A new article from Sonia

[Joesitz (RIP)]

Yes, but most people's issue isn't with the calculations - it was with the awarding and not awarding of points (i.e Evan was given more credit than he should have and Johnny was not given enough credit).

That's not a mathematical solution, but a personal preference solution. Pesonal preference solutions vary in different eras and even that vary with different minds. Was the Birth of a Nation the greatest film ever or is one fixated on the one scene of miscegination(sp)?

Joe

Much of what I read from Sonia's discourses are they contain a vast amount of opinions mixing in with some relevant facts. Hard to agree or disagree with her, but always interesting because she isn't a fan of a skater, but a fan of skating. I like that.

Joe

 

[ChrisH]

I don't think rounding earlier makes the calulation easier.

{10, 13, 16}

(a) Divide these numbers by 7, then add.

10/7 = 1.43
13/7 = 1.86
16/7 = 2.29

Total = 5.58

True, yet your rounding before adding made the above calculation simpler.

(b) Add first, then divide at the end

!0 + 13 + 16 = 39.

39/7 = 5.57

The last method avoids the rounding error of the first, AND it's easier because you only have to divide once instead of three times.

True, yet this 2nd calculation was less informative and redundant. The protocols have the scores for each skating element and each program component. It's nice to see the points for each part rather than just the sum total. It's easier to break down the calculations allowing multiple people (if the calculations must be done by hand) to take part in the calculations and to spot errors.


---

I just read the propositions from Skate Canada from Dr. Rossano's site, concerning points for jump combos. That is so confusing, how are people supposed to understand that?

With the current rules, there are already large indirect bonuses for doing combinations, so I disagree with these proposals which seek to increase their importance. The biggest problem with combinations is that judges are limited to grades between -3 and +3 like with solo jumps. So if a skater does a jump combination in the short program consisting of a triple lip landed with two-hands down plus a singled toe-loop (which requires an automatic -3 grade), the lip and two-hands down are ignored by the system.

 

[Jaana]

Yes, but most people's issue isn't with the calculations - it was with the awarding and not awarding of points (i.e Evan was given more credit than he should have and Johnny was not given enough credit).

Unfortunately Weir´s problem is that he does not skate the whole content of his freeskate programme and his programme (even if skated fully) is not choreographically as demanding as that of Lysacek.

 

[Mathman]

True, yet this 2nd calculation was less informative and redundant. The protocols have the scores for each skating element and each program component. It's nice to see the points for each part rather than just the sum total. It's easier to break down the calculations allowing multiple people (if the calculations must be done by hand) to take part in the calculations and to spot errors.

I agree. I think that is the reason the ISU does it this way. It displays more nicely in the protocols.

 

[Medusa]

Hmm, I have a question concerning COP and don't want to open a new thread. In the "free" program, are the ladies required to do a spiral sequence? Because I am sure there are ladies out there who would give up their dogs for being allowed to do another step sequence instead of the spiral sequence. Can men do a spiral sequence (and get points for it, of course)? Shawn Sawyer has such a pretty one.

 

[gkelly]

Hmm, I have a question concerning COP and don't want to open a new thread. In the "free" program, are the ladies required to do a spiral sequence? Because I am sure there are ladies out there who would give up their dogs for being allowed to do another step sequence instead of the spiral sequence. Can men do a spiral sequence (and get points for it, of course)? Shawn Sawyer has such a pretty one.

Under the current well-balanced program rules for the long program, ladies can do a maximum of one step sequence and one spiral sequence. Men can do a maximum of two step sequences.

They're not required to do either, but they won't get points for doing something else (spiral sequence for men, second step sequence for women) instead, so they might as well do the element they can get points for.

That applies to senior and junior competition, for which the rules are set by the ISU.

Within the US, at the novice level only one sequence is allowed for points and both sexes currently have the choice between spiral sequence or step sequence in the long program. If they do both, only the one they do first will get points. At intermediate and juvenile levels, step sequences are point-earning elements and spirals are only considered as connecting moves so there's no need to do them in a sequence.

