Ptichka said:
2. If they insist on throwing out random scores, put in a rule that if a judge from a skater's country is on the panel, then that judge's scores automatically get thrown out. (The "other side" will always consider them unfair)
Except they're not randomly choosing which scores to throw out skater by skater, but rather for the whole competition segment.
I once had a discussion with someone about doing this with the ordinal system -- in that situation it would make no sense at all, because one judge might be marking high throughout and another low, so the ordinals only make sense if one judge's marks are used for the whole event. (E.g., 5.7/5.7 might be first-place marks from one judge and fifth place from another.)
With adding up total scores, it does make a bit more sense. Although all it would achieve would be that, if judges consistently score their own skaters higher than the rest of the panel, their high scores will be thrown out (or kept, but still be high, if another judge happens to be using a higher range throughout).
If they were really trying to cheat against "the other side," they would also be consistently undermarking their skaters' perceived closest rivals, and your suggestion wouldn't protect against those artificially low scores. Nor is there any way to figure out in advance which skaters a cheating judge would consider a threat who needs to be undermarked, because it partly depends on pre-event buzz and partly on how the skaters actually perform during the competition.
So throwing out the home country judge scores would really be more a cosmetic solution than an actual one.
3. I like the idea of having separate jusges do PCS. If not, break the components down even further. The more it's broken down, the more objective it will be. This (the streamlining of PCS) is what I consider CoP's main problem.
My initial reaction when we were first introduced to the system last season was that one possibility to protect against individual judges deliberately or accidentally controlling the results by using larger ranges between their scores (for the medal contenders or whichever skaters we're worried about) would be to add up the GOEs and component scores by column, attach the same base mark as determined by the callers to each column, and convert each judge's total scores into ordinals and crunch those numbers by OBO or majority. I.e., the most complicated aspects of both system, but gaining the benefits of both. *Really* tedious to do without computers, though.
My other thought was, if they're just going to add up the scores anyway, and identifying levels of elements plus assigning grades of execution plus giving component scores is too much for any single individual to manage in real time, which is why they needed to separate the caller and judge functions, perhaps they need to break it up differently. Because grading the execution of all the elements *and* rating the components does appear to be too much to do at the same time.
So maybe, for singles, you would have:
-one set of officials identify and grade the jumps and spins (these are relatively easy for anyone who's motivated to learn to identify even if they never skated themselves, so it would be a good first assignment for judges from smaller, newer skating countries who have trouble fielding experienced judges)
-one set to identify the difficulty of the step (and spiral) sequences and also to rate the skating skills and transitions (this definitely takes more skating knowledge and experience)
-one set to rate the performance, choreography, and interpretation (these officials should have a significant degree of training in the arts, probably from other sources in addition to ISU seminars)
Obviously for pairs and dance the breakdowns would need to be different.
If you combine the identification (calling) and grading functions in the same individuals, then maybe instead of some aspects being rewarded or penalized in the level of the element and others in the GOE, there would just be a certain number of bonus points and deductions available for each element that the judges could award for both added difficulty, outright errors, and positive or negative quality.
But now that the existing system is in place, I doubt they'll actually reconsider the basics that drastically. Although putting in more checks and balances to the calling function and letting a separate panel assign component scores would be less drastic, so maybe we can hope for it.
