Even in the graduate school setting there were two problems. One problem was the continued problem of shirking. The people who were comfortable with a lower grade could rely on those who were not to keep the group grade up. In one case, the presence of a peer-evaluation system was hijacked by a small cluster of shirkers who threatened to uniformly down vote the most active person if he or she did not do most of the work. This soured me on peer-evaluation systems.
The second problem is that groups naturally specialize. This is fine with group assignments that serve as a summative experience that primarily teach coordination. If there is specific course material or skills intended to be taught in the class, though, the group arrangement can allow students to specialize to avoid these lessons. In my grant writing class, for example, it was common for people to specialize in a way that only one person got actual experience with budget writing and another with schedule design -- with some specializing in the design of the oral presentation or proposal layout. I have had to change the design of the course to ensure that people would have to develop a budget, a schedule, etc. for themselves -- in addition to the larger group project.
Given this skepticism, I got much more out of the discussion of group-based instruction than I had expected. The key lesson for me was the importance of diverse group construction and the design of projects to allow people with a variety of skills to shine. This leads me back to some interesting mechanics in role-playing games.
Application to RPG design:
One of the mechanics that has interested me a great deal over the past couple of years is the skill-system in Gumshoe RPGs (an RPG that focuses on investigation stories). As a brief overview, investigative skills in Gumshoe RPGs do not require a skill roll to succeed at an action. Instead, if you are skilled at Chemistry - you simply get whatever clues are related to chemistry in a scene. The skill indicates that you can grab the spotlight in a scene that includes a link to chemistry. The character development process allows people to choose their skills. In doing so, the players choose when they are likely to be in the spotlight in the game. The players are advised to coordinate to ensure that the group includes characters with a diverse set of skills. The people designing the adventures, then, are advised to balance scenes to allow all of the characters to be in the spotlight.
This approach (though not solely because of this game system) has influenced my design of adventures a great deal. I spend almost as much time trying to balance spotlight as I do the raw design of stories. Different game systems have different mechanisms for balancing spotlight. Cypher System games have the skill system that serves a similar role as in Gumshoe (though there is still a chance for failure). Dungeon and Dragons requires a little more dramatic design choices because one mostly has to balance across social encounters, exploration encounters, and combat encounters -- a much more coarse basis for spotlight balance.
This all feeds into game design at a key point. The basic wisdom is that one should reward what one wants to see in play. If you provide all combat encounters, players will eventually invest only in combat skills (see also DnD 4e). If you balance across types of encounters, players will be comfortable with balanced characters OR with being really good at one type of encounter and being quiet or passive in others.
Application to Instructional Design:The game design wisdom (players will invest in what you reward) has important implications for instructional design as well. If the group assignments reward specialization - that is what you are likely to get. Furthermore, instructional assignments also have a bit of the spotlight logic to them. Motivating every member of a group will require having enough elements of the project that each member can have the spotlight at some point. This makes the design of good group assignments similar to the design of good RPG adventures.
However, this returns us to the issue of specialization. The spotlight strategy for Gumshoe and Cypher System games is based on the desire for specialization in the group. The design is supposed to incentivize players to specialize in something relevant to the story and then share the spotlight with others. This is not obviously the case in instructional design. If the goal of the assignment is to provide for general training across a range of skills, you would not want students to specialize and trade the spotlight across tasks. If the assignment requires 1-2 key class-specific skills and several more general skills (say, presentation design or report writing), the spotlight approach will allow some students to practice their general skills but never add the new, course-specific skills.
My current approach to this problem is to have a hybrid syllabus with a large group project combined with smaller individual assignments to reinforce the course-specific skills. In my understanding of educational lingo, I have individual formative assignments with a group summative assignment (grossly). Thinking about these assignments in terms of spotlight management has helped me better consider how to design group assignments requiring diverse contributions -- and to better situate these assignments within the educational goals of the course.