of two components: organization scheme and organizational structure. Going back to the literature example, the genre facet would present groups of items arranged by topic (scheme) and displayed alphabetically (structure). Now this is only a rough example, and it assumes that the topic of a literature resource determines its genre. But I hope you get the idea.
What Abe wants us as IAs to do with this is to move beyond our current topical model of sitemaps and wireframes to one that is not dominated by topic—one where we think of objects and the information space they exist within.
Tagging and Beyond: Personal, Social and Collaborative Information Architecture (Or, Social IA Live: Five Challenges for Information Architects)
Gene Smith, Danah Boyd, Scott Golder, Jane Murison, Rashmi Sinha, Mimi Yin
Reviewed by: Chris Baum
Each presenter in this particular session played a clear role, which in the end made this thought-provoking exercise more like short sprints rather than a unified whole. Still, the presentations were interesting, and each provided a unique perspective.
Gene Smith (the Model) provided the context. Even as tagging and rating functionality have migrated from augmenting the user experience (Ebay, Amazon) to co-creating content (del.icio.us, wikipedia), we underutilize the wisdom of crowds. IA practice should encourage “good structure” rather than enforce it or expect the audiences to do all the work.
Scott Golder (the Network) discussed his ideas of social software as a rich network of disparate items and their relationships to one another. The lack of common experiences and language problems combine to make tagging for ourselves very easy (free association) and tagging for others very difficult. Over time, even we change so the associations we made previously are no longer valid. IA should help show the connections between ideas and contexts to ease these transitions.
Rashmi Sinha (the Pattern) noted that tagging encourages independence while allowing for easy aggregation. You “see” people based on their tags–while more abstract than blogs, your tags may be more telling of state of mind. IA should help migrate from software focused on existing social connections and most popular, most tagged, etc. to conceptually mediated connections that reveal the wisdom of crowds.
Dana Boyd (the Sociologist) challenged IAs to explore designs that start with the individual, then encourage the social by serendipity–don’t try to control behaviors, look at barriers as incentives, make the aggregate more visible, and utilize most passionate users to improve the system for the benefit of everyone.
Mimi Yin (the Interface) chided us for trying to force tagging and social software into the way we’ve always done things. She wants us to explore what we COULD do with software that capably enables users to understand information without having to actually re-experience it.
The vignettes stoked necessary ruminations and revealed different challenges that we face as we try to create effective social software.
The strict faceted classification model: an effective alternative to free-form tagging
Reviewed by: Fred Beecher
Travis began his talk with a discussion of how modern uses of faceted classification fail to leverage the power of the faceted model, that is, true orthagonality. In a strict faceted model, a given item can have only one value per facet. In many modern instances of faceted classification, each facet can have multiple values. Travis said that it may make sense for an item to have multiple values, but there are better ways to do this.
Travis’ example of desserts was appropriate. He applied two facets to a set of desserts, the first being “confection” and the second being “flavor.” So a dessert could be a pie, a cookie, or ice cream, and it could have a flavor of chocolate, cherry, or pecan. But wait… isn’t it totally valid to have a chocolate pecan pie? Yes it is. But Travis’ point is that if we are just going to apply multiple facet values to an item, we might as well just be using a simple tagging system.
Travis’ proposed solution was to create what I’m calling “multidimensional facets” (hooray for new buzzwords!), facets that are essentially groups of binary values. Obviously, not all facets should be multidimensional, for example, a “pie-cookie” would indeed be a highly improbable confection. However, chocolate cherry pecan is a perfectly reasonable flavor. In the dessert example, “flavor” would be the multidimensional facet. It would have three binary options, cherry yes/no, chocolate yes/no, and pecan yes/no. This allows the possible values of the multidimensional facet to be orthogonal among themselves, and by extension, to any other facets the item may possess.
Montreal, Paris, Dakar: Conducting an International Intranet Needs Analysis
Reviewed by: Jorge Arango
I’m usually drawn to presentations that promise to address issues of cross-cultural IA work, so I was attracted to this one early on. Isabelle Peyrichoux presented us with an engaging case study of an intranet project she helped develop for the French-Speaking University Agency, which required that she work with diverse team members in Canada, France, and Senegal. She outlined some of the challenges she faced along the way, including such potentially explosive issues as differing attitudes about gender roles in the workplace, and concrete tips on how to manage some of these situations.
While the case study was interesting, I was somewhat disappointed that the nature of the project—which was limited to three countries whose cultures are heavily influenced by France, and therefore one language—avoided some of the more common (and challenging) aspects of developing a cross-cultural site, such as bridging language barriers, or working across radically different cultures.
From Pace Layering to Resilience Theory: the Complex Implications of Tagging for Information Architecture
D. Grant Campbell, Karl V. Fast
Reviewed by: Donna Maurer
I commented at the end of this session that it was the best I had heard. While there were many great presentations, this was one of the best as it was so thoroughly considered. It was easily the best presentation on tagging because it moved beyond rhetoric and noise and examined what may really be happening.
