h engine crawled the Web.
The web site itself may have changed. You can check the current page or check for previous versions at the Internet Archive. Yahoo! is not affiliated with the authors of this page or responsible for its content.
Statement of Research, Teaching, and Service Statement of Research, Teaching, and Service

James A. Landay
Assistant Professor, EECS Department, UC Berkeley


Since coming to Berkeley at the start of 1997, I have led the EECS departments research efforts and curricular
improvements in human-computer interaction (HCI). The HCI area at Berkeley is now poised to be one of the top
programs in the world. My group, the Group for User Interface Research (GUIR), has received positive feedback
from the HCI community, and we have made intellectual impact through regular publication at the most selective
conferences. In addition, I have revitalized and added new HCI courses at both the graduate and undergraduate
levels, resulting in strong student interest in the area. The advice and support of Professors Larry Rowe and John
Canny, as well as the support of the EECS and Berkeley community have aided me greatly in this undertaking.

Research
Todays graphical user interfaces limit our use of computers in several important ways: 1) they restrict who can use
computers, 2) they constrain where computers can be used, 3) they do not let users communicate in the same ways
as they naturally do with other people, and 4) they leverage few of our innate perceptual, motor, and cognitive
abilities. My research is focused on overcoming these limitations through informal user interfaces. Informal user
interfaces are designed to support natural, ambiguous forms of interaction without much up-front recognition and
transformation of the input [11]. They are a reaction to the restrictive interfaces of current applications.

These restrictive interfaces are often a result of tradition. Computers have had the most impact in applications where
they have replaced humans in tasks at which humans are poor, such as performing complex calculations. I use the
term formal user interfaces to describe these precision-oriented interfaces. As computers become more powerful,
less expensive, and more widespread, we expect them to assist us in tasks that humans do well, such as writing,
drawing, and designing. Unfortunately, the historical design bias towards support of precise computation has
resulted in interfaces designed to facilitate structured input rather than natural human communication.

The primary tenet of my groups work on informal user interfaces is to bend computers to human modes of
interaction, not the other way around. I am a leader in moving the HCI field to look at this new direction for user
interfaces. We have built several systems with informal user interfaces and have completed several usability studies
that show the benefits of these representations over traditional styles of interaction. The systems we have built have
had impact on both the research and industrial communities.

Design tools
We are researching problems that will lead to computers that support the human properties of ambiguity, creativity,
and communication using speech, writing, gesture, and sketching. In many domains the formal approach clearly
presents an obstacle to effective task performance. One such domain that we have investigated extensively is user
interface design itself. This research led to our work on Silk [11], Denim [13], and Suede [9]. These informal design
tools, Silk for GUIs, Denim for web sites (see
http://guir.berkeley.edu/denim
), and Suede for speech user interfaces
(see
http://guir.berkeley.edu/suede
), allow designers to focus on the task at hand rather than disrupting their
creativity with the specific details of a particular look and feel. These tools have had impact in both the research and
practitioner communities. For example, Denim has now been downloaded over 4000 times and Suede is in use at
some of the leading speech UI development firms, including Nuance.

Much of our work on design tools has been motivated by an in-depth study of current practices in the web design
field [16]. The results of this study led directly to our design for Denim and more recently the Designers Outpost
[8]. Outpost is a tangible user interface that combines the best properties of physical and electronic interfaces.
Designers write design concepts with a regular pen on standard post-it notes. The notes are then placed in relation to
each other on a large electronic wall. Designers can easily move and manipulate notes due to their physical form,
while at the same time virtual operations are also possible, such as showing relations between notes, annotating
notes, or exporting the design representation to Denim). We have harnessed computer vision to track the physical
notes as well as manage the ink that is written on the surface of each note. Our evaluations with professional
designers and a demonstration of Denim and the Designers Outpost at the recent ACM1 Exposition in San Jose
generated excitement for the work.

Development infrastructure & tools
One common theme of my work has been to ease the development of informal user interfaces. We have followed
this both for designers who might use Denim or Suede and for computer scientists building new tools. For the later, we have developed a toolkit, called SATIN [5], for ink-based applications (see
http://guir.berkeley.edu/satin
) and
quill [14, 15], a design tool for creating gesture-based interfaces (see
http://guir.berkeley.edu/quill
). Previous
toolkits for pen-interfaces had little support beyond recognizing simple gestures. SATIN was designed to support the
needs of informal, pen-based user interfaces and includes a scene graph for manipulating and rendering objects;
support for zooming and rotating objects, switching between multiple views of an object, integration of pen input
with interpreters, libraries for manipulating ink strokes, and widgets optimized for pens.

Quill is the first design tool for gesture-based interfaces. Previous tools simply let a designer enter example gestures
for a recognizer to train with. In addition to support for managing and editing gesture sets, quill offers extensive
advice about the entered gestures to help the designer create a gesture set that is easier for a computer to recognize
and, more importantly, is less confusing to users. Quills advice is based on a computational model of human
perceived gesture similarity that we derived by running several experiments with human subjects. An early version
of quill has been released and is being used by other researchers in the HCI community.

Evaluation tools
Supporting design is necessary, but not sufficient for achieving usable computing systems. One of my senior
colleagues has repeatedly stated that measurement was what helped move computer performance forward. Taking
his advice, I recently started a major project on Web user experience measurement.

Research has shown that on the Web, many more test participants (20-100) are needed to find the usability problems
on a site than the 5-8 participants that have been traditionally recommended for testing desktop software.
Unfortunately, testing 20-100 participants is far too expensive and time consuming with conventional methods. The
goal of WebQuilt (see
http://guir.berkeley.edu/webquilt
) [3, 4] is to make it easy to automatically capture the
interactions that a test participant has when performing an experimental task on a Web site.

The important distinction of this work is that our proxy architecture enables evaluations to be carried out in a
remote, asynchronous manner without having to control the users web client or the target sites server, making
competitive comparisons easy. This methodology also scales to the sizes necessary to find all the problems on a site
as well as to make valid statistical conclusions about the data. Furthermore, we have harnessed our experience with
semantic zooming user interfaces to create a novel visualization of the captured traces. Using this type of
information, we can start to measure performance for Web sites and give designers tools that allow them to directly
discover problems and thus improve their designs. WebQuilt allows evaluations to take place at a lower cost and
with less work, thus furthering the goal of establishing evaluation standards for Web user interfaces.

Shared note taking
Any application in which absolute precision is unnecessary could benefit from an informal approach. We have
applied this idea to domains other than interface design. In particular, the NotePals project (see
http://guir.berkeley.edu/notepals
) has explored the use of informal interfaces for collaborative note taking [1, 10].
This system captures and provides access to personal notes, presentation slides, and documents of interest to a
workgroup. Its lightweight process, interface, and hardware distinguish it from previous systems. Group members
can share notes with each other by synchronizing with a shared repository that they can view using a desktop-based
We