Under the old system, for several years in the early 2000s, ladies were supposed to do one step sequence and one spiral sequence and men were supposed to do one step sequence and one "field moves sequence" (could include spread eagles, Ina Bauers, shoot-the-ducks, etc., as well as spirals).

Before that, in the mid-late 90s, such sequences were suggested but not required, and even before that there were no such guidelines, there was just an expectation that skaters would do connecting moves of some sort between their jumps and spins.

Personally, I would like to see more options for both sexes. For junior and senior men's and ladies' LPs, let them have the option of choosing one or two step sequences (of different shapes), a spiral sequence, and/or a field moves sequence. Have a maximum number of total elements, so if they chose to do all four of those they'd have to give up a jump and/or spin element in exchange.

 

[Medusa]

Personally, I would like to see more options for both sexes. For junior and senior men's and ladies' LPs, let them have the option of choosing one or two step sequences (of different shapes), a spiral sequence, and/or a field moves sequence. Have a maximum number of total elements, so if they chose to do all four of those they'd have to give up a jump and/or spin element in exchange.

Me to! More choices I mean. Skaters could emphasize their strengths more, be a bit more creative and also adapt the program better to the music. If you have slow lyrical music it might be better to do a ballet-like circular steps and some graceful moves in the field - if you want to bring the house down you can dance yourself through peppy step-sequences... And I would love to see KVD Perren do a spread eagle sequence - if he gives up one of his spins for it (so painfully slow *brrr*).

 

[rain]

As always, I agree with Bianchetti about some things and disagree about others.

First, her lament over the random dropping of a certain number of scores doesn't make any sense. She claims that depending on which scores are counted and dropped there can be two different winners of the same competition possible. Isn't that always the case when there's a tough call between two skaters? Hypothetically, the only way to get the cut and dried result she seems to be hoping for is to have judges from every nation on every panel. There has always been and continues to be random choice between who gets to have a say (judge) in the final result of a competition, whether it's after they've written down marks or when they're pulling the names from the hat seems to me to be entirely immaterial.

On the argument that secret judges has hurt the sport, I entirely agree. Unfortunately, I imagine that's the most important part of the NJS to the ISU. Make it so nobody can see behind the curtain. No more scandals. That is something that is very easily remedied, whether using COP or 6.0.

My other problem with the article is that Bianchetti seems to be suffering slightly from nostalgia and it's clouding her memory of the big problems that 6.0 itself had. For example, I personally don't look longingly at the day when only the top three after the short had any chance of winning the gold unless there was some extraordinary circumstance, such as in the case of Sarah Hughes gold medal win. It failed to take into account how close those top skaters could actually be ranked together — we see it now as literally sometimes tenths of points in the scores. That person sitting in fifth just a hundredth behind the third place skater should have just as much right at a chance to win, IMO. And that's just one example that sprang to mind.

 

[gsrossano]

First, her lament over the random dropping of a certain number of scores doesn't make any sense. She claims that depending on which scores are counted and dropped there can be two different winners of the same competition possible. Isn't that always the case when there's a tough call between two skaters?

The point is, that when the results are so close that the random selection determines the results, the best answer(most reliable, most statistically significant) is the one where all the judges are used.

Of course, adding even more judges would give even a better answer, but given the panels that you have, the statistically best result is the one that uses all the judges.

 

[Joesitz (RIP)]

The problem with not dropping the highs and lows is that in a close contest
any skater who has a similar national judge in the system may sublimely favor that skater. It would only be in a close contest, imo. Collusions are no longer easy the way things are set up, praise to be. There still exists in immigrant countries that ancestral heritage will prevail albeit sublimely, imo.

I think it is working with regard to collusion. There hasn't been a beep about it since the scoring was changed.

Joe

 

[gsrossano]

Sonia is not prosposing to eliminate the single trimmed mean.

