Retired from Competitions?

[David Franklin]

Originally posted by RobC
OK, hands up - according to your link, I may have misinterpretted how to apply the skating system. However, in all the ballroom competitions that I have scrutineered, I have never had to deal with a real life situation as twisted as your example, and since I have previously mentioned that I have not actually sat the scrutineers exam, it's a fair cop.

Having done some web-browsing, and having some knowledge of the skating situation, here's my understanding of the various rules:

Ballroom is ranked as RobC describes (see How Ballroom Competitions are Judged) - the error was in describing it as the Skating system. Under this system, the number of 1st places is the prime decider of placement (ties broken by # of 2nd places etc...)

Figure skating used to use the system described by Andy McGregor, generally called "majority ordinal". This system tries to make sure a majority of judges "agree" in some sense about the rankings. It is somewhat more complicated than the ballroom system.

Around about 1998, there was a change to OBO ("one-by-one") ranking - a (much) more complicated system which takes account of every "1 vs 1" comparison (i.e. the ranking is based upon looking at each pairing of competitors and which each of the judges preferred). OBO was sufficiently complicated that a lot of competitions didn't go over to it.

In the last year or so, the system has been changed again - this time to a "directly scoring" system (i.e. marks count, rather than just placements) - the whole 6.0 scoring system is now deprecated.

Dave

[Andy McGregor]

Originally posted by David Franklin
Having done some web-browsing, and having some knowledge of the skating situation, here's my understanding of the various rules:

Ballroom is ranked as RobC describes (see How Ballroom Competitions are Judged) - the error was in describing it as the Skating system. Under this system, the number of 1st places is the prime decider of placement (ties broken by # of 2nd places etc...)

I've had a look at the site and it seems to be the same as the one I call 'Relative Placement'. The only difference is that it goes on to consider more than one dance before deciding the overall winner - something we don't need to do in MJ.

Originally posted by David Franklin
In the last year or so, the system has been changed again - this time to a "directly scoring" system (i.e. marks count, rather than just placements) - the whole 6.0 scoring system is now deprecated.

Dave

The website David gave us the link to is titled "How ballroom dancing competitions are judged" it then goes on to describe the "Skating System" which seems to me to be the same as I'm calling "Relative Placement". This is clear evidence that some people in ballroom are using the 'Skating System. Does David or anyone else have a link to a site explaining the "directly scoring" system as, if it's fairer we should be encouraging MJ comp organisers to use it.

I have placed below my argument against direct scoring of MJ comps but I'd love to see the argument for a better system than the one I've written - because I can always change it!

In my whole method the judges do directly score couples. But only to find individual judge's rankings of couples. I think it would require a huge judges training programme to ensure that judges scored consistently if raw scores were used to calculate the final placings of competitors. To use raw scores in the absence of this cosistency would be to give some judges many times the influence than others. For instance, if a judge marked couples A and B with 9 and 2 and another 6 judges marked those same couples 7 and 8 respectively the final score would be A=51 B=50. So, even though 6 judges had marked B as the winner A had the highest score!

The above example is extreme but it must happen to some extent or other where raw scores are used. Judges who give bigger differences in scores will always have a bigger influence - but they will be able to defend their scoring as they truly did score the better couple higher - at least in their opinion.

Because MJ is not as formal as Ballroom in terms of exams for dancers and exams for judges there is no consistency of scoring. In this situation it is, in my opinion, COMPLETELY UNFAIR to calculate the winners of a final using raw scores!

[David Franklin]

Originally posted by Andy McGregor
The website David gave us the link to is titled "How ballroom dancing competitions are judged" it then goes on to describe the "Skating System" which seems to me to be the same as I'm calling "Relative Placement".

Um... I screwed up - that's not the site I was looking at when I decided ballroom judged things differently. But I can't find the site I was looking at...

Does David or anyone else have a link to a site explaining the "directly scoring" system as, if it's fairer we should be encouraging MJ comp organisers to use it.

I think it's on the ISU website but it's highly technical and I don't think it's relevant - basically because like you I don't believe in direct scoring for dance. The OBO solution is "supposed by some" to be better than majority ordinal, but is sufficiently compilcated to require computers IMHO.

I have placed below my argument against direct scoring of MJ comps but I'd love to see the argument for a better system than the one I've written - because I can always change it!

I think the more "objective" the sport, the more sense direct scoring makes. It also makes a difference if the number/difficulty of "scoring opportunities" might vary between competitors. The advantage is in allowing results to be combined. It also makes it much more explicit how you judge between three easy tricks performed flawlessly and four difficult tricks performed with mistakes.

In modern jive, I don't believe it makes much sense (because it's not all about tricks), with the possible exception of airsteps...

Another big difference is that in skating you combine the scores of two routines. The way the scoring worked under ordinals, you could be a clear 3rd after the short program, win the long program by an infinitesimal amount, and that would win the gold medal. It didn't make an awful lot of sense.

In my whole method the judges do directly score couples. But only to find individual judge's rankings of couples. I think it would require a huge judges training programme to ensure that judges scored consistently if raw scores were used to calculate the final placings of competitors. To use raw scores in the absence of this cosistency would be to give some judges many times the influence than others. For instance, if a judge marked couples A and B with 9 and 2 and another 6 judges marked those same couples 7 and 8 respectively the final score would be A=51 B=50. So, even though 6 judges had marked B as the winner A had the highest score!

This is where the whole "arrow theorem" thing of no solution being perfect comes in. Because even if that judge really and truely believes scores of 9 and 2 are appropriate, his score isn't going to matter. In general, this is a feature not a bug. But suppose couple B made a serious rules infringement that should be an automatic disqualification, but only one judge saw it. Under an ordinal scheme, it is very unlikely for that judge to be able to affect their position.

FWIW, most "direct" scoring systems will discard the 'outlier' results. Typically the lowest and highest scores are discarded, so a single rogue judge cannot easily affect the results. This is (essentially) the system used in gymnastics and diving. If one judge is significantly outside the other scores, the theory is that someone will hopefully report/investigate it. [With the exception of the ISU, where it seems the plan is generally to cover it up as best as possible ].

Anyhow, I'm certainly not advocating this for Modern Jive.

Dave