Use of technology in figure skating tech calls
- Thread starter jinhamasaki
- Start date
Really?
Only one person like the data operator doing a sanity double-check of the images and the resulting data would make the possibility of another huge scandal rather likely.
I'm all for technical assistance, but where there is technical support humans tend to trust blindly, therefore any supporting system must be very very good.
- Joined
- Jan 1, 2013
Yep, sounds like a great idea. Give someone who isn't a judge (and who at international events is provided by the host country) the responsibility. Way to go.
This^
I mean for goodness' sake, it was bad enough when people after Sochi were claiming that absolute proof of a scandal was that the data and replay operators were Russian. Imagine if they were given actual power? Holy toledo. I'm not a judge. I don't want that on me.
Okay. Honest question here. Who here actually understands how a judging and technical panel work? How many posters? There just seem to be a lot who don't have a clue.
+1 seems to be too much power to one guy.
Id understand if maybe the whole technical pannel would be responsible for that. For example, the automatic video analysis would be avaliable for them as a sort of additional info - if the automatic system flagged a huge underrotation (i guess that, while not really precise, it could do quite well in saying "underrotation above 60 degrees here, please look at slowmo, humans, to see if it is UR / downgrade"), the tech pannel would have to view the slowmotion to check it and then make a decision.
Also, the software results could be avaliable maybe not for fans (too much potential butthurt and blablabla - "it was 89 degrees, and it still got dinged, you cheaters!!!") but for the skaters, coaches and feds - in case of some issue like "the coach seen the jump was ok, the software said it was 30 degrees short, and it was still dinged as UR", there would be some base to complain about the results (as i understand, ISU gives quite a short time for that).
Id understand if maybe the whole technical pannel would be responsible for that. For example, the automatic video analysis would be avaliable for them as a sort of additional info - if the automatic system flagged a huge underrotation (i guess that, while not really precise, it could do quite well in saying "underrotation above 60 degrees here, please look at slowmo, humans, to see if it is UR / downgrade"), the tech pannel would have to view the slowmotion to check it and then make a decision.
Also, the software results could be avaliable maybe not for fans (too much potential butthurt and blablabla - "it was 89 degrees, and it still got dinged, you cheaters!!!") but for the skaters, coaches and feds - in case of some issue like "the coach seen the jump was ok, the software said it was 30 degrees short, and it was still dinged as UR", there would be some base to complain about the results (as i understand, ISU gives quite a short time for that).
- Joined
- Jan 13, 2014
If it's not available to the public then its of no use. Judges won't be held to interpret what the computer says. Look what happened to VW when someone from the public scrutinized VW's results :think:
Unless computer is 100% precise, releasing the results of the computer analysis to the public will be a shitstorm source, nothing more.
Public tends to think computers are better than human judges when the computer results are more favorable to their skaters. And believe that the computers suck and are useless when the result is not favorable.
Also, public tends to not know what is precision, error etc. So for example, a result like 89 will be interpreted exactly, while it could be for example 89+-10.
So yeah no, i would release those result to the skaters and their coaches and the feds, but not to the public.
As for endorsing the judges to do stuff... For example, it is quite easy to code a program that will not allow the tech pannel to proceed to the next element if the system flagged the jump as close to UR.
Honestly, i would rather just start asking more from the judges - for example require a formal discussion and explanation of ALL judges if one judge gives a +3 GOE on something that another one gives a -1 GOE, to go with the protocols, instead of giving public computer results and letting it shitstorm.
During post-program reviews, you mean? They need to be able to proceed to the next element in real time during the actual program.
In principle I like the idea of public statements about reasoning for scores, but how would it work in practice?
Just flag outliers? In many of those cases, I suspect that the judge's explanation will be "oops, data entry error!" If the computer can instantaneously identify outliers that would raise concern, it might be better to have the computer question the judge to verify the entry before finalizing the scores to be announced. That would make the K&C wait as much longer as it would take for the judges to confirm yes or no they did want to enter that GOE for that element; in most cases that should only take a few seconds.
So if a judge made a mistake on applying mandatory GOE reductions or missed a change in the call during the review period or
entered the GOE for a previous or following element in the wrong box, they would be alerted to the mistake and have the opportunity to fix it.
But flagging outlier scores for the judges would also alert judges that their score was out of line -- so if they're concerned about staying in the corridor, they might change the score at that point even if they hadn't made an error, to try to bring themselves back into line.
But what if they guess wrong about which direction they were out of line and overcorrect in the wrong direction?
