Relative difficulty of types of jumps: triples versus quads

[Mathman]

Obviously they don't put sensors on Jupiter or the exoplanets themselves. So what techniques do they use to measure the rotations?
And are they measuring rotation with respect to a fixed point or to the orbit of the planet doing the rotating or to a point on earth, which is traveling?

Could similar techniques be used on a smaller scale to measure jump rotation? And if so, would it be cost effective to do so?

I am not an expert, but my impression is that in reality none of this has any relevance or application to anything so complicated as figure skating. In the case of Jupiter — visible to the naked eye, arduously studied for millennia, and close enough that it presents a detailed disc even with the kind of telescopes that they had in the 1600s — just eye-balling while appealing to the pendulum clock that had just been invented was quite adequate for any practical or scientific purpose of the time.

For up-to-the-minute searches for exoplanets circling stars a billion miles away, the method is actually pretty cool. As the planet rotates, information recovered from spectrographic analysis of chemicals in its atmosphere is blue-shifted on the edge of the planet that is turning toward the viewer and red-shifted on the side that is turning away, due to doppler effects. In principle you could measure how fast a skater in a white costume is spinning by comparing the redness of her red side with the blueness of the blue.

As for your first question, astronomers would try to calculate the actual number of rotations with respect to the distant fixed stars, which calculations are possible due to the detailed knowledge that we have of the motions of the earth (camera).

For figure skating, I think the best plan in theory would be to mount a stationary camera way high in the air (build taller skating arenas) that could cover the whole ice surface at once. This would minimize distortions of perceptions in angular measurement and would eliminate moving camera complications The role of the sensors would be to determine the exact times of lift-off and landing. Cheap except for the part about building taller buildings.

 

[Anna K.]

I wouldn't rule out sensors attached to skates. It's somehow funny that we can hear music based on data collected from the skates do but can't make out of this anything that could help to develop the sport
In theory, such sensor is capable to provide lots of of interesting data:
- mileage (distance covered)
- travelling speed & speed changes
- travel trajectory & curve
- pressure on skate (contact with ice)
- distance covered on one blade (interrupted and uninterrupted)
- force of impact (toe assistance)
The number of rotations can be calculated based on the length of the jump, distance traveled in the air, and the trajectory right after the landing (if the skater needs to turn i.e. from landing forward). It is even easier when counting spin rotations or calculating spin speed and, thanks to digital technologies, such calculation and sorting/ranking the data can be done in an instant.
It says nothing about the arms and body position but this is fine because no machine can tell anyway if the body/arm position is aesthetic or not. So, we can expect an assisting tool from this, not a decision maker. Which I guess is also fine
The only problem is, such tool requires development to fit for specific tasks. It would also need prolonged testing and data analysis to create an effective software. The development part is always expensive although the production part may be comparatively cheap; and, above everything, somebody needs to order such tool to start with. But, once created, it could be great assistance for judges, interesting intermediate data pool for journalists and general public, and useful tool for coaches and skaters who want to improve their skating and compare it with the best performances.

 

[eppen]

This and the quads data are super interesting thanks for posting these.

It seems the loop is just not a jump skaters like. The attempt and success rate of a triple loop is significantly less than a triple lutz and more recently even less than a triple axel …. but I don’t think people could argue that means a triple loop is harder than a triple lutz - and certainly not a triple axel.

Just totally ignoring the whole rotation discussion...

This made me think about what the jumps are used for, esp. triples, and how this affects their numbers.

3T is the preferred combo 2nd part from time immemorial (well, end of 1970s) and it gets done by most skaters 1-3 times/competition. Hence lots of attempts.

3S was until 2010s attempted usually on its own or as the first part of a combo. Since the mid 2010s it has started to become a common 2nd part of a combo. Hence perhaps a slight increase in attempts.

3Lo has been and is still usually a single jump and gets predominantly done in FS only. The almost obligatory 4th jump in the free... It is attempted in general about as often as 3S despite the multiple uses of the 3S. The spike in 2022-23 is due to 3Lo being part of the junior SP requirements.

3F and 3Lz are also often single jumps or first parts of a combo. Attempted in both programs depending on whether the guy can do two quads. As I said, higher BV and probable good consistency will bring more points in addition to quads and 3As, so more frequently used.

Some additional insights could come out of this if I (or someone?) had the time and energy to look at what's behind the simple raw data...

E

 

[Mathman]

The only problem is, such tool requires development to fit for specific tasks.... The development part is always expensive although the production part may be comparatively cheap; and, above everything, somebody needs to order such tool to start with.

I agree that this is the problem, perhaps an insurmountable one.

We can compare it to a pharmaceutical company that invests 100 million dollars to develop a new drug. Once the R&D phase is complete, the product costs 2 bents a pill to produce. It is then marketed for $500 a pill and eventually the company recoups its investment and makes a big profit.

