Conclusions
Further studies
We could further improve our study by looking at :
- How these correlations happen for each judge. Are there judges that are more correlated than others ?
- How do these correlations evolve during the tournament ? Between the first, second and third etc. rounds.
Why these ideas should be implemented at the 2019 French Physicists' Tournament
As we have shown above, the comparison of grades between tournaments serves as a discriminating experiment, just like we do in physics !
Let's recap the points we made:
Regarding the overall environment of the tournament:
- Most teams get away with very late preparation time leading to false results
- There are too many problems. Most teams cannot cope with such a burden.
- Due to the tournament style working of the IPT, there is a tendency for teams to present tweaked data in order to look great and surprising.
Regarding the grading procedure at the tournament:
- The Reviewer role is tremendously important to spread the grades of the Opponent (and the Reviewer)
- There is a positive correlation between all three roles. Let us analyze why such correlations appear :
- Reporter–Opponent : As the Opponent cannot present any additional result, this role is bounded by the presentation content of the Reporter. If the Reporter is good, the Opponent will have more points to discuss whereas if the Reporter's presentation lacks important details, the Opponent will not be able to present new ideas without being told that he presented his own results.
- Reporter–Reviewer : Same reasons as before, if the Reporter's presentation is good it is easier for the Reviewer to lead the discussion in a productive direction.
- Opponent–Reviewer : There comes a trickier problem, the Opponent and Reviewer role are the two most correlated roles. That comes from the fact that the two roles are quite hard to distinguish. Even though the Reviewer has a clear incentive to lead the discussion, the fact that the Opponent does not bring anything new leads the spectator to wonder what are their differences.
Realistic ideas to implement:
When we thought about ideas to better the tournament, we knew that it would be crucial to keep its spirit intact, we therefore believe to have come to a realistic list that can be implement in the next edition :
- Reduce the numbers of problems to 12-13, with 2-3 vetoes, 1-2 temporary and 1 permanent
- Implement a commission that would look at the feasibility of problems before they are revealed
- For the FPT, implement the role of the Reviewer
- Change the language with the rule associated with what the Opponent can present.