forUse: The Electronic Newsletter of Usage-Centered Design
#19 | January 2002
= = = = =
* Watch for us in ACM interactions! We’re being featured in the upcoming design issue.
* Object Modeling and User Interface Design (Addison-Wesley, 2001)
(http://www.amazon.com/exec/obidos/ASIN/0201657899/conslockltdsoftf) is a great book
edited by Mark van Harmelen. It includes our phenomenally popular paper (130,000
downloads!) “Structure and Style in Use Cases for User Interface Design.”
* Performance-Centered Design Platinum Award of
Excellence 2001 – STEP 7 Lite
1. Instructive Interaction.
Instructive interaction is a way to make novel user interface features self-teaching and instantly usable right out of the box.
Instructive interaction is solidly grounded in a model for single-trial learning. Most learning is a trial-and-error process or requires repeated performance and reinforcement. Single-trial learning is a special case where, under just the right conditions, once is enough. The best known form of single-trial learning is the paradigm popularly known as the Sauce Béarnaise Syndrome. If you eat something new, then some hours later experience acute nausea, after that, just the smell of that something will trigger a feeling of nausea. Short of smell-enabled touch screens in cafeterias, it is hard to imagine much use of this in user interface design.
Anticipatory learning is another, more broadly useful form of single-trial learning. In this learning model, the user (1) sees an unfamiliar feature and recognizes it as novel, (2) anticipates its function or behavior by making a guess about how it works, (3) tries the feature out, and (4) discovers that the guess was right. Bingo! The user has learned how to use the new feature. This pattern of unfamiliarity, anticipation, and immediate reinforcement for correct guessing can be exploited in innovative designs that are original but still understandable on first encounter.
Instructive interaction does require careful design with a considerable amount of discipline. Novel features must be devised such that the user’s guesses will invariably be right, and the user must feel confident that trying things out is safe. But the payoff to user and designer alike is tremendous. Designers can confidently break with convention when needed for more efficient, more dependable, or more flexible use. Users can compress the learning curve and make efficient and effective use of even completely unconventional software.
We’ve been using instructive interaction in our own design work for several years. We have developed an array of visual and interaction designs that make use of it, and it’s a key part of the breakthrough work that led to our winning the 2001 Performance-Centered Design competition (http://foruse.com/pcd/). We first introduced the idea in this newsletter in a brief note in the February 2000 issue (http://foruse.com/newsletter/foruse2.htm) but have not been able to give the topic the comprehensive treatment it warrants. Now we have completed a blockbuster new paper that details instructive interaction, with dozens of specific techniques and numerous practical examples. Check it out at (http://foruse.com/articles/instructive.htm). It’s a monster at over 500K for the .PDF file, but worth every byte.
2. Coming Attractions
Over the mid-winter hiatus, we have been putting the finishing touches on a complete redesign of the foruse.com Web site! The official go-live date is still a closely guarded secret, but we’ll be making the switch-over in time for the next newsletter.
The site was first created in 1999 as part of our contractual obligation to Addison-Wesley, publisher of our book, Software for Use. Originally designed as a simple six-page site to augment and support the book, it just grew and grew to the point that the original “architecture” no longer makes sense. Those of you who are regular visitors to the site know that there may be a lot of information there, but it is not always as easy to find as you might wish. The organization is not always transparent and the look-and-feel is beginning to look and feel a bit dated. One might say that the foundation is beginning to crack under the weight of three years accretion of materials and features. It has long been clear to us that neither cosmetic surgery nor simple page shuffling was going to be enough, so we gritted our teeth and started over with a complete redesign from scratch.
We tackled the rework entirely through usage-centered design. Equipped with several years of page-view and click-through stats from several hundred thousand visitors and guided by hundreds of emails and feedback forms from you, we had a pretty clear idea of how the site was really being used and by whom. We identified and modeled user roles, then prioritized them by frequency and by importance to our business. We developed task cases to support these roles and ranked them by importance to users, to the business, and by historical and anticipated frequency. We then devised abstract content and navigation models along with a basic visual design and navigation scheme that would support users and their tasks. We also set certain project-specific design criteria to guide the detail design. For example, the new site is designed to reduce user clicks and increase the likelihood of success for focal task cases.
The way the design emerged and evolved as the new site was built was an interesting process. In fact, we will be writing up the project as a case study to share with all of you. Watch for it and the new site, coming soon to a browser near you!
3. Shuffling Cards for Fun and Profit
We develop a lot of our working models, including those for the new Web site, on index cards. We brainstorm lists of candidate user roles or task cases directly onto index cards, then sort, combine, or refine them until we are satisfied with the set. Once we have a working set, it’s easy to rank them in terms of expected frequency or importance just by shuffling the stack of cards in order.
It can be useful to get multiple perspectives by having different people sort the stack: for example, all the members of the modeling team, or a user, a manager, and a business analyst. After each person finishes sorting, we flip the stack over and number the cards on the back using a different color or location for each rater. This way raters aren’t distracted or biased by seeing how others sorted the stack. Sometimes we print out and distribute multiple copies of the set of cards.
The fun begins when we combine and compare rankings. It’s easy to get a combined ranking just by adding up the numbers on each card, then resorting the stack in order of this “sum of ranks.” One legitimate question is whether much stock can be put into this combined ranking. If everyone ranks the tasks in a completely different order of importance, the average (or sum) of the ranks is not very meaningful. A measure of concordance--or agreement among raters--can indicate how objectively similar the rankings are. We use Kendall’s W, computed with a handy little Excel spreadsheet, to check the level of agreement, which varies from 0.0 (completely random) to 1.0 (perfect agreement).
Just putting the numbers into neat columns for side-by-side comparison often reveals interesting patterns, such as whose perspective is wildly out of line with all the others. We don’t always try to resolve these issues, but discussing what might account for the differences can be enlightening.
Spreadsheet template? Did we say template. We’ve had a lot of requests for the spreadsheet template, but have always been reluctant to pass around something that was just hacked together for personal use and never meant for distribution. However, if you do this kind of modeling in your own work and are prepared to use the template on an “as-is” basis, you can download it at (http://foruse.com/resources/tools/ranking.xls).
4. Conference Update.
Thanks for your proposals for sessions at forUSE 2002, the First International Conference on Usage-Centered Design (25-28 August 2002 in Portsmouth, New Hampshire).
It’s not too late to submit a proposal, but the of 15 February deadline is fast approaching. We’re looking for regular sessions, panels, tutorials, demonstrations, practitioner reports, research reports, and teaching experiences. We’ve updated the Call for Participation at (http://foruse.com/2002/cfp.htm) to provide more details about sessions and benefits. Presenting at forUSE 2002 pays off in free registration, publicity, and stipends up to $1500. Details on the conference are at (http://foruse.com/2002/).
5. Train with the Best
We’ll be in Portsmouth, New Hampshire, the week of 6-10 May 2002 to teach another public training seminar on usage-centered design. The seminar is hands-on, practical, and intensive and will be covering everything from agile card-based modeling techniques to instructive interaction, so don’t miss it. Details are on the Web at (http://foruse.com/seminars/).
And if you can’t come to us, we’ll come to you. We offer custom-tailored seminars for clients anywhere. For details, send an email with information about your group and its training needs to: (mailto:email@example.com).
= = =
To unsubscribe, send email with the word "unsubscribe" (no quotes) in the subject and message to: <mailto:firstname.lastname@example.org>