What is the difference between Choreography and Interpretation?

[emma]

Thanks, these explanations do make a lot of sense and are very helpful.

 

[Joesitz (RIP)]

The closeness of the PCS scores, imo, proves the point that the 6.0 system still prevails on presentation.

What does it matter to break down these well intended facets of the PCS scoring system. The end result will be:

1. The music, whatever it is,is sufficient

2. The costume fits the skater and the music

3. The jumps are solid and the spins do not travel. (also covered in tech)

4. His/her grimaces are real.

5. Are there any nuances?

6. Is there a change of pace in the music?

7. Is the skater an entertainer or a competitor?

8. The home town of the skater

9. The roar of the crowd.

I think the above, and more, are better guidelines for producing the PCS score.

Joe

 

[gsrossano]

Emma asks:
The skater who deserves a larger gap because her skills are unbalanced is the exception, not the rule.

Sorry, I do not agree. In general, the best skaters (top 1/4) are decent in all skills and their marks PC marks should be similar. Similarly, the worst skaters (bottom 1/4) are weak in all skills and their PC marks should also be similar.

But for the middle half you expect (and do see) a mix of skills, and you expect the PC marks to move around -- significantly. In general, people can only discern noticeable differences in a subjective skill at the 10-15% level. That is, if you judge two skills and you perceive that one skill is "noticeably different" from another, then in assigning a value, they should differ by up to 15%, which is 1.5 points on a 10 point scale. A difference of 0.25 is 2.5% on a 10 point scale, barely perceptible by even trained observers of whatever the heck they are observing. So when I see ISU judges separating their marks consistently by only 0.25 points, and the marks from one judge to the next differ by a whole point, that says to me the judges are not marking the PCs independently as they should. You can't pick out a single judge in any one case and say they are right or wrong, but as a group something is funny.

U.S. judges are more willing to move their marks around. The biggest variation is usually in Transitions and Interpretation.

Many skaters/coaches do not include many (if any) transitions in a program. The skater may have reasonable skating skills, but between the elements they have few steps or movements, may have just a few lame half hearted arm movements. For these skaters the TR markends up in the toilete. I have gone done as much as 2.5 points in TR compared to SS skills when I see this, and I am not alone.

For Interpreations it is not uncommon to have a program that follows the rules of design built into the CH mark and so deserve a decent score there, but the performacne has no heart, not soul, no sense of the timing or the character of the music, no nuances. So the IN mark goes down significantly compared to the CH and PE marks.

Perhaps in certain championship events, where all the skaters are decent all five PC marks should be similar most of the time, and unbalance is the exception, but in a typical competition, and even at a Regional CHampionships, the range of skill of the skaters is very large, and a significant fraction of the skaters have distinctly unequal skills.

 

[gsrossano]

Emma asks:
The skater who deserves a larger gap because her skills are unbalanced is the exception, not the rule.

Sorry, I do not agree. In general, the best skaters (top 1/4) are decent in all skills and their marks PC marks should be similar. Similarly, the worst skaters (bottom 1/4) are weak in all skills and their PC marks should also be similar.

But for the middle half you expect (and do see) a mix of skills, and you expect the PC marks to move around -- significantly. In general, people can only discern noticeable differences in a subjective skill at the 10-15% level. That is, if you judge two skills and you perceive that one skill is "noticeably different" from another, then in assigning a value, they should differ by up to 15%, which is 1.5 points on a 10 point scale. A difference of 0.25 is 2.5% on a 10 point scale, barely perceptible by even trained observers of whatever the heck they are observing. So when I see ISU judges separating their marks consistently by only 0.25 points, and the marks from one judge to the next differ by a whole point, that says to me the judges are not marking the PCs independently as they should. You can't pick out a single judge in any one case and say they are right or wrong, but as a group something is funny.

U.S. judges are more willing to move their marks around. The biggest variation is usually in Transitions and Interpretation.

Many skaters/coaches do not include many (if any) transitions in a program. The skater may have reasonable skating skills, but between the elements they have few steps or movements, and may have just a few lame half hearted arm movements. For these skaters the TR mark ends up in the toilette. I have gone down as much as 2 points in TR compared to SS skills when I see this, and I am not alone in doing that.

