I spent the last week attending the Human-Computer Interaction Consortium, HCIC. HCIC is a membership-based organization that (whether by design or not was a topic of discussion) is not very publicly-visible. At any rate, I had a great time, hanging out with many interesting people, listening to fascinating talks, and participating in great discussion. Here are some of my notes on the talks that were presented (the theme of HCIC 2014 was “mobile”). HCIC has a different structure than many conferences; there are a few hour-long talks, followed by a 15-minute presentation by a “discussant”, whose job it is to further comment on the talk and encourage discussion by the audience. The discussion phase is extended as well, allowing the conference attendees to get deeper into thinking about a particular topic.
Tapan “Tap” Parikh, UC Berkeley School of Information
Tap’s research group “studies the design and use of information and communication technologies for sustainable development”. His talk focused on what he calls “representation technologies”—technology used to represent and communicate knowledge—and how to increase diversity and widen access to such technology. He presented two projects: Awaaz.De and LocalGround.
Awaaz.De is something like a phone-based message board service. The initial project, targeted at agriculture, allows farmers in India to call a number and ask or answer questions. For example, “I want to grow cotton. Which weather environment is best?” Interestingly, their research results may indicate that farmers tend to trust their nearby peers more than experts. Here’s a PDF of one of the latest papers on the project.
The second project Tap talked about was LocalGround, a system for helping to document geographically-linked local knowledge. He gave an example from a low-income neighborhood in the East Bay of San Francisco, where the tool was used to collect community impressions about locations around a school, as well as to try to influence the planning process for a new city park. The talk ended on a bit of a down note (as Tap laughingly acknowledged), as data collided with politics and lost. The latest paper on LocalGround can be found here.
Mobile Community Apps as an Innovation Infrastructure
John and Jess talked about several mobile apps they’ve developed to help build community, particularly in State College, the town where Penn State is located. Although many non-profit agencies often exist in communities, John and Jess pointed out that they are frequently quite siloed, not interacting or sharing much with each other. Their goal is to enhance “community innovation” by linking agencies and community members to each other. They showed three pieces of research: Lost State College, an historic walking tour of State College that adds the ability for community members to share their own perspectives on history; Future State College, a similar application that allows users to browse in-place the 10-year master plan for the town and comment; and Local News Chatter, which collects locally-relevant news with local tweets.
Beyond the Smartphone: Rich Mobile Shared Experiences
John Tang, Gina Venolia, Sasa Junuzovic, & Kori Inkpen, Microsoft Research
John Tang (properly pronounced, I learned “Tong”) works for Microsoft Research in Redmond, WA, but lives here in the SF bay area. He uses a telepresence robot to interact with his colleagues at MSR, and we got a fun demo of him driving the robot through the halls on the Microsoft campus. His main thread of discussion, however, was on using telepresence-like technology to share experiences with others. He showed several projects—Experiences2Go, PortaProxy, and ProxyWear—that allow remote people to view and, to a limited extent, interact with remote events. One example (of Experiences2Go) was of grandparents remotely attending their grandchild’s birthday party.
Capture and Playback for Designing Context-Aware Interactive Systems
Mark Newman, Mark Ackerman, Stanley Chang, Manchul Han, Perry Hung & Jungwoo Kim, University of Michigan
Mark “Epic Beard” Newman presented some work very much in the vein of part of my own dissertation work. The idea is that for developers of context-aware systems, testing and iterating on the system can be very difficult and time-consuming. The RePlay system that Mark and colleagues built allows developers to record sensor information and then use it for repeated testing of a context-sensitive system. They studied the system by having developers create an application for location-based content delivery. One issue that came up—which I experienced in my own work—is that of running studies on complex software like they wrote. Another concern I had (and likewise had no answers for) was that their system (like my own) was highly optimized for the kind of data they were using, making it difficult to generalize to other problems.
The discussant for this paper, Gregory Abowd, had some interesting points. He opined that Mark was working on the right problem, and that tools for programmers of Ubicomp systems are still in their infancy. He pointed out that we design and implement 2D GUI interfaces using 2D GUIs, but that hasn’t translated to the new paradigm of multiple devices situated in various places in the world.
Usable Privacy and Security for Mobile Devices: With Great Power Comes Great Responsibility
Serge Egelman, UC Berkeley
Serge talked about the difficult problem of allowing users to understand and control the kind of access that applications (especially mobile applications) have to a system. He posited as an example trying to discover which application on a phone might be responsible for excessive SMS charges, and showed that to even find out which applications have permissions to send SMS takes over a dozen steps on Android! He compared the Android approach of “ask once at install” to the iOS approach of “ask once when necessary” and pointed out the flaws—a major one being that apps want to use so many things in the system that they ask all the time and users become habituated and tend to ignore the security and privacy implications of the questions. One possible solution he put forth is attribution—allowing users to see what application has recently done what. He gave two examples of how this might work: an ongoing notification icon in Android showing which app is currently playing sound, and a note on the wallpaper setting for the phone showing which app most recently changed it. Here is a recent paper discussing some of the issues.
Mobile Support for Face-to-Face Social Interaction
Jaime Teevan, Merrie Morris, & Scott Saponas
Jaime’s short talk was about how phones might add value to social interaction rather than simply being a distraction. Her talk was augmented by a system whereby people could give thumbs up or down during the talk, as well as answer questions. There was some interesting discussion—mainly engendered, I think, by her initial photograph of her family at dinner engrossed by mobile phones—about whether such technologies actually increase social interaction or simply provide another layer of distraction. A general consensus was reached that such applications could be quite useful in targeted settings such as the classroom, but it was unclear whether a similar benefit could be realized in less-focused situations such as a family dinner.
Other notes and thoughts
I had an interesting discussion with Steve Johnson, a Ph.D. student at UW Madison, about telepresence robots and how they might communicate “presence” when someone is using them, even if the user is not doing anything active with the robot. I was thinking of how I could avoid my commute to San Francisco by working via robot, except that mostly I program all day. How could the robot, sitting still at my desk, give the impression that I’m there and available to talk? We discussed subtle motions and even (very subtle!) smell-based cues.
Don Norman had some things to say (which were met with varying degrees of appreciation) about the difficulty of research work influencing products. This is one of my own fundamental conflicts, although now that I am in a more product-focused organization, I’ve discovered that it’s nearly as difficult to influence product anyway.