There were three entrants in the category of planners requiring hand-coded control knowledge. As in the 2000 competition the teams competing with these planners were allowed time to reformulate domain descriptions to include domain-specific control knowledge. As the results show, control knowledge can dramatically improve planner performance. However, it is difficult to assess the cost-effectiveness of hand-coding control knowledge. Fahiem Bacchus fahiem observed the need to quantify the time and effort required to identify and encode useful control knowledge in order to be better able to judge the trade-off between the fully-automated and hand-coded approaches. However, it is very difficult to measure the effort involved. In principle it should be possible to bound the time allowed for domain reformulation, but then differences in team sizes and in the experience of team members become very important in comparing what has been achieved by different participants. The third competition, like those before it, did not explore these factors, so it is impossible to make judgements about the relative effectiveness of the solutions to these problems offered by each of the hand-coded planning systems. This remains a very important open issue for future competitions to address.