For Interpretation it is not uncommon to have a program that follows the rules of design built into the CH mark and so deserve a decent score there, but the performance has no heart, not soul, no sense of the timing or the character of the music, no nuances. So the IN mark goes down significantly compared to the CH and PE marks.

Perhaps in certain championship events, where all the skaters are decent, all five PC marks should be similar most of the time, and substantial unbalance is the exception, but in a typical competition, and even at a Regional Championships, the range of skill of the skaters is very large, and a significant fraction of the skaters have distinctly unequal skills.

--- addition ---

I went back and looked at the marks from several club competitions I have in my computer. For one in the Midwest the judges kept the spread in the PC marks at about 0.25 points. For a few competitions on the Pacific Coast a spread in the PC marks of 0.50 to 1.00 was not uncommon.

 

[Mathman]

In general, people can only discern noticeable differences in a subjective skill at the 10-15% level.

I have always wondered if this was the rationale for marking skaters from 0 to 6.

Seven gradations = 14 per cent difference in skill level.

 

[gsrossano]

I have always wondered if this was the rationale for marking skaters from 0 to 6.

Seven gradations = 14 per cent difference in skill level.

It is also why there are 7 GoEs.

 

[Mathman]

OK, here is a little mini-case study.

I looked at Alissa Czisny, because she is an example of a skater that does many things exquisitely, while at the same time keeping her fans on pins and needles hoping she can get through the program. So we might expect to see some kind of fluctuation in the marks from one type of skill to another.

Unfortunately, there is no Program Component titled “little black Audrey Hepburn dress” (10.0).

I have not the slightest notion of what we should be looking for in these numbers, or what conclusions they support, in the context of the questions under view on this thread.

The first two numbers are TES and PCS (long program), and the last five are the program components Skating Skills, Transitions, Performance/Execution, Interpretation and Choreography.

Skate Canada: 53.57, 54.00; 6.90, 6.45, 6.90, 6.75, 6.75.

Highest = Skating Skills
Lowest = Transitions
Spread (lowest to highest) = 0.45

Cup of Russia: 33.43 (4 falls), 53.61; 6.05, 5.65, 5.65, 6.00, 5.90.

Highest = Skating Skills
Lowest = Transitions/P&E
Spread = 0.40

U.S. Nationals: 62.45 57.14; 7.18, 6.75, 7.21, 7.14, 7.43.

Highest = Choreography
Lowest = Transitions
Spread =0.68

Four Continents 48.78, 53.61; 6.83, 6.46, 6.57, 6.71, 6.83.

Highest = Skating Skills/Choreography
Lowest = Transitions
Spread = 0.37

Worlds: 47.66, 57.65; 6.46, 6.29, 6.50, 6.46, 6.57.

Highest = Choreography
Lowest = Transitions
Spread = 0.28

Can Alissa learn anything from this? Can we?

Her weakest area was Transitions in every event. (Looking at the performances, I agree. But IMHO opinion there is no point in her trying to get more fancy. Her first, last and only task is to land those jumps.)

Her strongest areas were Choreography and Skating Skills. Again, I have to agree. Perhaps her Interpretation scores suffered a little because technical mistakes prevented her from maintaining the character throughout.

In Skating Skills, I assume that is where the judges reward her for great technique on spins and spirals.

 

[gkelly]

gsrossano raises a lot of interesting points. Maybe I'll try to engage some of them when I can avoid rambling too much.

I went back and looked at the marks from several club competitions I have in my computer. For one in the Midwest the judges kept the spread in the PC marks at about 0.25 points. For a few competitions on the Pacific Coast a spread in the PC marks of 0.50 to 1.00 was not uncommon.

Does this refer to spreads within each skater's marks, or within a whole field?

 

[gsrossano]

Does this refer to spreads within each skater's marks, or within a whole field?

That is the spread in the panel average for individual skaters. The spread for individual skaters for individual judges is a bit greater.

Within a whole field in Novice and below, the marks typically span a factor of 2-2.5, say in the 2's for the bottom skaters and 4s to 5s for the top skaters. TES also has a big spread in a whole filed. As low as in the low 10s up to high 20s (even for Novice).

 

[Joesitz (RIP)]

That is the spread in the panel average for individual skaters. The spread for individual skaters for individual judges is a bit greater.

