Skaters Welcome New Marking System (Lalique Spoilers)

[adrianchew]

http://www.dailytimes.com.pk/default.asp?page=story_18-11-2002_pg2_14

[Debbie S]

Former US champion Michael Weiss admitted Saturday that his victory in the Trophee Lalique may not have been possible without the new system of voting in figure skating...Under the previous system if a skater was fifth going into the final it would be virtually impossible to win.


Huh? Am I missing something? How does the fact that 14 judges mark and only 9 unnamed judges' scores count make it easier to pull up from 5th to 1st? The ordinal/OBO/factored placements system is still being used. Secretive scoring still doesn't change the fact that any skater who's 5th in the SP needs the top 2 SP skaters to have meltdowns in the LP in order for him/her to have a chance to win.

[Artemis]

Sorry, can't help but laugh that the skater endorsing the new system is one who feels he's benefitted from it. A valuable opinion indeed! (((sarcasm off)))

[pilgrimsoul@work]

Debbie, I share your confusion on Mike's assumption that the new judging system helped him at Lalique. These may be stupid questions, but I'm gonna ask anyway - help me out, my more math-gifted fellow posters. Would it be possible for Michael to move from 5th to 1st under both the old and new judging systems? If the answer is yes, could it occur under the same circumstances? Is a leap of this magnitude more possible under the new system, and if so, why? I really want to understand how far a skater can move up after a bad short, and conversely, how far down a skater can drop overall with a good short and bad long. It appears to me, from Mike's Lalique victory, that the placement spectrum has widened considerably, and I am really curious as to what scenarios are now possible.

[adrianchew]

I have to say from the events so far - I do think the new system has been effective for improving the judging in ice dancing.

[Ellyn]

Originally posted by pilgrimsoul@work These may be stupid questions, but I'm gonna ask anyway - help me out, my more math-gifted fellow posters. Would it be possible for Michael to move from 5th to 1st under both the old and new judging systems? If the answer is yes, could it occur under the same circumstances?

Yes and yes. Do you want examples of what happend with the factored placements at Lalique and at other events where similar things happened in the past without random judge selection?

Is a leap of this magnitude more possible under the new system, and if so, why?

I can think of only 2 more reasons why it might be more likely to happen with the current system.

1) If different judges are randomly chosen to "count" in the short and long programs, so that in essence there's a different panel of judges for each program, it's entirely possible that more of the judges whose scores counted for the short program were those who happened to be "biased" against a certain skater's skating (whether for reasons of national bias, stylistic preference, or pet peeves/preferences on technical issues, i.e., dishonestly or completely honestly) and the judges chosen for the long program include more who are similarly biased in favor of that skater's skating.
I'm not sure if that is the case or not that different judges are chosen for different programs, but if they are, that will sometimes magnify the effect of certain skaters skating much better or worse in one program than the other and produce big jumps up or down for that skater; other times it will mitigate the effect of large differences in the skating and produce smaller jumps than might otherwise be expected. The general trend would probably be toward more rather than less movement compared to using the exact same judges for both phases.

2) With the current state of anonymity and lack of immediate calls for explanations at the post-event judges' meetings, judges may feel less pressure to try to conform to what they think the overall result is going to be and just mark what they see on each day. At least that seems to be the reasoning in favor of the anonymous/random selection system by its proponents.

I really want to understand how far a skater can move up after a bad short, and conversely, how far down a skater can drop overall with a good short and bad long.

This is an effect of the factored placements and is dependent purely on how many phases there are to the competition (e.g., if there's a qual round that counts or a compulsory dance or two, OD, and free dance, then the free program counts for less than when there's only short and long program, but how many skaters/teams "control their own destiny" going into the free could vary from 1 to 5 depending on the mix between the first phases instead of always being 3 under the 1/3 short-2/3 long format) and what order all the other skaters finish in to "help" a skater who wins the free but does not "control his destiny."

It appears to me, from Mike's Lalique victory, that the placement spectrum has widened considerably, and I am really curious as to what scenarios are now possible.

There's no change in scenario. And one example is far too small to draw any conclusion about whether such jumps up from 4th, 5th, of 6th to 1st place will be more or less common under the new system. It always could happen given just the right circumstances, and occasionally did. This just happened to be one of those occasions that happened to involve random judge selection.

[dcden]

I agree with Debbie and New Yorker, that the new system is different only in the way the scores are presented to the audience (i.e. we don't know which of the posted scores are actually used). But the OBO system used last season is identical to the one used this season, as far as I know.