The ISU, however, does not have the hundred million, nor is figure skating overall sufficiently big business to finance the venture. The only way I can see this happening is for figure skating to piggyback onto existing technology or technology developed for some other (more lucretive) purpose.

It says nothing about the arms and body position but this is fine because no machine can tell anyway if the body/arm position is aesthetic or not. So, we can expect an assisting tool from this, not a decision maker.

Although... I can imagine an artificial intelligence judge that could be taught to recognize what humans generally regard as esthetically pleaseing positions and movements of the body. This would be a fascinating exercise in the field of AI, whether it ever amounted to anything of value to figure skating or not.

 

[Anna K.]

I agree that this is the problem, perhaps an insurmountable one.

We can compare it to a pharmaceutical company that invests 100 million dollars to develop a new drug. Once the R&D phase is complete, the product costs 2 bents a pill to produce. It is then marketed for $500 a pill and eventually the company recoups its investment and makes a big profit.

The ISU, however, does not have the hundred million, nor is figure skating overall sufficiently big business to finance the venture. The only way I can see this happening is for figure skating to piggyback onto existing technology or technology developed for some other (more lucretive) purpose.


Although... I can imagine an artificial intelligence judge that could be taught to recognize what humans generally regard as esthetically pleaseing positions and movements of the body. This would be a fascinating exercise in the field of AI, whether it ever amounted to anything of value to figure skating or not.

That's why I am rather looking towards DIY department. For our technologically advanced youth, building a gadget is an absolutely achievable task. Of course, such tool won't be used as a judging aid. But it can be used as a training aid if the young technology genius is, by coincidence, doing figure skating is his/her free time
If such DIY gadget goes viral, then it's possible that it gets adopted as a judging aid as well. We know that ISU is susceptible to whatever goes viral

In fact, this is an interesting topic: what technologies skaters currently use or would like to use in training? It's a bit off-topic here, but in the Lutz Corner...

 

[Magill]

These two bits could be interesting for some of you as this discussion is turning towards technology used for assessment, whether for training or actual judging purposes.
The first one is technology apparently currently used sort of routinely in Japan to assist training/periodical assessment of a skater's jumps, the other one is a research paper from Nagoya University, published and accepted into a few science conferences last year, focused on jump assessment based on its translation to 3D image and then assessed by AI based on its data library (instead of 2D video recordings). To summarize, it turned out that 3D-based technology is much more accurate than video-based assessment although still imperfect. The research paper is just an example for discussion and just one of numerous similar studies which, I believe, have been going on without much external financial support.

https://arxiv.org/pdf/2408.16638

 

[Mathman]

About Yuzuru Hanyu's contribution to this endeavor (summaries of his project are available in English, if not the whole thesis), the salient point is that Hanyu is a skater, not an electrical engineer. As I understand it the main thing that he wanted to investigate is whether the type of sensors that were already in use could be comfortably and unobtrusively attached to boots and clothing without interfering with free movement or affecting performance. His conclusion, no problem. And also not outrageously expensive.

The idea of using such technology in training, not necessarily for judging, is quite promising. Plus you get some cool computer-generated footage of ghost skaters twirling about. like in the Mihara clip. You could make a science fiction movie out pf it: magic invisible figure skaters from outer space save the world!

Edit: About the second clip in the post above, if Hanyu can still do that now (August, 2024), I hope he decides to return for the 2026 Olympics.

 

[Magill]

About Yuzuru Hanyu's contribution to this endeavor (summaries of his project are available in English, if not the whole thesis), the salient point is that Hanyu is a skater, not an electrical engineer. As I understand it the main thing that he wanted to investigate is whether the type of sensors that were already in use could be comfortably and unobtrusively attached to boots and clothing without interfering with free movement or affecting performance. His conclusion, no problem. And also not outrageously expensive.

The idea of using such technology in training, not necessarily for judging, is quite promising. Plus you get some cool computer-generated footage of ghost skaters twirling about. like in the Mihara clip. You could make a science fiction movie out pf it: magic invisible figure skaters from outer space save the world!

I think Mai's clip is not the same technology which was used by Yuzuru, still I do think his thesis was a significant contribution to work of others interested in the subject, not only because it was pioneering, but also precisely for the reason you're quoting : he is a skater, and more so, the one whose credentials cannot be reasonably put into any doubt. Furthermore, he also graduated from the Human Information Science Department which is a kind of academic department where this kind of research would be done. So you can say his credentials are of double value as both a skater and an academic researcher,
As for Yuzu clip, I have seen quite a few researchers using footage of his skating to create data bases for their reference libraries, and I don't find it surprising. I mean, who better to learn from than the best, and who might have better credentials from a double OGM?