Within a whole field in Novice and below, the marks typically span a factor of 2-2.5, say in the 2's for the bottom skaters and 4s to 5s for the top skaters. TES also has a big spread in a whole filed. As low as in the low 10s up to high 20s (even for Novice).

Isn't that the same thing done in the five groups of 6 in Seniors? Regardless how one skates, whatever group the skater is in after the SP his marks in the FP will have a cap on them, and that skater will not move up. If it's a top skater who just happened to skate a poor SP, he will get the best score for the FP but not enough to move him to a higher level group. (Sarah Hughes at Worlds 03). How many skaters have you seen move up to the higher group after the SP? (Takahashi and Lambiel were in the last group to skate).

Getting out of a low group to a higher group is near impossible despite that the judges in the CoP do not compare but judge what they see. I think the judges are conerned about staying within the limits rather than to genuinely score what they believe. JMO .

Joe

 

[emma]

OK, here is a little mini-case study.


Unfortunately, there is no Program Component titled “little black Audrey Hepburn dress” (10.0)..

Thanks for the analysis, MM. So, I understand that you see these PCS marks as differentiated among components and fairly accurate (in your opinion of course) in terms of reflecting Czisny's skating....so, I think I'm following pretty good so far, right?

Ok so my question is: i don't quite follow the 10-15% thing and how that relates to the number 6 (me and math don't jive very well sometimes), but I think my question is does the range of variation seem to fit what you would expect (given what we have been discussing)....or is this a ridiculous question?

 

[Mathman]

Thanks for the analysis, MM. So, I understand that you see these PCS marks as differentiated among components and fairly accurate...

I think they are OK in the sense that I understand why Alissa's marks in certain areas were higher than others. I am undecided about whether all five marks are too close together or not.

The suspicion is always that in the back of the judges' minds they are saying, "Who's up next? Oh yes, Alissa -- a 6.5 to 6.75 skater. OK, let's see how she does."

Joe also makes a good point about whether the judges subconsciously reserve a certain range of marks for the final group to skate, for the second-to-last group, etc.

Ok so my question is: i don't quite follow the 10-15% thing and how that relates to the number 6

Actually, that's a pretty cool question.

As gsrossano mentioned, pyschologists and people who study how the brain works have determined that most people can discriminate between about 7 different levels, but not more.

For instance, you show a subject 7 sticks of different lengths. Then you hide them and pull them out one by one and ask the subject, is this the longest? the second longest? etc.

Most people will get them all right if there are only 6 sticks. If there are 7, people start to make mistakes. If there are 8 or more, most people can't do it at all. (Out of 8 sticks, is this one (seen in isolation) the 4th longest? or is it the 5th longest? -- that's too hard.)

So...when asking someone to make a judgement about the quality of something, you should offer only seven choices. The numbers 0, 1, 2, ..., 6 -- that's 7 choices. For instance, when you ask someone, on a scale from 1 to 10, how do you like my new hairdo? -- that's putting too great a burden on the judge (he can't decide between a 3 or a 4). You should instead ask him for a rating from 0 to 6..

The fraction one-seventh is approximately equal to .14. So scoring from 0 to 6 is the same as scoring in increments of 14%. That is, 0 = 0% to 14%, 1 = 15% to 28%, etc.

Gsrossano's point is that for the GOEs, the seven categories are -3, -2, -1, 0, +1, +2, +3. The average person can (just barely) tell the difference between a +1 and a +2 performance, so this works out fine.

But for the PCSs the judges must decide whether a skater's performance deserves a 6.25 or a 6.50. This is dicing it too finely -- it is impossible for judges to tell the difference on a consistent basis. The PCSs go from 0 to 10, so there are 40 different gradations, 0 to 0.25, 0.26 to 0.50, etc. This is too many and makes the judges' burden impossible.

...but I think my question is does the range of variation seem to fit what you would expect (given what we have been discussing)....or is this a ridiculous question?

On the contrary, IMHO that is THE question. I don't know the answer. Statistical methods are not of much use, because they all depend on some sort of randomness in selection of the data. I have read a lot of discussion of this question from a mathematical point of view, but have come away unimpressed.