The idea of pace layering comes from Stewart Brand (a previous IA Summit keynote)–complex systems can be decomposed into multiple layers, where the layers change at different rates. The “fast layers” learn, absorb shocks and get attention; the “slow layers” remember, constrain and have power. One of the implications of this model is that information architects can do what they have always done–slow, deep, rich work; while tagging can spin madly on the surface. If it is worth keeping, it will seep down into the lower layers.
Resilience theory explains the role of change in complex adaptive systems. Key aspects of resilience theory include:
- Change is neither continuous nor chaotic; it is discontinuous, patchy, and non-linear
- The destabilizing forces as important as stabilizing forces
- Constant yields indicate false stability
Grant and Karl discussed whether resilience theory is relevant as a way to examine the tagging phenomena.
How can information architecture address challenges to the Web in third world and developing contexts?
Reviewed by: Jorge Arango
When I first saw the title of Jason’s presentation in the Summit schedule, I knew this would be one I couldn’t afford to miss. Being from a “third world” country myself, I thought there would be much I could learn from a colleague’s experiences in a similar environment. However, I wasn’t expecting to be energized and encouraged to forge ahead in what can sometimes be a very frustrating environment. But this is exactly what I got: Jason’s presentation turned out to be an impassioned call to arms to those of us who are trying to bring coherence and design to the web in underdeveloped areas of the world.
Jason is originally from South Africa. He spent three years working in the UK, and recently returned to work in his country of origin. Like many others who’ve worked in “developed” countries and later return home to try to apply what they’ve learned, he seems simultaneously energized by a desire to improve things and somewhat frustrated at the harsh realities of the environment he lives in.
The audience was kept entranced as Jason showed slide after slide of South African Internet cafés, many of them in the poorest areas of Johannesburg. He also highlighted some of the challenges and obstacles he faces in his day-to-day work as an IA: a disconnect between the demands of customers and the realities of the infrastructure of the country, monopolistic—and therefore, expensive—internet access, a sense of inferiority (“why can’t we design like they do overseas?”), and more. All the while, Jason showed a deep empathy with users, and a clear desire to help improve things.
Unfortunately the audience was so wrapped up with the description of the current situation in South Africa that they started assailing Jason with questions and comments in the early stages of the presentation. While some of these sidetracks proved valuable, they caused the presentation to fall seriously behind schedule. As a result, the second half of the talk—in which Jason showed case studies of actual projects he developed in South Africa—was rushed, and eventually curtailed. The presentation had been scheduled on a pre-lunch slot, and many people left as Jason spoke well into the lunch break (he politely asked the audience if they minded this). Some of us stayed to the very end, and beyond: the conversation continued into the dining room.
There was a palpable sense of energy and wonder during this presentation. For many there, seeing the hardships many go through to access the Internet was an inspiring revelation. For me… well, as I told Jason after the session: “I feel like I’ve found a long-lost brother!”
Information Architecture for the Spatial Web
Matthew Milan, Michael MacLennan
Reviewed by: Jorge Arango
The “Spatial Web,” as I learned in this presentation, has to do with maps: the representation of physical space in a virtual medium in such a way that it conveys information in a useful way.
The presentation was roughly structured in three parts: the first was an overview of developments in mapping over the past four decades. This was followed by an explanation of how these mapping tools have impacted the way geographic data is presented online, with its advantages and limitations. Matthew and Michael then moved on to the third part, which was the core of the presentation: an argument for “user-centered mapping,” which focuses on the real-world needs of users.
This concept is characterized, according to the presenters, by web mashups that employ map data to help build meaning for users by relating the geographic information to data relevant to their lives. As examples, they used the much-blogged-about Google Maps / Craigslist mashup, and the “Gawker stalker,” among other sites.
Matthew and Michael also outlined clear and succinct points on how the online map-use experience can be structured more effectively:
- Know the user’s location
- Understand the user’s purpose
- Control hierarchy with scaling
- Filter with distortion or abstraction
- Label with effective symbology
While these points addressed concepts specific to the design of online maps, some of them (e.g., “Control hierarchy with scaling”) have broader applications in IA, and are therefore of interest to a more general audience.
Sorting in an age of tagging: How Information Architects can use sorting to address just about any research question
Reviewed by: Christian Crumlish
This year’s Summit had a recurrent theme of tagging and folksonomies. Monday was tag day but there was talk of tags all weekend. Rashmi Sinha of Uzanto started her talk by asking “Who’s sick of hearing about tagging?” before plunging in.
She began by discussing how tagging is cognitively easier and more natural than categorizing. She told us about “The man who could not sort.” A man was asked to sort email into three categories. He couldn’t do it, saying, “This is a waste of time.” It didn’t represent him. The test was torturing him and he finally gave up.
