CHI 2007

CHI 2007 Conference Trip Report

May 28, 2007

In late April 2007, I travelled to San Jose, California for the 25th annual SIGCHI Conference on Human Factors in Computing Systems (CHI 2007). This is the premier conference for Human-Computer Interaction research and I was privileged to attend and to present (as a poster) my research.

The conference was a four day event with over 2600 attendees from around the world. There were many courses, panel discussions, paper presentations and hospitality events to keep everyone busy. In fact, there was so much going on that I quickly became overwhelmed with the scope and eventually resigned myself to absorbing very little of the overall content.

In this report, I will record my impressions of the conference and some of my half-baked ideas that came from attending sessions and informal discussions at the conference.

Day 0 - Sunday

There is not much to say about the travel day, other than:
Late Saturday night + very little sleep + very early flight + long flight = dead tired.

Having little sleep Saturday night was, of course, entirely my fault. But I couldn't miss a chance to see The Weakerthans, could I?

Day 1 - Monday

The conference started with the Opening Plenary talk given by Bill Moggridge. His talk was a series of case studies on the design of products taken from his new book Designing Interactions. One by one, he showed how the design of the product came, for a large part, from nowhere based on intuition. The main point of his talk was not that great ideas suddenly appear from out of nothing but that the whole design process serves to accumulate knowledge to the point where a leap of intuition, a crystallization as he called it, happens to occur. Some of the case studies were a bit tedious, however, overall it was a good talk and I am interested in reading his book.

For the rest of the day, I attended talks. Some highlights were:

An exploratory study of input configuration and group process in a negotiation task using a large display was an investigation into how competition and social dominance are affected by how many many mice are available to interact with a large display. They found that there was better discussion with a single mouse. People collaborated more with one shared mouse versus one mouse each.

White rooms and morphing don't mix: setting and the evaluation of visualization techniques. This was a very interesting talk with a nonintuitive result: You would expect that a physical environment with a lot of movement and noise would be distracting and would affect the performance of using a visualization system. They found the opposite result, that users performed better in a dynamic environment (a coffee shop) than in a white room with all distractions removed.

They suggested that dynamic environments tend to change our way of perceiving (and perhaps our cognitive behaviour as a whole). In particular, visualizations that use motion offer better performance when viewed in a non-static environment. The environment somehow primes the user for that sort of perception.

Many people seem to take advantage of this by choosing to work in a coffee shop for a short period of their day. Personally, when I am stuck or unmotivated it helps to sit in a coffee shop and work there for an hour. This different and dynamic environment brings out different thought patterns. However, it may just be the change of scenery or comfort level that explains the effect. (I think we are social beings who are, on some level, more comfortable in a social situation, despite any social hangups any individual may have. At least more comfortable than in a sterile whiteroom.) The talk reminded me of how Richard Feynman would work on very hard physics problems while sitting in a strip club in Pasadena. It seems like the last place a heterosexual male would be able to concentrate but it worked wonders for him.

Move to improve: promoting physical navigation to increase user performance with large displays. This talk compared physical navigation (ie walking and touching a display) with virtual navigation (ie indirect interaction with, say, a mouse). It was a nice, thorough study. The presenter made an insightful comment motivating interaction techniques for large displays: You will never have a large enough display for all of your data. Even if the display covers the entire wall or an entire room, the data will inevitably grow larger, faster.

How pairs interact over a multimodal digital table. This study investigated how voice commands to a computer system are used both as commands and as part of the conversation when pairs are collaborating. The subjects integrated voice commands into their conversation 98% of the time. This is interesting because it demonstrates that voice interaction is different than other input. It also shows that voice interaction is ideally suited for collaborative systems because users can, and do, dual purpose their voice commands, potentially becoming more efficient in the process.

After the day's session, there was reception that I skipped to work on my final essay for a course on Formal Logic. I sure wish I completed that before I left for the conference.

Day 2 - Tuesday

On Tuesday morning, I skipped the early morning awards session (although, I am sure they would have been quite interesting) to finished my Formal Logic essay. With that done and submitted, I headed to the Gaze and Eye Tracking session to see if I can begin to understand what this stuff is all about. We have an old eye tracker in the lab and there is some discussion of purchasing a new one, so I was interested in what these things were actually used for in practise. The first two talks were about using the eye tracker to improve the layout of search result pages. This is interesting but not really applicable to my interests.

The third talk in the session (EyePoint: practical pointing and selection using gaze and keyboard) was the presentation of an interaction technique based on an eye tracker. The major problem with using an eye tracker for interaction is what they termed the midas touch problem, where it is very hard to detect when the user wants to select and when they are just looking (which is analogous to the story of King Midas who wished for the ability to turn everything he touched to gold). The presenters' solution was to use the eye tracker for pointing (a natural extension to looking) and the keyboard for selection. It is a really simple idea but had never been done before because using eye trackers as an input device was previously limited to helping people with full paralysis. The presenters motivated the need for a general purpose eye tracker-based interaction by predicting the inclusion of eye trackers on consumer laptops in the very near future. Very interesting indeed.

