Part of the pain involves the setting up of a system or process that either monitors individual effort within the group or allows students, through a peer review process, to influence and moderate the group mark so that individual contributions are recognised.
But let's now consider the students, themselves, the natural competition between humans, especially young ones and the desire to "win", the strategic approach to learning and anything to do with it, and the opportunity that group work, however carefully it is designed, provides for the achievement of individual objectives. Let us also consider the natural sense of justice and fairness that is embedded in most human nature and the potential for both punishment and forgiveness for group members. In short, Games that students can play.
Consider the following:
Free riders - perhaps the most obvious beneficiaries of group work. These students rely on the efforts of group members and their unwillingness to punish too harshly.
Dominators - those able to select weaker group members, dominate the group effort but exact unequal benefit by negotiating favourable peer review scores in exchange for better than expected group marks.
Collaborators - these students join "friendship" groups and agree that all members of the group will score each other equally in the peer review - whether this is justified by effort or not.
Traitors - those who agree scores with their group as collaborators but then cheat their friends in order to gain higher marks individually.
To avoid these issues, which happen rarely in practice, group work is often reduced in scope and impact in later stages of degree programmes. No one single mechanism for peer review will iron out all problems but perhaps that is just the point. To avoid the "learning effect", whereby the above student types "game" the peer review system, tutors need to deploy a variety of methods over time.
Tutors need to "game" the " gamers" to achieve the full benefits that collaborative learning can bring.