People really struggle with the idea of the one correct category to place each item in. (Even we information architects struggle with this—imagine how non-webgeeks feel!)
Tagging works because it maps well to the cognitive process of free association. Also, it’s fun. There is self-feedback, social feedback. You don’t feel obliged to balance your organizational scheme in the moment.
However, findability is still the missing bit. “Here’s where IA comes in,” said Sinha. “How do you add sorting, exploration, discovery?”
She compared sorting and tagging in terms of cognitive cost, richness of data, and ease of social aggregation:
|Higher cognitive cost||Lower cognitive cost|
|Richer data||Less rich data|
|Harder to aggregate socially||Easy to aggregate socially|
To improve existing categorization interfaces, Sinha recommended not whisking away an item as soon as it’s added to a category, to aim for flatter schemes, and permit nonexclusive categories.
She said, “Categorization is going to make a comeback. These are all fashions,” and the audience applauded. She recommended an essay called Don’t take my folders away! Organizing personal information to get things done, which talks about the feeling of satisfaction that comes from filing things in folders.
She recommended trying the classic IA exercise—card sorting—with tags. Ask subjects to brainstorm tags for Apple (the computer). They might come up with:
Then calculate co-occurrence and do hierarchical cluster analysis. Sinha pointed out that tagging works because the web has become social. She cited findings from a recent Pew Internet Report:
- Internet and email play important role in maintaining dispersed social networks.
- People use the internet to maintain contact with sizable social networks.
- People use the internet to seek out others in their networks when they need help.
- There is a concept of networked invidualism (connections are individual-to-individual).
She made an observation that may seem obvious but is actually worth really thinking about: People hang out on the web just for fun. Not just some people, 40 million people a day (in the United States). And not just men: 34% of men and 36% of women hang out on the web every day.
Tags make the web a shared experience through:
- Other social characteristics
- Social play
- Eavesdropping (my addition)
Sinha suggests that tagging allows for shared browsing, which is a way of socializing without having to deal with the kind of strife and flamewars that arise on email lists.
On the subject of tag clouds as a navigation device or form of menu she acknowledged that they “are not the future.” Menus are structured, stable over time, comprehensive. Tag clouds are unstructured, relatively unstable, and not comprehensive, but they let current stuff bubble to top. For example, many websites wanted to respond to hurricane Katrina. To do so, most companies had to add an explicit link to their homepage, but Flickr and Delicious didn’t need to do anything different. The community did it for them.
Comment from audience: Cloud shows relative importance, something easier to assess than absolute importance.
Sinha wrapped up by discussing some ideas for designing social systems. “Serve the individual’s selfish goal,” she recommended. Create a symbiotic relation (to avoid mob behaviors, the tragedy of commons). Think about when the individual should feel alone, when part of group, and how to encourage social sharing.
These systems don’t design themselves. They just seem to do so when those above considerations are given careful thought.
Sinha recommended we in the audience try these things:
- Create an account on MySpace
- Read Emergence and The Wisdom of Crowds
- Play a multiplayer online game (such as World of Warcraft or Second Life)
- Play with an API (Google maps API for example)
- Think about what is fun on the web (not just tasks and work)
A spirited question-and-answer session followed that invoked Erich Von Hippel’s research on lead users at MIT, Tom Coates’ article on tag drift (tracking the change in meaning of the Ajax tag on Delicious), and the search for social applications in the local space, beyond Dodgeball (one audience member mentioned a site called Socialight out of the Interactive Telecommunications Program (ITP) at New York University which allows you to add stories to buildings, anything from “this is a great coffee shop” to “there were three murders here in 1932, and everybody says this house is haunted.”
Rashmi Sinha’s talk was one of the best ones I saw all weekend, helping further my understanding of the viral popularity of tagging and the proper design of social software.
5 Minute Madness
Reviewed by: Jess McMullin
I love 5 Minute Madness. I hate 5 Minute Madness. For those of you who haven’t been to the Summit, 5 Minute Madness is an open mic session during the conference closing. Anyone can get up, and say anything, for up to 5 minutes. And they do.
I love 5 Minute Madness because for me it expresses the intimacy of the IA Summit–the conference as a whole is small, and there’s a chance to connect with industry luminaries and rising stars alike in hallways, bars, and yes, during a crazy open mic where anything goes. The openness of 5 Minute Madness, the sheer lack of structure in a discipline renowned for structure, the grandeur and banality and gratitude and inspiration that emerge in simple unrehearsed words–for me, it captures the Summit, and our community, like nothing else.
And that’s why I hate 5 Minute Madness–it means that the Summit is coming to an end, and that it’s another year until I’m back again, learning, growing, and exalting in the amazing experience put together by an amazing community. I’m looking forward to next year already.
And if you want to hear just how 5 Minute Madness went in 2006, CD Evans (one of my annual 5 Minute Madness highlights) has kindly posted an MP3 of the whole thing (13 mb).