 

[rain]

The point is, that when the results are so close that the random selection determines the results, the best answer(most reliable, most statistically significant) is the one where all the judges are used.

Of course, adding even more judges would give even a better answer, but given the panels that you have, the statistically best result is the one that uses all the judges.

And what I'm saying is that when there's a judging panel there will necessarily always be an element of selection, and whether it takes place before an event or before the results are announced is, when you think about it, largely irrelevant.

 

[gsrossano]

And what I'm saying is that when there's a judging panel there will necessarily always be an element of selection, and whether it takes place before an event or before the results are announced is, when you think about it, largely irrelevant.

Absolutely wrong from a mathematical point of view. Go read a good book on statistics. You might start with the classic book by Bevington.

 

[Mathman]

Absolutely wrong from a mathematical point of view.

It is? There is a subtlety because of the business about some judges sitting for both the SP and the LP, and others coming in and out, but otherwise, doesn’t the basic idea go like this?

Method #1. There is a pool of a thousand judges. 9 are selected at random. The 9 watch the competition and give their scores.

Method #2. There is a pool of 1000. 12 are drawn at random. After all 12 watch the competition, 9 of the 12 are selected at random to count.

It seems like in either case, each judge has 9 chances in 1000 of contributing to the result.

Of course, it would be better to have all 12 sitting judges count, but that’s a different question.

IMHO the reason the random draw is stupid is not because of mathematical considerations, but just because it’s stupid. As I understand it, the purpose of the random draw (way back in the days of the Interim Judging System) was to prevent mobsters and Federation heads from being able to tell whether the judges in their pockets had doublecrossed them or not. hwell:

 

[gsrossano]

Still not the point.

You have 12 data sets and you throw 3 away. The measurement uncertainty for the nine data sets left over is worse than had you used all 12 data sets. The point is not about the chance of one judge being selected. It is about the total number of data sets in the final calculation. The measurement uncertainty is less (a better result) with 12 data sets. For random differences of opinion the measurement uncertainty decreases by the square root of the number of samples.

What it comes down to is an election poll with 1000 respondents is a better measure of the opinion of the total population than a poll with 500 respondents, which is a better measure of the opinion of the total population than with 1 respondent.

 

[Mathman]

That part is clear. I think the point that Rain was making was this. However many judges scores you decide to keep (let's say it's 50), it doesn't matter whether those 50 were selected all at once or in a two-stage draw.

If the rule was, seat 9 and count them all, that would be the same as seating 12 and counting 9.

 

[gsrossano]

That is true. But the comment was made somewhere above; what's the big deal about whether there is a random elimination of judges or not.

Sonia's point about the random draw (and mine) is this. If you brought 12 judges to the competition and they all score the competition, then throw out three sets of scores, that really doesn't do anything to stop misconduct, and you pay for it by degrading the statistical accuracy of the calculation.

If a random draw does not buy you anything in terms of stopping misconduct and it costs you in the quality of the results, then having it (the random draw) is a bad thing. It is better to use all the judges you have. The closer the results, the more judges you would like to have.

So what Sonia and I would like, is to bring the same number of judges to the competitions as now, but use all of them in the scoring, and not just some of them. (Actually in U.S. competition I would like to use even more judges in competition than we do, but the cost makes that problematic.)

 

[Mathman]

I think we are all saying the same thing now.

To me, the most convincing way to word the argument goes like this.

Fix the number N of voting judges. The more the merrier, the only restriction being the expense involved. (Right now it's nine; 12 would be better; 20 would be better yet.)

Now the question is, do we want to pay the expenses of three extra dummy judges to sit beside the real judges twiddling their thumbs?

 

[feraina]

Now the question is, do we want to pay the expenses of three extra dummy judges to sit beside the real judges twiddling their thumbs?

Exactly! Why throw all that extra money away and not count their contributions? Either let their votes count, or spend the money on something else useful, like better software for tabulating scores which actually follows the written rules.