I don't know how to go about studying this question. I think a person would have to know WAY more about figure skating and figure skating judging than I do in order to have a well-supported opinion about this.

 

[gkelly]

But for the PCSs the judges must decide whether a skater's performance deserves a 6.25 or a 6.50. This is dicing it too finely -- it is impossible for judges to tell the difference on a consistent basis. The PCSs go from 0 to 10, so there are 40 different gradations, 0 to 0.25, 0.26 to 0.50, etc. This is too many and makes the judges' burden impossible.

If 0 is not skating (e.g., clinging to the boards) and the highest value in the score is the best skating we've ever seen or expect to see, I don't know how to research whether it would be better to use a scale with the highest value set at 6.0 or 10.0 or some other number, or whether to allow judges to subdivide the marks by 0.5 or 0.25 or 0.1 or some other fraction.

Tell me which scale to use, and I'll adapt my thinking to fit it. After you do the research, you can tell me that I need to adapt my thinking further to work with the existing scale correctly or that we should adopt a different scale instead and I'll need to adapt again.

But I can point out that in a meaningful competition there will hardly ever be a need to use the full range marks.

(An event like the Taipei JGP last year where you had the Indian pair who were essentially beginners competing against the eventual junior world champs, or the first round -- previously qual rounds, now short program -- at Worlds where you have world medalists some of whom may be all-time great skaters competing against skaters who may be able to attempt the senior short program required elements but whose overall quality has no hope of coming close to top 24 are exceptions.)

Really there are two scales at work -- a macro scale of 0 to 6 or 0 to 10 that tells us what level of competition we're looking at, and a micro scale of decimal places that lets us distinguish between skaters at the same general level.

If I'm on the ice and a teenage skater I don't know also gets on and starts stroking around to warm up, I can make a good guess even before she does any elements as to whether she would most likely compete as no-test or preliminary or open juvenile or novice or senior -- whether her scores would most likely be 1s, 2s, 3s, 4s, or 5s. I can't score her skating with any finer distinction than that, even if I get to watch her skate a program with no one else on the ice.

That's the macro level.

But if I'm watching, say, an Open Juvenile competition with ten skaters at the same general level, I can distinguish which of them are at the top of that group, which at the bottom, and which in between not only overall but also broken down to distinguish to some degree between the IJS components or the old presentation criteria.

Which of those criteria should be pooled together into the same score and which are unrelated enough that they should receive separate scores is open to debate. Under the IJS, we have to use the breakdown of the components that the system specifies; under the old system we have to lump them all together.

So for Open Juvenile maybe the IJS component scores would end up ranging from 1.75 to 3.25. That's your 7 gradations right there. That's the micro scale appropriate to this small competition of skaters at approximately the same level.

Put some of the same skaters into an Intermediate competition, or mixed-age Juvenile, with some of the top intermediate or juvenile competitors, and you'll need a wider range for the whole event because the real differences between the best and worst skaters will be larger. Maybe you need a 7-point scale for the bottom third of the field, 7 grades for the middle of the field, and 7 for the top contenders, with some overlap possible.

For each individual skater?

Well, if there isn't a significant difference between how they do on one component and how they do on another, then judges should just give them the same mark for all.

The 0.25 or 0.1 subdivisions within the marks should represent the smallest increment where judges can perceive a meaningful difference within the micro scale for that competition (or subcompetition if the field is on several levels as in the previous paragraph).

Which means if you want to limit the possible gradations to only 7, then even an unbalanced skater who is at the bottom of her level on one set of skills and at the top for another set should never have marks differing by more than 1.5 with 0.25 increments, or more than 0.6 with 0.1 increments. The well-balanced skaters would get marks that vary only by one or two increments, whether at the top, bottom, or middle of the range, and less balanced ones would span the range. Sometimes a skater will be extremely unbalanced, though, in which case it would be appropriate to go outside that range.

 

[emma]

weeeeeeeeeeeeeee, this is fun! Thanks MM for the explanation - I'm still mulling over the latter part (the 40 odd gradations as waaaaay more than the 6-7 we can normally handle is clear but at the same time leaves my in a hazy/fuzzy speechlessness) and I'm still ponderng gkelly's following post...

