CHI'94 Trip Report

by Jakob Nielsen on June 1, 1994

The annual CHI (computer-human interaction) conferences are the main meeting for user interface professionals. CHI'94 took place in Boston in April and was the largest such conference ever, with 2,632 participants (see the table for a list of the organizations sending the largest number of people). My overriding conclusion from the conference is the emerging trend toward making user interfaces engaging and emotionally appealing. This experiential perspective is in contrast to earlier conferences which were dominated by a cognitive science perspective on the use of computers for problem solving and skilled task performance.

Organizations with more than 25 participants at CHI'94.
AT&T, GIS, Bell Labs 71
Hewlett-Packard 70
IBM 65
Sun Microsystems 58
MIT, Media Lab, Lincoln Lab 50
Apple Computer Inc 49
Digital Equipment Corporation 43
Microsoft 43
Lotus 37
Xerox, PARC, EuroPARC 37
Bellcore 36
Carnegie Mellon University 29
University of Toronto 26
U S WEST Technologies 26

The new approach to user interfaces was best expressed by Tim Skelly from Microsoft Research who implored us to remember that "people are subhuman," by which he meant that we are all motivated and engaged by more basic instincts than the supposedly "higher" levels of cognition. Skelly participated in a panel entitled "What HCI Designers Can Learn from Video Game Designers." This panel title was clearly a misnomer since video game design is a special case of human-computer interaction design, meaning that game designers are user interface designers. Anyway, this panel was one of the best events at the conference. Another Microsoft researcher, David Thiel, discussed his work on use of sound in games and more traditional user interfaces. In games, sound can be used to indicate upcoming events (like the approach of an enemy), allow users to assess the current system state in minimum time, and provide instantaneous feedback (for example, different sounds can be used for swinging a sword depending on whether you miss or hit). In video parlors, very intense sounds have to be used to rise over the ambient noise, but in office settings, Thiel was working on sounds in the 4 dB range that would sound somewhat like the head moving on your harddisk or the keys clicking on your keyboard. Such low-intensity sounds provide valuable feedback in current systems, and it seems likely that better sound effects will improve user interfaces as long as they are not too intrusive.

Another indication of the trend toward engagement increasing in importance compared to problem solving was the job opportunity message board. Instead of the usual, slightly boring 8.5x11 resumes, several job applicants had posted multimedia resumes. The most creative resume was probably the wooden box hanging on the message board. The box had a door with a handle that was begging to be opened (an affordance if ever I saw one), and upon opening the door you found a pile of resumes and heard a recorded voice tell you about the person and his job search goals. Another observation from the job board was that the announcements of job openings took up twice as many tackboards as resumes posted by job applicants, signifying the continued growth and healthy career prospects for the user interface profession. Many of the job openings were with the usual suspects like Apple, Microsoft, Sun, and the telephone companies, but there were also many new employers looking for usability experts, including several brokerage companies and a railroad company.

There are several reasons for the increased emphasis on engagement in user interfaces. One important trend is the beginning dominance of home computing, with more than half of all the computers sold in the United States this year going into the home. Many home use interfaces are clearly designed to be engaging; more than necessarily highly productive. One example shown at the conference was the new Creative Writer program for 8-14 year old kids. In this application, the alert dialog boxes are accompanied by a loud barking sound effect and a picture of a watchdog called Spike, and there are special menu commands to show jokes or fun animations that do not have anything to do with the user's task. One of the problems with engaging interfaces is that they can be more highly dependent on local culture than cooler and more abstract interface designs. For example, the Creative Writer spelling bee icon (shown in the icon bar below) would score low in international usability since very few non-English speakers would associate a picture of a bee with the functionality of a spelling checker.

The Undo-Egg icon from Creative Writer

Icon bar from Creative Writer
Multiple-level Undo in Creative Writer. Here, the user has undone four levels of prior actions.

One element I particularly liked from the Creative Writer application was the "Undo Egg" icon . Interface designers have struggled for years to come up with an appropriate icon for Undo. Prior suggestions have included a car backing up and an Uh-Oh Face, but none have been successful. Research in icon usability has found that icons depicting objects are easier to understand than icons depicting functions. Concrete items are easier to iconize than abstractions, and of course, Undo is an abstract function, making it particularly hard to depict in an icon. My informal user tests with three users indicate that the meaning of the Undo Egg is not immediately intuitive either, but that it is very memorable. When you click on the icon, it changes to a whole egg (you can put it together again, even without all the King's horses) accompanied by an egg-shell sound. The design even addresses the issue of multi-level undo that has been struggled with in mainstream productivity interfaces for a long time. Creative Writer allows 20 levels of undo that are visualized by a series of eggs as shown in the figure: for every action you undo, one more egg heals, and for every redo, it cracks open again (with a bird squeak sound).

Starfire texturemapping sequence
Sequence of detail shots from SunSoft's Starfire film showing a texture map of a person being applied to a mannequin object.