Later on, I visited the LC Technologies Eyegaze Systems booth and tried out their eye tracking system. I was quite impressed by the accuracy and ease of use. I was also very impressed by the fact they are a small company that takes great pride in using their technology to help people with disabilities lead a better life.

Furthermore, I was surprised to discover that it was very hard for me to use a dwell-based selection interaction when the system showed a cursor. My eyes (and apparently most others') tried to focus on the cursor instead of the button underneath and the small amount of inaccuracy and feedback delay of the eye tracker caused me to chase the cursor around the screen. Once the cursor was turned off, I could select with ease. This would have implications for an interaction technique that augments a tradition mouse or touchpad with eye movement for pointing, such as the EyePoint technique just described. I wonder how they deal with that.

In the afternoon I attended two sessions on Mobile Interaction Techniques. There was some interesting work, in particular: Shift: a technique for operating pen-based interfaces using touch and Intimate interfaces in action: assessing the usability and subtlety of EMG-based motionless gestures.

In the evening, I worked on a demo for a meeting the next day and then went for street tacos and excellent Vietnamese with my advisor, Pourang Irani.

Day 3 - Wednesday

Wednesday morning was the time my poster session (Comparing visualizations for tracking off-screen moving targets). I set the poster up early in the morning and it was available for viewing all day, along with the rest of the posters in the common area of the conference center. The break between the two morning sessions was the assigned time for me to stand next to the poster and describe my work to anyone interested.

The presentation went as well as could be expected with the large amount of work competing for attention. A number of people expressed interest and I was happy enough with that.

Pourang and I had a lunch meeting with Patrick Baudisch and Carl Gutwin to discuss a project for the summer to be submitted to next year's CHI. The meeting went well: it was great to meet Patrick in person and we came away with a concrete plan for completing the work.

In the afternoon, I attended some paper presentation sessions. Bubbling menus: a selective mechanism for accessing hierarchical drop-down menus was interesting and I made a note to show it to one of my lab mates who is working in that area.

Command line or pretty lines?: comparing textual and visual interfaces for intrusion detection was an interesting comparison of using a command line interface and using a visualization interface for investigating network intrusion events from log files.

This talk is part of a trend that I noticed in HCI, which questions the use of visualizations for large amounts of data. This trend became clear to me after the conference when I read Don Norman's recent column in Interactions magazine, The next UI breakthrough: command lines. The general idea, as I understand it, is that once a dataset becomes large, the only efficient way to use the data is with search (which is generally done via standard text entry on what could be called a command line). The network specialists investigating an intrusion from the talk above were using the command to issue complex grep commands to search through the log data for interesting entries, which was more efficient than the visualization interface tested.

No one has seriously considered using a visual representation for interaction with such things as the Internet (not recently anyway, I hope) because there is just too much data. The same may well apply to network logs, maps, your laptop's filesystem, source code, etc.

The idea of using the immense parallel processing power of our visual systems for searching and identifying patterns in data is still an important idea. But for those researching interfaces that take advantage of this, will increasing be asked to justify their use of visualization over an SQL query, or a grep command.

Pointing and beyond: an operationalization and preliminary evaluation of multi-scale searching at first appeared to be a rather ugly and boring comparison of pan-zoom, overview+detail, focus+context, etc. I eventually realized that the comparison itself was not be point, but the authors were instead presenting a generic framework complete with a software framework for the task. It is a great idea.

The evening was full of hospitality events put on by the conference sponsors. I only made it to the Microsoft event which featured book signing by Bill Buxton and others, along with an extremely drawn-out raffle where despite great anticipation, I won nothing.

This was followed by a large party sponsored by Google at The Tech museum. There was excellent food and drink (including Sierra Nevada Pale Ale, my favorite beer) and we were given full reign on the museum. Imagine 2000 geeks set loose in a technology museum - it was great fun.

I continued the evening with a 7-11 wine run and late night philosophical discussions with a group of students.

Day 4 - Thursday

Despite the late night, I was up bright and early (for me anyway) to finish a draft of my M.Sc. thesis proposal due that day. Unfortunately, I ended up missing my roommate Jared's talk to get the draft completed. Once again, I should have finished up all my work before I came to the conference.

I attended The I in CHI discussion panel, which I enjoyed. The discussion was a bit unfocused but it was great to listen to some of the leading people in the field talk about their ideas for the future of CHI. One gem that I scribbled down after a brief mention, was the alt.chi paper The three paradigms of HCI. As a new student, who is trying to figure out the scope of the HCI community and how it works, I found this paper helped.