 

[Mathman]

Gkelly's post is what I meant when I said that a person would need an extensive knowledge of figure skating and figure skating judging in order to come to a defensible opinion. Just looking at the numbers out of context (which is all a humble mathematician can do, LOL) is basically worthless.

Here are a couple of points that particularly struck me.

If I'm on the ice and a teenage skater I don't know also gets on and starts stroking around to warm up, I can make a good guess even before she does any elements as to whether she would most likely compete as no-test or preliminary or open juvenile or novice or senior -- whether her scores would most likely be 1s, 2s, 3s, 4s, or 5s.

OT -- I went to a local charity show last month. During warm-ups, mixed in with the club skaters of all ages and skill levels was the headliner, Yuka Sato. A different world.

Back on topic. That macro-micro thing is certainly the key. Some skaters are just better than others, and an experienced judge can see this. That is why the judging of figure skating can never be like a "real sport," where any fool can see whether the baseball went over the fence, whether the soccer ball went in the goal, whether the golf ball went in the hole, whether the tennis ball cleared the net, whether one football team scored more touchdowns than another, whether the high jumper cleared the bar or whether one horse's nose is ahead of the other's at the finish line.

IMHO, to expect the CoP or anything else to change figure skating from a judged sport to a measured one is unrealistic.

I don't know how to research whether it would be better to use a scale with the highest value set at 6.0 or 10.0 or some other number, or whether to allow judges to subdivide the marks by 0.5 or 0.25 or 0.1 or some other fraction.

Tell me which scale to use, and I'll adapt my thinking to fit it.

I don't know how to research this, either. And my own personal opinion is that it doesn't make any difference.

If you gave the judges a sack of different colored teddy bears and told them to throw the gold one at the skater they thought was best and the silver one to the second best, etc., to me that would be as good a system as any. (Actually, that pretty much describes the 6.0 ordinal system.)

But still, I like the New System. Although I don't think that the judges really need it, it is great for the fans. By checking out the protocols and adding up the points, the general public can begin to get a feel for the kind of thing that goes into the judging.

However, the judges themselves don't need all these numbers to know their own minds. IMHO we see this over and over. The judges decide who they think deserves to win, then they make sure that the points come out that way, even if they have to give inflated PCSs. Like Plushenko at the Olympics, Slutskaya at 2005 Worlds, Meissner at 2006 Worlds. These skaters clearly outskated everyone else. The purpose of giving out marks at all is to make sure that the best performance carries home the top prize. (Really, the numerical scores have no other purpose than this.)

JMO.

 

[gsrossano]

So for Open Juvenile maybe the IJS component scores would end up ranging from 1.75 to 3.25. That's your 7 gradations right there. That's the micro scale appropriate to this small competition of skaters at approximately the same level.

It's not about having seven gradations, it about being able to discern a delta (difference) of only 10-15%. So it's seven gradations FULL SCALE, not seven gradations within a smaller range.

If I were asked to judge heights, and the heights were 5.01 feet to 5.07 feet and I have seven graduations of 0.01, I am screwed, because there is no way I can pick out a difference of 0.01 feet without a good ruler.

Say in Novice where the scores typically run from 3.5 to 5.0 then, a difference of 0.25 is 5% of the greatest value, and discerning a difference of 5% reliably is very difficult even for a trained observer.

Also implicit in a lot of your discussion (the macro/micro thing) is you are thinking in terms of relative placements. The macro number (say scoring in the 3s instead of the 4s) puts you in the rough ball park in an absolute sense. But then using the tenths of a point to set an order of finish is relative judging. That is pretty much exactly the way it is done under 6.0. For example, in a typical Juvenile event (formerly under 6.0) the mid-point would be a 3.0 and the range would usually be form 2.5 to 3.5 The midpoint is a rough absolute measure of where the average Juvenile should be, but the tenths are used in a relative sense to place the skaters.

When Mathman talked about 40 steps in a 10 points scale being too many, I was wondering if anyone would come back with "but 6.0 has 60 steps in a 6 point scale, isn't that even worse."

Well, for absolute judging it would be worse, but in 6.0 the judges place the skaters first, then assign a score second, and nobody cares if the exact numbers are meaningful in an absolute sense, and differences in points from one skater to another have no great significance.

