How to Have a Great HICSS Experience

Guidance for someone used to ACM conferences

HICSS is more interdisciplinary and topically diverse than ACM conferences. With an acceptance rate of ~45%, HICSS includes both polished papers and some aimed at discussion and feedback en route to journal publication. With over a dozen parallel sessions, you can’t scan the program just before a session begins to decide what to attend. However, an hour’s preparation can yield an experience that is as consistently interesting and high quality as at ACM conferences where papers are more tightly screened by program committees. If your interests are eclectic, a HICSS experience can be better, due to the topic diversity.

Tools
The Program booklet has a high-level calendar of events, papers organized by sessions with Best Paper Nominees starred, and a list of all authors who registered early (one author for each accepted paper must register early). You also get a book of abstracts and a CD with the papers.

Method
First I mark each registered author whom I recognize, independent of their area. This takes advantage of a “fish-eye lens” effect – someone I recognize from an unrelated field is likely to be good. On the high-level calendar I mark the sessions each of their papers are in, then examine the other papers in those sessions. If a session looks interesting as a whole, I indicate that on the calendar; otherwise I just note whether the paper will be presented first, second, or third. When done with all the authors, I do the same thing for Best Paper Nominations. These papers are likely to be polished. Because of the diversity of HICSS, most of them are not of interest, but a good many individual papers and sessions look intriguing. Finally on the calendar I mark the minitrack or minitracks that have high priority. Last time that was two of the Digital Document minitracks and Kevin Crowston’s Open Source minitrack.

For almost every time period, I ended up with one to three possibilities of interest. Sometimes I attended one session for the first paper and another session for the next. Last year one session had nothing, so I talked with some people, then spent time with my accompanying family.

Conclusion
With lower selectivity, HICSS takes off the “parental controls” enforced by ACM program committees. You can’t wander into a session and expect the polished if sometimes boring quality found at CHI. But you’re given the tools to design a great conference experience. And I haven’t even mentioned the fact that with lunches and evening social events included every day, there is a huge opportunity for conversations.

HICSS 39 Trip Report

Joi Ito gave an engaging plenary talk on emerging technologies and practices around intellectual property, focusing on film and music and how the industries are not responding well to the digital change.

Dan Russell gave a talk focused mainly on studies at IBM Almaden, followed by interesting data from studies since he joined Google. (Dan has also worked at PARC and Apple and mentioned a recent talk at PARC describing different research lab cultures.) At IBM he studied how people quickly skim large numbers of documents under time pressure using paper and online tools. After two IBM tools fared worse than paper, they did careful video analyses to understand why and built a new prototype that did much better (for what seem plausible reasons). The search experiments at Google found patterns in eye-movement data and user behavior, for example in response to small differences in duration of returning a page even at short intervals and moderate variability in such durations. (Incidentally, IBM pension plan changes announced last week will affect longer-term employees after two years; some might be easily recruited then if not now.)

Other speakers of interest included Charles House, an Intel Research Director of 10 years following 30 doing the same at HP, a very engaging speaker with insights into RTC whose team (including Cynthia Pickering who spoke at MS last year) is tackling asynchronous interaction with my favorite innovation of 2005, the “asynchronous meeting”; danah boyd, who attended Social Computing Symposia here, now a Yahoo employee and SIMS grad student, who spoke on analyses of Friendster data and sent me nice as-yet unpublished papers on blogging and tagging; Annie Tat, a Calgary grad student with a very nice 3D visualization tool for MSN Messenger repositories; and, for anyone working on disaster modeling and response planning, John Linebarger, an impressively knowledgeable fellow from Sandia who after interesting work is being moved to work on semantic nets, gave a clear and thoughtful talk.

Alan Dennis described qualitative studies of IM use by execs and high-level managers in a distributed high tech and pharmaceutical companies (think, for example, Cisco). Behaviors are consistent with those in some MS managerial meetings, but more widespread and formal in less collocated settings. For example, job interviews and contract negotiations are held exclusively by phone in many highly distributed groups. From the perspective of the candidate or contractor, there is one interviewer, but several people are listening and IMing questions to the person talking. Partly driven by necessity, it seems efficient. Another paper by Dennis carefully analyzed when people approach other people for information or knowledge, looking at many possible contextual correlates. For example, is the information present in a formal company document repository? To Dennis’s surprise that made no difference (my paper focused on the general inadequacies of such repositories, and I cited this as supporting evidence when giving my talk).

One fascinating session comprised three skillful talks on “the dark side of knowledge management.” None were very technological but all had clear implications for technology design and use. Two included frameworks and numerous examples and will require reading the papers to fully assess. The other looked at legal and behavioral issues around organizational ownership of knowledge in employee heads – some surprising legal case history and evolving practices.

Microsoft Research participation included Danyel Fisher’s visualization paper that won a Best Paper award, Tammara Coombs Turner’s paper on technical newsgroup communities, Shelly Farnham’s paper on Wallop, and my paper on emerging technologies and knowledge management. There was heavy participation from IBM Watson HCI people, UW Information School (eight faculty and some students), and others.

Sessions I attended that were perhaps most interesting for revealing the intensity of research in the areas were on Open Source and cross-cultural differences in team dynamics.

Odds-and-ends: An e-government session covering identity card practice and plans in Europe, and a proposal for a semantic web e-government portal for Europe that drew a sharp response; an MIS paper on the “new frontier” of affective aspects of design driven by online consumerism, again recapitulating the course that CHI took for commercial software.

Overview of the HICSS Conferences

The Hawaii International Conference on System Sciences is organized along 9 tracks:

Collaboration Systems and Technology
Decision Technologies for Management
Digital Media: Content and Communication
Electronic Government
Information Technology in Health Care
Internet and Digital Economy
Knowledge Management Systems
Organizational Systems and Technology
Software Technology

Each track is divided into multi-session mini-tracks, averaging 8 mini-tracks per track. Each mini-track has one Best Paper Nomination, leading to about 75 such papers (about 10% of submissions or 20% of acceptances), and each track has a Best Paper, leading to 9 Best Papers (1% of submissions).

In January, 2006 HICSS also had six multi-session symposia:

Case and Field Studies of Collaboration Technologies
Electric Power Systems: Reliability, Control, and Markets
Hot Topics in Negotiation Support Systems
MOCHA Design: Mobile Computing Hardware Architectures Design & Implementation
Security Informatics
Skilled Human-Intelligent Agent Performance: Measurement, Application, & Symbiosis

HICSS gets about 1000 submissions and usually takes a little over half. Last year it accepted 48%, 450 papers. There were typically 14 or 15 parallel sessions. Days began at 8 a.m. and ended at 5:30 or 6:00. A premium is placed on discussion, and most people attend most sessions. The mini-track format tends to hold people together for half a day to a day or longer. HICSS also effectively promotes informal discussion by covering lunches and a 6pm-7pm social hour with the registration fee.

The papers vary from polished to somewhere between a workshop and a conference standard. Most authors see publication at HICSS as a step toward journal publication. Many participants are from IS/IT (for many of whom this is the premiere conference), which prizes journal publication more than CS does.

Topics can extend across multiple tracks; I noticed this for open source, disaster response, and semantic nets. Also, papers relevant to one track or mini-track can be in another. With so many parallel sessions and this kind of overlap, it’s essential to plan sessions and papers in advance to get the most from the conference. I'll post another blog entry on how I did this in 2006 to have one of my best conference experiences.