This is similar to a within-subjects design except that a group of people need to work together in each condition, for example on some form of shared software or CSCW system. If anything, group dynamics make the differences between groups larger than individual differences, so there is even greater advantage in terms of statistical power. Unfortunately the interference effects between conditions are also greater and it is harder to match groups in the way that you can individually match subjects.
Used on pages 110, 111