The only way I can see someone saying that the current scoring system benefits them is to consider OBO's method of handling ties. I can't recall any other time when there were TWO ties in one stage of one discipline, much less two ties in an entire competition. The ordinal method still used in the US would probably have resolved the ties, but without knowing which scores were given by which judges, there's no way to know for sure. In any case, the fact that OBO uses fewer tie breaking procedures than the old ordinal method ("judges in favor" vs. # of judges in majority, TOM, and TO) means that you're more likely to get such ties, in which case a skater like Weiss would be ranked (tied for) 5th in the short rather than possibly being ranked 6th. Had Weiss been ranked 6th in the short, he would not have won the competition.

[dcden]

Wow, good points Ellyn, I didn't think of those. I agree that the effects of (1) (if the judging panels do indeed differ from SP to LP) would contribute to more jumping around in the rankings from one stage of the competition to the next. I'm a little too cynical to think that (2) actually happens, but it's possible that some judges out there actually have a good conscience ;)

[Debbie S]

Originally posted by dcden
I can't recall any other time when there were TWO ties in one stage of one discipline, much less two ties in an entire competition. The ordinal method still used in the US would probably have resolved the ties, but without knowing which scores were given by which judges, there's no way to know for sure.

Actually, dcden, your post reminded me of why I was taken aback at the ties in the SP. I thought one of the main arguments for OBO was that it prevented ties (along with flip-flops). The ordinal system used at U.S. Nats actually makes ties more likely b/c it just looks at each skater's placements from judges and not necessarily their placement relative to other skaters. For example, last year in the men's SP, Tim G. and Todd E. tied b/c each had four firsts and seconds. But the "culprit" of the tie was 1 judge who placed Matt S. 1st, Todd 2nd, and Tim 3rd. That ranking didn't resolve the tie b/c Todd also had 1 3rd (I assume that judge placed Tim 1st and Matt 2nd), so the numbers of each placement were tied bet. the two.

If OBO was used, Todd would have won the SP outright b/c that judge ranked him ahead of Tim - to put it more simply, 5 judges placed Todd ahead, so Todd would have ranked ahead of Tim overall, even though he did not get any more firsts than Tim did.

[dcden]

Originally posted by Debbie S
I thought one of the main arguments for OBO was that it prevented ties (along with flip-flops). The ordinal system used at U.S. Nats actually makes ties more likely b/c it just looks at each skater's placements from judges and not necessarily their placement relative to other skaters.

True, but while OBO looks at skaters' placements relative to one another, it doesn't make much of a distinction as to whether a particular judge placed skater A ahead of skater B by one ordinal or by several (absolute placements). The only time this comes into play in OBO is when using "judges in favor" which is the SOLE tiebreaker in OBO. In the ordinal method, the absolute placements are used for TOM and TO, the 2nd and 3rd tie breakers (the 1st tie breaker, # of judges in majority, just counts the # of judges who gave the majority placement or higher, but doesn't care how much higher). My hunch is that neither method is significantly better than the other in preventing ties, but since OBO was hyped as a new method to reduce ties, I expected OBO to be a stronger scoring system than I feel it is.

The Tim vs. Todd example does indeed get resolved under OBO, but that's just one example. Because of the new method of displaying scores in ascending order, it is impossible to say whether the ordinal method would have resolved the 2002 Lalique ties (maybe we'll find out when the scores are released in 2016 or whenever they decide to make these scores public).

There is another interesting example from last year's Lalique ladies' LP (http://www.usfsa.org/events_results/results/200102/lalique/ladies-free.htm). Here, under OBO, Sarah beat Sasha beat Maria beat Sarah, and all three beat everyone else. So that meant it came down to judges in favor (equivalently, total ordinals), which decided in favor of Sarah, with Maria in 2nd. But what if the Italian judge had placed Sarah 3rd and Sasha 2nd? Sarah would still beat Sasha 5-4, but then all three would have the same # of judges in favor... resulting in a THREE-way tie for 1st in the free program! The ordinal method would have resolved the ties under both the real and hypothetical scenarios.

One could easily come up with examples where one method resolves a tie which the other method does not, so I think the only way a helpful conclusion could be reached is if someone did an objective historical study on competitions in years past, to see how OBO performs vs. the ordinal method in preventing ties. Sandra Loosemore's site (http://www.frogsonice.com/skateweb/obo/) is the best example I've seen of such a study, but I admit it's biased against OBO.