A couple talks in the afternoon, Hard lessons: effort-inducing interfaces benefit spatial learning and Strategies for accelerating on-line learning of hotkeys explored the idea of annoying users to help them learn. They, of course, didn't use the term annoy but the idea is to make it artificially harder for the user so they are forced to think about the task more (and therefore retain the information better, in the first talk) or motivate them to learn a different interaction method (ie, hotkeys, the second talk). This seems like a very bad idea and although there was improvement in performance, that doesn't mean that users would tolerate the annoyance in a real product. I am being somewhat harsh regarding this research but I do recognize the contribution of the work. It is very interesting to consider that the generally accepted rule of making the interface as easy as possible to use may not always be the best. I am sure there is a place for this idea, just not in learning hotkeys.

One of the more rewarding areas of work in HCI and computers in general, is using technology to help third world communities. It is a noble cause, however, I am not quite convinced it is the best approach or use of funds. I attended one talk that could be classified in this area (it is also more generally applicable), Multiple mice for retention tasks in disadvantaged schools. The study looked at how the number of mice, and the strategy for using those mice, affected retention of an English lesson for students in a rural Indian school. Despite my reservations, it is promising to see work like this being done. I hope to see more.

The closing plenary talk by Niti Bhan was also on the topic of using technology to help rural communities in India. Unfortunately, the talk was simply awful. It started with a 20-30 minute (I swear) silent slide show of photos that were loosely connected with either India or technology. The audience was respectful for the first bit and then grew increasingly restless, eventually resorting to crowd stunts, like the wave. Some people left disgusted before she even got on stage. Once the talk actually started, she had great difficulty winning back the audience with a rambling and unfocused talk about her life and what she plans to do in India. I left with very little understanding of her plans and no insight into the problem or potential solutions. It was very disappointing and I am not sure who to blame, Bhan or whoever hired her.

After Bhan's talk, my faith in the conference was restored with a rousing advertisement of sorts for CHI 2008 in Florence, Italy.

Following the closing plenary, I went with a group of students for burritos and then Mojitos at a great Cuban lounge with amazing live music.

Some statistics from the conference:

• Attendees: 2622 (700 students) from 45 different countries.
• Proceedings Acceptance Rate: 182 of 840 submissions, 22%.
• Extended Abstracts Acceptance Rate: 212 of 582 submissions, 36%.

After the Conference - San Francisco

After the conference I took a train up to San Francisco and stayed for a week at the infinitely cool Green Tortoise Hostel at the edge of the China Town and North Beach neighbourhoods. I was a bit concerned that the place would be overrun with frat boys but the crowd turned out to a excellent mix of (relatively) young world travellers from the UK and Australia, peppered with the occasional North American who, for whatever reason, had time to hang out in San Francisco (like me).

I loved the city. It is beautiful and brimming with culture. The locals are laid back and surprisingly helpful. The city parties well into the night with great music and crazy people. It turned out to be a good idea for me to stay at a hostel, which besides being much cheaper than a hotel, provided a steady stream companions for exploring the city. It would have been quite boring otherwise and I would not have met some very interesting people.

The trip wasn't all fun, however. I was holed up in my bunk for two days after catching an exotic (I like to think so, anyway) virus that sapped all of my energy, made my head pound, clogged my sinuses, and ripped apart my throat. I'm sure (well, not really) that I caught it, avian flu style, from an Australian tourist who just came from touring poultry farms in Indonesia. After a couple days, I recovered enough to venture back out and continue my study of San Franciscan culture.

I didn't play the part of a tourist very well; I didn't go to any museums, the Golden Gate Bridge, Alcatraz, wine country or many of the other things on the typical tourist trail. I did, however, spend lots of time in the streets and I think I got to know the city as well as I could. I went to have a holiday and to see and do things I enjoy in a new and exciting environment, not to be a tourist. In that respect I have no regrets, except perhaps Alcatraz. Next time, I guess.

Final thoughts of the trip

Attending CHI 2007, along with my little side-trip to San Francisco, was an amazing experience. The thrill of travel and meeting so many enthusiastic and interesting people has reinvigorated me after eight months of gruelling coursework. After some recovery time, I am ready to take on the HCI world and work my butt off to get to Florence, Italy next year.

Italy, and the spring and first love all together should suffice to make the gloomiest person happy. - Bertrand Russell

---

On the trip, I received a number of suggestions for books to read. Both HCI-related and general fiction:

• Designing Interactions by Bill Moggridge.
• Sketching User Experiences by Bill Buxton.
• How Maps Work by Alan M. MacEachren.
• The Man in the High Castle by Philip K. Dick.
• Howl, and Other Poems by Allen Ginsberg.
• Big Sur by Jack Kerouac.

I want to thank my advisor, Pourang, for giving me this opportunity and for providing generous financial support. I also acknowledge the Faculty of Graduate Studies, Graduate Students Association, Department of Computer Science, and the Alumni Association for their financial support. And lastly, thanks to the conference organizers and all of the great people I met in San Jose and San Francisco.