Engaging communication is not just relevant for the user interface itself but also for the process of improving the development practice of organizations. No number of dry talks or memos can convince development managers as much as a short time spent in the usability lab observing real people struggle with their precious product. Similarly, video envisionments of concepts for future user interfaces get people much more engaged in the ideas than words do. Bruce Tognazzini from SunSoft gave a presentation about a new film called Starfire he had produced to show possible computer use in the year 2004. The film shows a large number of advanced interaction techniques across a variety of platforms, from a full-desk workstation to a portable notepad. Instead of a technology-focused presentation of the new interface ideas, the film presents them as natural parts of a scenario in which a user prepares for and participates in a strategy meeting. For example, she wants to show a presentation of a planned car model with an attractive male model clipped from another car commercial. She does this by picking out images of the person from the commercial (having the computer recognize the person as a 3-D object as he moves around in the 2-D image), and texture mapping the resulting graphical model onto a mannequin in her own presentation. The figure shows a detail from the film as the user "pours" the texture map from the computer model to her own presentation. Having a texture-mapped mannequin, the user then proceeds to position it to enhance the looks of the car presentation.

In contrast to some other video prototypes of future user interfaces, the Starfire film was deliberately limited to technology that might be reasonably expected in the next ten years, and as it turns out, initial research results on user manipulation of a computer-modelled person were included in the CHI video show. Christian Breiteneder and Simon Gibbs from the Universities of Vienna and Geneva, showed so-called interactive video actors, where an actor would be digitized in multiple positions, that could be combined on the computer to allow, e.g., the actor's head to look in the direction of the cursor. Currently, these interactive video actors were not fully flexible 3-D object models, so they could not be used to support the exact interactions shown in the Starfire film, but they do show that Starfire is not as far-fetched as some think when they first see the film.

Photo of FingeRing user's hand


The trend toward more engagement in user interfaces is also related to the progressively more personal and portable nature of some computers. It is quite common in computer workshops these days to see more people taking notes on laptop or even handheld computers than use paper notepads. In her keynote address, Joy Mountford of Apple noted that this trend changed the users' role from that of "worshipper" in front of a set of control panels to that of "wearer" of the technology. The ultimate in portable interfaces so far is probably the FingeRing system shown at CHI by Masaaki Fukumoto and Yasuhito Suenaga from NTT in Japan. FingeRing was billed as a full-time wearable interface and consists of five rings that in principle could be worn permanently as shown in the photo (though I think the NTT engineers should consult with a fashion designer to improve the looks of the rings). The rings contain accelerometers that can detect sudden changes in finger speed, meaning that a user with a ring on each finger can "type" by tapping the fingers on any somewhat firm surface. By chording multiple finger taps together, users can input 31 different symbols with FingeRing, meaning that it can be used for text input as well as for many forms of menu-oriented or selection-based dialogue techniques once a matching output mechanism has been worked out.


Much of the work presented at CHI'94 was still related to traditional user interface issues. For example, Francesmary Modugno and Brad Myers from Carnegie Mellon University had addressed the problem of user-level programming in graphical user interfaces. Their solution combined a visual programming language with programming-by-demonstration to overcome the problems inherent in textual scripting languages that look very different from the interface which the user is accustomed to.

Most current user interface standards employ a hidden clipboard to store the user's objects between cut and paste editing operations. The fact that the clipboard is hidden is known to cause user anxiety and also limits the flexibility of what can be done with the clipboard. To overcome these problems, Eric Bier and colleagues from Xerox PARC had designed an interface where the clipboard was represented by a so-called toolglass. A toolglass is a transparent (but visible) mini-window that can be moved by the user to locations where clipboard contents need to be sucked up or deposited. By enlarging the toolglass, the user can achieve the effect of a multi-object clipboard that temporarily stores several objects. In a related study, Paul Kabbash and colleagues from the University of Toronto had user-tested two-handed use of a toolglass and found that it was 16% faster than traditional one-handed interaction techniques for a combined drawing/color change task. The toolglass is suited for two-handed input because the user's dominant hand can control the command cursor with the mouse while the other hand moves the toolglass around with a trackball.

The CHI conference thus showed progress on several of the well-known problems in current user interfaces as well as exciting new trends toward experiential interfaces. There were not quite as many results of immediate interest to everyday development projects, though a few papers did cover improved usability methods. For example, Jürgen Koenemann-Belliveau and colleagues from IBM extended the traditional critical incident approach (look at events that stand out during user testing due to major interface catastrophes) to a "critical threads" method where one considered a collection of episodes that together explained a user problem. Also, maybe the most important result in the entire conference was shown in a poster by Tom Landauer from Bellcore. By comparing products released over the years he had found that the measured usability of released products increases by an average of 11% per year even though proper usability engineering methodology can improve products by about 50% on the average. The difference in these two numbers indicates the huge potential for enhanced productivity if more projects would follow recommended user interface practice. (Later, Landauer published an in-depth discussion of these issues in his book The Trouble With Computers: Usefulness, Usability, and Productivity.)

Share this article: Twitter | LinkedIn | Google+ | Email