But in IJS the marks are supposed to be have significance in an absolute sense and the differences in the numbers are also supposed to have significance in the absolute sense, with each skater compared to an abstract standard and not each other.

One more example of the difference between the absolute and relative judging, is Mathman's sticks example. If I give you 18 sticks, say between 3.5 and 5.0 inches long, and ask you to place them in order of increasing length, if you compare them one to another you will probably do a decent job, and without having to know how long any one of them is in an absolute sense. On the other hand, if I show them to you one at a time and ask you to tell me how long each one is in inches (without access to a ruler) and put them in the order of your length estimate you will not do nearly as well.

In the absence of markers (a measuring reference such as a ruler) human perception is significantly better at relative judgements than absolute judgements. That is why the math underlying the 6.0 calculation method is superior to the IJS calculation method -- IJS relies on absolute judgements AND it does not provide the judges a decent ruler. (In fact for the PCs there are no rulers at all.)

 

[gkelly]

If I were asked to judge heights, and the heights were 5.01 feet to 5.07 feet and I have seven graduations of 0.01, I am screwed, because there is no way I can pick out a difference of 0.01 feet without a good ruler.

True. But you could probably estimate people's heights (barefoot and standing up straight) within the inch knowing or expecting that they're all going to be between 5'1" and 5'7", even though you also know that there's a much wider range of potential human heights. Inches are meaningful differences to the trained human eye; hundredths of feet are not.

Also implicit in a lot of your discussion (the macro/micro thing) is you are thinking in terms of relative placements.

I don't think it's possible to get away from that entirely in the context of a competition. The alternative would be to say that everyone who falls within the same general range should get the exact same score. Which, if they all do the same elements, could mean a lot of ties.

If the results are determined by absolute scores, it's not really where the numbers fall on the macro scale of how this competition compares to all possible levels of skating that determine the results. For the PCS it would be the size of the gaps between each skater's scores (and how those compare to the gaps in technical base score, leaving out GOEs for the moment). Whether the judges use a scale of 3.5-5.0 or 5.5-7.0 is completely irrelevant to the results if they're consistent about it -- the actual scores would be 10 points higher for everyone in the latter case, but the standings would be the same.

The judges who use the narrowest gaps between skaters would have the least effect on the ultimate results, and those who use the widest gaps would have the most control of the results. That's my biggest concern about an absolute scoring system. Assuming everyone is judging honestly, some judges will have a disproportionate effect just because of the way their brains happen to translate perceived differences into numbers. And if anyone does want to cheat, it will be those who can learn to use the largest possible gaps without having their marks thrown out as high and low who will have the most effect.

Of course, you could also make the PCS increments larger or smaller or factor them differently to best balance the size of the differences in tech content common at a given level, similar to the way the GOEs scale according to the base values and the


Another way to handle that would be to award +3 to -3 for each component to a predetermined base mark for that level. E.g., if the base mark for novice is 4.0 and the increments are 0.25, then +1 on a given component in a novice competition would translate 4.25.

One more example of the difference between the absolute and relative judging, is Mathman's sticks example. If I give you 18 sticks, say between 3.5 and 5.0 inches long, and ask you to place them in order of increasing length, if you compare them one to another you will probably do a decent job, and without having to know how long any one of them is in an absolute sense. On the other hand, if I show them to you one at a time and ask you to tell me how long each one is in inches (without access to a ruler) and put them in the order of your length estimate you will not do nearly as well.

With practice at recognizing the differences between 3.5", 3.75", 4.0", etc., I think many people could learn to recognize lengths with fairly good accuracy within the quarter inch. Yes, occasionally there would be mistakes and I might assign a 3.85" stick a score of 4.0 and a 3.90" one a score of 3.75 and thus get them in the wrong order, but if you got a panel of nine trained observers to rank the sticks by estimating each one's length without direct comparison to each other or to a ruler, I think you would end up with a pretty accurate ranking.

