I had an interesting discussion with Amy Bruckman the other day (http://www.cc.gatech.edu/~asb). I was asking her about a project to attempt to create "virtual communities" in classrooms, and she raised an issue that I hadn't even considered: Maybe it's not possible. The notion of "community" implies willingness or even desire to be part of the community, usually to some advantage of the participant. The particular advantage that Amy plays off of is fun: Making the community someplace fun to be. Amy's concern is that the culture of the classroom tends toward only one goal for participants: Grades. Grades are not fun.

At first, I was uncomfortable with Amy's analysis. I've been part of class-related virtual spaces where I, at least, had a real sense of community. But then I came to the realization that those are the rare instances in undergraduate education where students are really interested in the topic, not just taking it for a good grade in a required course, or it was an instance in graduate education, where students are much more likely to be interested in the discussion as fun.

Now I've come to suspect that "community" is perhaps an unrealistic goal, at least for traditional undergraduate classrooms, not without a large change of culture. I think it's time to think about social interaction spaces other than "community." The village square of a MOO is one kind of virtual place that parallels a physical space where we interact with one another. While it has some tremendous advantages (e.g., the support of other participants, the mixture of learning and socializing), it has some disadvantages, too. Communities involve committment. It takes time to become and remain a full, contributing member to a community.

Other social interaction spaces may make different trade-offs. I've heard people refer to IRC (Internet Relay Chat) as a virtual "cocktail party." Much of the work in the CSCW community seeks to create spaces for virtual "meetings." Cocktail parties and meetings are two kinds of social spaces where people interact that are not in themselves communities.

• Cocktail parties and meetings don't require as much committment as a community. One can attend a cocktail party or meeting only once and never re-meet the same people.
• They take place over a much shorter period of time. I think this plays a role in terms of the cost of a "willing suspension of disblief." One of the lessons I've learned about MOO's from Amy is that they are real -- the participants involved have suspended their disbelief and accept that the textual penguin is a penguin. The disbelief that I have to suspend is that the social interaction that I want students to engage in is about learning, and not about grades. I may be able to get that suspension for a short time, while I may not for a longer time.
• The downside of each of these is obvious. The support of a large number of participants that is there for a community is not there for a cocktail party. There probably won't be the same engagement for a brief interaction activity that there is for an ongoing community membership.

It's worth exploring these alternatives to community for educational social interaction. I don't necessarily think that cocktail parties and meetings are our best models in the physical world, though. Here are a few possibilities of what might be:

• Lecture: I don't mean lecture here as "stand forth and deliver," but the best examples of what can happen in a lecture period. I try to make my lectures interactive and engaging, while still providing useful information in a prepared format. I ask students questions, I respond to their questions, I use small group activities, we review and critique student and group work, and sometimes we even do group hack sessions.
• Museums: Have you ever been to a party that takes over a museum, or go to a museum with a big group of friends? I find that to be a great learning and social experience. You float from exhibit to exhibit and join in the cluster of friends around that exhibit. The people bring you up to speed on the exhibit and the discussion, and you discover that you have a local expert who can really add to the exhibit. Each of you gets to be teacher and student. You learn more about the exhibit than what's there, and more about your friends, too.
• Code Review: The software engineering term sounds too formal, but the interaction I'm thinking about here are two or more people staring at the same code (on paper or screen) and talking about it -- either to literally review it, or to look for bugs, or to figure out what it does. I've done this a few times and find it to be really fun and useful. You have to work at really sharing the visual space, but when it does work, each person's talents can really shed new light. My wife is amazing at figuring out detailed algorithms that I gloss over, and my friend Luke is brilliant at picking out the strengths and weaknesses of a data structure, and Brian really knows how to relate math to a program to come up with new tricks or insights.

Each of the above, in my experience, strongly relies on the physical world -- the lecture room where we can hear and see each other, the museum exhibits that we can crowd around, and the shared real-time code view combined with high bandwidth face-to-face communications. Capturing the benefit of these in a virtual space is a challenge. Even more challenging and exciting is extending these to make them work even better as a learning and social activity than they do in the physical world. Perhaps these kinds of briefer but still beneficial events may be an alternative path to virtual learning spaces.

Side note: What kind of activity is it to design virtual spaces? Computer science? Architecture? Pyschology? Education? All of the above?

In the most recent Interactions, Dan Olsen, director of CMU's new HCI Institute, explains why he believes that HCI is a new science that grows out of computer science. He sees HCI design as requiring designers, inventors, evaluators, and modelers that are all of the above, and not just a team of single-discipline experts. Terry Winograd makes a similar argument in the book celebrating the 50th anniversary of the ACM Beyond Computation. Winograd talks about "interaction design science." Maybe Olsen and Winograd are right. Designing web pages and other shared virtual artifacts is not just about task and widgets anymore -- it's about interaction at many different levels for many different purposes.