Quote:
Originally Posted by billw
The randomness as well as the ranking point system bother me as well, and I am not sure it delivers the expected results. But that is another discussion.
|
One possible improvement to the point system is to have rank points based not on the losing alliance score alone, but finding the differential between the winning & losing scores and adding this amount to the winning alliance's rank points, and subtracting the same number of points (ouch!) for the losing alliance. That way, a strong alliance is not penalized by having low-scoring opponents.
On the other hand, perhaps FIRST is trying to add some uncertainty to the strategy of the game -- an alliance that sees that its opponent is low-scoring might strategize to help its opponent score some points near the end, benefiting them both. I suspect this is why 3053 didn't try for the atlas ball in a 1 vs. 1 match we played with them.
Quote:
Originally Posted by billw
Would it be an improvement if teams were sub-grouped into lots of four and then had to play three matches between themselves? Because everyone would be on the field, it might be much faster to run. You might even be able to repeat the regrouping three times (each team would then play nine matches).
|
Yes, it's faster to run, but I think there are limitations to this situation. Suppose that 4 of the best teams are grouped together. Half of the outcomes must be losses, whereas all 4 teams might have had all or or mostly all wins had they played some weaker teams. Conversely, in a grouping of 4 weak teams, at least 1, possibly 2 teams could have all wins, and look like an artificially strong team. Ideally, you want to see each team play with and against a variety of strong, intermediate, and weak teams. This also allows a team to show off the range of its abilities for scouting purposes. However, your idea has spawned a model I'm developing where clusters play in groups of 8.
The 2 models that have invaded my brain this past week look like the following:
Increment Model - For each match, separate team numbers by different increments.
Advantages:
• No duplicate alliance partners
• Minimal duplication of opponents
• Odd number of teams is easily dealt with
Disadvantages:
• No provision made for minimum spacing
Node Model - Form clusters of 8. Have each team play 7 matches (or fewer) against the other teams in the cluster.
Advantages:
• No duplicate alliance partners
• For tournaments of > 15 teams, minimum spacing of at least 2 matches
Disadvantages:
• Duplicate opponents are likely
• For number of teams not divisible by 8, special provisions must be made (if the remainder is close to 8, leave gaps and fill in with volunteers. If the remainder is close to 1, mix in the extra teams, extending the cycle length by 1 match.
The details take 9 pages.
and are attached. If you can see improvements or ways to overcome the disadvantages, please send them along.