Most of us can easily estimate whether the temperature is in the 30s, 40s, 50s, etc., on the macro weather scale. I once had a job going around my college campus while it was closed for January break, walking into each building and looking at thermometers mounted on the walls to make sure the room temperature was as close as possible to 50 degrees (to save energy without risking the pipes freezing). After a few weeks of checking thermometers 8 hours a day, we got to the point where we could walk into a building and estimate the temperature within one or two degrees just by feel. I couldn't estimate temperatures that accurately all these years later, nor could I do so outside the range of about 40-52 degrees even that month, but it is possible to train oneself to make fine distinctions on the micro scale.

IJS relies on absolute judgements AND it does not provide the judges a decent ruler. (In fact for the PCs there are no rulers at all.)

I agree there is lots of room for improvement at defining the benchmarks.

 

[SeaniBu]

For Interpreations it is not uncommon to have a program that follows the rules of design built into the CH mark and so deserve a decent score there, but the performacne has no heart, not soul, no sense of the timing or the character of the music, no nuances. So the IN mark goes down significantly compared to the CH and PE marks.

Perhaps in certain championship events, where all the skaters are decent all five PC marks should be similar most of the time, and unbalance is the exception, but in a typical competition, and even at a Regional CHampionships, the range of skill of the skaters is very large, and a significant fraction of the skaters have distinctly unequal skills.

This is one of the best threads ever IMO.

Gkelly's post

:agree:100%

However, the judges themselves don't need all these numbers to know their own minds. IMHO we see this over and over. The judges decide who they think deserves to win, then they make sure that the points come out that way, even if they have to give inflated PCSs. Like Plushenko at the Olympics, Slutskaya at 2005 Worlds, Meissner at 2006 Worlds. These skaters clearly outskated everyone else. The purpose of giving out marks at all is to make sure that the best performance carries home the top prize. (Really, the numerical scores have no other purpose than this.)
JMO.

:agree:100%, that is what really happens. Is that good? - is it meant as some way to make a career more important?

So make the "numbers" fit what ever fits in the fans mind, seems reasonable if the marks are just there for us anyway?

I agree there is lots of room for improvement at defining the benchmarks.

Could it be the sport should put more emphasis on a "season" much like many sports do? ~ usually by "play-off" style. It would seem that having a "score earned in year" as a "benchmark" might work. ???

IMHO, to expect the CoP or anything else to change figure skating from a judged sport to a measured one is unrealistic.

A balance???

 

[gsrossano]

Inches are meaningful differences to the trained human eye; hundredths of feet are not.

Really not relevant to my point -- which is there is a fundamental (and well established) limit to the difference in performance a human observer can reliably perceive and the ability of humans to accurately assign a numerical measure to any observable phenomenon. You (and the designers of IJS) give yourself and the average person far more credit than they deserve for what is humanly possible in this area. You may think you can make accurate, repeatable, unaided-eye measurements of something to 1.5% (1 inch out of 60), but I know for sure I can't, never could, never will. The well established limit in every study ever done on the subject is 10-15% at best, and for some types of observations no better than 20-25%. In addition to the real limits to human perception, another interesting aspect of such studies is that humans always think they are doing a much better job in their ability to observe, than they actually are doing when put to rigorous test.

And for those who would like to learn about these fascinating subjects (and see the source material for my comments -- that I'm not just making this stuff up), you are invited to wander down to you local university library and thumb through the Journal of Experimental Psychology, or the Psychological Review, or the journal of the Psychonomic Society. Or search the Internet for the phrases "human perception and information theory", "psychophysics", "Weber's law", "Steven's law", or search for the work of Stanley Smith Stevens (originator of Steven's law) or Louis Leon Thurstone.

The limit human perception plays in the ability of a person to make absolute measurements has been well known since the time of Galileo (late 1600's). Modern study of the effects of human perception on observations can be trace to Ernst Heinrich Weber in the early 1800s at Leipzig University, and others there in the mid-1800s. There are laboratories at the University of Chicago (est 1930s), and the University of North Carolina (est. 1952), among others, that are devoted to this (you can search for them on the Internet too). The laboratories at UC and UNC Chapel Hill were established by Thurstone who was a leader in these fields in the U.S.

(But what this has to do with the meaning of Interpretation beats me. Seems we have gotten a bit off topic!)

 

[Mathman]

(But what this has to do with the meaning of Interpretation beats me. Seems we have gotten a bit off topic!)

But it has everything to do with the question of whether the Program Component Scores can really do what the ISU advertises.