Edit: About the second clip in the post above, if Hanyu can still do that now (August, 2024), I hope he decides to return for the 2026 Olympics.

The clip was used for research in August 2024 but it comes from 2018 OG, of course. Yet Yuzu was just performing a full-scale rendition of Ballad No. 1 as a part of his Echoes of Life show staged December 2024 to February 2025.
In fact, the last show was held a week ago and he managed to nail a flawless performance there .
And this, while Ballad was skated some 50 minutes into the solo show, as some 9th or 10th programme, directly after he skated some 10-minutes-long classic medley of Brahms, Bach, some other Chopin and the likes. Here you can see play by play comparisons divided into two parts (I have posted it in the show's thread but repost here for your easy reference).
As for coming back for 2026 Olympics, he's being asked this question in almost every interview and he always says he has no such plans, he's enjoying his pro career and the artistic freedom that goes with it.

 

[Anna K.]

I wonder if the data provided by sensors in Mai Mihara clip can actually say something about the topic of this thread: the difficulty of jumps. It's clear that no videos, 2D or 3D, can deliver such data because involved loads are neither measured nor analysed: we only know if the jump is correctly assessed as a certain type of jump or not.

Thus far, the only data we can look up as data regarding the difficulty is statistics provided above by @eppen
But can the jump preference and the success rate completely define the difficulty of a jump?

If we suppose that it rather depends on physics involved (i.e. when changing/not changing a leg, direction, or curve), then what could the actual measured data say about it?
Could such data, in theory, change our attitude towards figure skating as a sport?

 

[gkelly]

Thus far, the only data we can look up as data regarding the difficulty is statistics provided above by @eppen
But can the jump preference and the success rate completely define the difficulty of a jump?

Not entirely, because the frequency of attempts of different kinds of jumps will be affected by the rules at the time, and also the Scale of Values under IJS (which has been adjusted a few times in the last 20 years) and the perceived preference of judges for specific jump content under 6.0.

If everyone "knows" that a 3Lz is more valuable than a 3Lo, they're going to spend more time training 3Lz so they can include more of them in their programs. If the values were equal and skaters were not allowed to repeat any jumps in the free skate, we'd get a better sense of whether (or by how much) 3Lz actually is more difficult.

If we suppose that it rather depends on physics involved (i.e. when changing/not changing a leg, direction, or curve), then what could the actual measured data say about it?
Could such data, in theory, change our attitude towards figure skating as a sport?

Well, it might change our attitude toward the ice jumping part of figure skating.

For the skating itself, you'd need to account for things like whether/by how much a FI bracket is more difficult than a BI three, a FO mohawk than a BO choctaw, etc. And also whether/how much the rules (for step sequence levels and also for Skating Skills -- or Transitions back when that had a separate score) encourage using the hardest turns as many times as possible or whether the rewards are higher for including maximum different kinds of turns with both backward and forward, inside and outside, and clockwise and counterclockwise entries.

And then both for on-ice turns (and other steps) and also for jumps, there would need to be ways to quantify the relative difficulty of different ways of combining two or more jumps, combining multiple turns on one foot, etc. And also combining various turns with jump takeoffs or landings.

And how variations in body position make a jump or a turn easier or harder or have no effect on difficulty per se but may affect the perceived quality or aesthetic value.

Experienced skaters, coaches, and judges, and commentators who were once elite skaters themselves and may or may not have also coached, will have a pretty good sense of what's relatively easy, what's moderately difficult, and what's really difficult. But individual skaters/former skaters themselves may have different experience about what was more difficult for them personally, which will influence their perceptions.

How valuable is it to parse differences in difficulty to the hundredths of points, or are whole points or tenths of points sufficient? (Not accounting for second-half bonuses, averaged GOEs, etc., that will introduce those hundredths (and finer, before rounding) mathematically.)

 

[Mathman]

Well, it might change our attitude toward the ice jumping part of figure skating.

For the skating itself, you'd need to account for things like whether/by how much a FI bracket is more difficult than a BI three, a FO mohawk than a BO choctaw, etc. And also whether/how much the rules (for step sequence levels and also for Skating Skills -- or Transitions back when that had a separate score) encourage using the hardest turns as many times as possible or whether the rewards are higher for including maximum different kinds of turns with both backward and forward, inside and outside, and clockwise and counterclockwise entries.

Thank you for this reminder. In many of these discussions, when it comes to difficulty and the appropriate response thereto in the IJS it seems like questions about jumps sweep everything else off the debating stage. Is a triple flip 8% harder than a triple loop (5.30 to 4.90)? Are we sure it is not closer to 6% harder ( 5.3 to 5.0)?

In the first version (2003) of the IJS it was 10% (5.5 to 5.0).