A report from Day One of the CHI 2001 conference (Computer-Human Interface, pronounced kie, rhymes with pie), in Seattle. Some general comments: apparently, SIGs (Special Interest Groups, essentially 50-150 hands-on professionals discussing topics, led by a small panel) are seriously second-class; presentation of papers is the big thing. Some of the SIGs got bumped to smaller rooms after people couldn't get into the rooms where they were presenting papers and were screaming and pounding the walls. Go figure. I'm skipping the papers because most look pretty dull and how can you tell in advance if the authors can actually present or if they're just going to recap the paper?

I saw IBM's 200-pixel-per-inch, 3840x2400 flat-panel monitor. It's gorgeous! The demo cycled between x-rays, high-altitude photography, and similar stuff. It looked as sharp as, or sharper than, high-quality printed paper. I asked to see some text, and the guy brought up a full page in a three-inch-high window. It was completely legible! Amazing! $30,000!

Keynote:

Chairman Bill was the keynote speaker; I showed up early and got a good seat. I've never seen him speak in person before; he has an eerie, Buddha-like calm which is totally at odds with all the descriptions of his persona. Has he always been this way on stage? Most of his time was taken up by other people's demos (Reader; Priority manager; Tablet). He did acknowledge the irony of Bill Gates keynoting a conference on usability and computer interfaces ("Microsoft has had its share of failures ... the paperclip ...").

First session: Tips and Tricks for a Better Usability Test

Very well run. Covered three areas, about half of the 90 minutes was taken by people in the audience asking questions and making generally useful comments. Pretty much as well-done a SIG (Special Interest Group) as I could imagine.

International testing

• At a minimum, a native, living in the target country, should review the product/site.
• When you do usability testing for another country, that testing must be done in the target country, with natives of that country, in that country's language (even if the product is in English). Also check references for whomever you contract to do testing.
• Date formats: only formats in which the month is in letters, not numbers, are safe.
• 25% of usability issues are different across different countries (example was a user quitting after getting a useless error message when they tried to buy 2 kg of coffee from a site that wanted "Quantity ____ lb")
• To make a business case for I18N testing, simply translate your web site to that language and back and present the results to management.

Testing with experienced users

• They can help build usability specifications (no idea what these are; will follow up)
• Have experienced users watch novices do usability tests; the experienced users will make plenty of great comments/suggestions
• Test with data sets that are as big as your users typically use
• Differentiate between domain expert and product expert; try to get people who are both
• granting access to developers can be a better incentive than cash for getting experienced users to participate
• Differentiate between ease of use and ease of learning

Test Results

• Put a fridge with beer in the observation room to attract developers
• Use the KJ method to analyze results (collaborative method) Developers who observe just a few usability tests will be much less likely to create new usability problems
• Someone should observe the developers who are observing the test
• SIG presenters have stopped using questionnaires - not getting useful data and marketing kept abusing the data. Exception: interview the user while they are filling out the questionnaire; then burn it once they're gone
• Open issue: store usability results with bugs or separately.
• What severity scale to use?

Second session: Current Issues in Assessing And Improving Information Usability

Run by two academics, who do this SIG every year. Promising start - we shouted out possible topics to add to the list they proposed, then voted as a room on what four topics we were going to talk about, for 20 minutes each. Topics I voted for included "Does incremental usability improvement work?" and "Migrating Windows applications to the web." Unfortunately, neither made the cut. We spent a lot of time on stuff I wasn't very interested in. I heard a lot about card sorts, which are apparently a pretty common IA technique: write down lots of subjects on index cards, shuffle them, and then give them to users to organize into groups. Variations include: using category names; using blank category cards ("separate these 50 cards into groups, then name the groups"); having content owners and IAs do separate card sorts and then comparing the results; including actual content on the cards; doing heterogeneous card sets. infodesign.com is supposed to have good one-page instructions.

Book title "Minimalism since the Nuremburg Funnel"

In short, this session was borderline useful. Oh yeah, and we spent about an hour on (or rather, wandering off from) the second topic and didn't really get to the third or fourth. Annoying.

Third session: Practicing information architecture

Run by two consultants and an IA from CapitolOne. Pretty close to a total waste of time in terms of learning anything about the details of information architecture. Fairly informative (in the last 30 min) about the politics and nature of IA. What I got out of it was: Information Architecture is essentially a fancy version of library science: how to arrange and index information so that it can be found and retrieved. Extending the label past that, to include navigation, usability, interaction architecture, and other closely related disciplines renders IA a meaningless label. This SIG set the trend that anything having to do with IA included a half-hour argument about what IA is. Book title: Paco Underhill: Science of Shopping

Show of hands: who's working on a web application, not a web site? 90% from a room of ~100.

Also cruised the poster room and saw some interesting stuff: PARC survey of what people use the web for; Wichita State study of where on the screen 304 participants expected to find titles, links, ads, etc; Graphical User Profiles. will summarize later.