Also, the public may very well want to know why all/almost all the judges gave negative GOE to an element that looked perfectly good to them. Or why they gave +3 to one skater's good element but only +1 to another skater's good element that look fairly similar to a casual viewer. A program that flags the outliers for explanation won't flag decisions where the judging panel agreed with each other and disagreed with the general public for reasons that were not obvious.
The public may also want to know why the tech panel called certain levels on spins or steps (or other leveled elements in pairs and dance).
There's no way for the computer to guess which tech panel decisions, or which consensus decisions of the panel, are going to be controversial to the public.
That was just one example...
Id actually want the judges to explain their GOEs for example... I dunno, it could be even as easy as replacing the value input by checkpoints - so instead of typing 2 goe, judges would check the points that seem to apply and then select out of possible goe values for that combination of checkpoints. Same for tech pannel - for example, for a step sequence level 4, the skater must check a series of requirements - so the tech pannel could check/list the elements (for example, a full list of steps with CW and CCW columns, and then the tech pannel inputs the number each step was done - probably wouldnt take that long as they theorically need to do something like that in their head) and then the classification is automatically done on that.
Thats something i would trust to a computer, as it is just a matter of accounting points.
And lots of other stuff.
Id actually want the judges to explain their GOEs for example... I dunno, it could be even as easy as replacing the value input by checkpoints - so instead of typing 2 goe, judges would check the points that seem to apply and then select out of possible goe values for that combination of checkpoints. Same for tech pannel - for example, for a step sequence level 4, the skater must check a series of requirements - so the tech pannel could check/list the elements (for example, a full list of steps with CW and CCW columns, and then the tech pannel inputs the number each step was done - probably wouldnt take that long as they theorically need to do something like that in their head) and then the classification is automatically done on that.
Thats something i would trust to a computer, as it is just a matter of accounting points.
And lots of other stuff.
So the details protocols that now fit two skaters per page would need a whole page for each skater, but there would be a lot more information available.
Actually, it would probably need more than one page per skater if every judge's GOE points for each element were listed.
Should these extra-detailed protocols be published for all events?
Most people read them electronically as PDFs, not on paper. If they do want they can print out relevant pages but we wouldn't expect printing whole protocols (and the attendant waste of paper) to be a matter of course.
Actually it would probably add a lot more time for each tech panel member to name every turn and step out loud and the data input person to input them all.
It would be a lot simpler just to name the features awarded. But for singles step sequences there are only four features available, so if a skater earns level 4, we know the tech panel awarded all of them.
If the skater earned level 3, it might be interesting to know whether the reason was not getting credit for "complexity" of turns and steps or not getting credit for the upper body movement (most likely) or the clusters.
It's possible to get the turns in both directions feature 2 without the variety or complexity required for feature 1, but I'm not sure it's possible to get credit for complexity in feature 1 without also including enough turns both ways to earn feature 2.
In ice dance it's more complicated.
And there would still be questions about the PCS, which are more subjective and harder to explain the process of quantifying the qualitative assessments. It would certainly be interesting to know what the judges were thinking to come up with the numbers they arrived at, but I can't think of any easy way to communicate that.
Actually, it would probably need more than one page per skater if every judge's GOE points for each element were listed.
Should these extra-detailed protocols be published for all events?
Most people read them electronically as PDFs, not on paper. If they do want they can print out relevant pages but we wouldn't expect printing whole protocols (and the attendant waste of paper) to be a matter of course.
Actually it would probably add a lot more time for each tech panel member to name every turn and step out loud and the data input person to input them all.
It would be a lot simpler just to name the features awarded. But for singles step sequences there are only four features available, so if a skater earns level 4, we know the tech panel awarded all of them.
If the skater earned level 3, it might be interesting to know whether the reason was not getting credit for "complexity" of turns and steps or not getting credit for the upper body movement (most likely) or the clusters.
It's possible to get the turns in both directions feature 2 without the variety or complexity required for feature 1, but I'm not sure it's possible to get credit for complexity in feature 1 without also including enough turns both ways to earn feature 2.
In ice dance it's more complicated.
And there would still be questions about the PCS, which are more subjective and harder to explain the process of quantifying the qualitative assessments. It would certainly be interesting to know what the judges were thinking to come up with the numbers they arrived at, but I can't think of any easy way to communicate that.
Well yeah, PCs are still an issue =/
kinda wish they would change the PCs to something like a comparison - is this skater better than a number of previous ones - for example the last 3 (having a few will make it easier to compare as you could place skater C between skater A and B, instead of just saying C was better than B). And then just scale it up giving the best skater a 10 and the worst skater a 0, something like that?
kinda wish they would change the PCs to something like a comparison - is this skater better than a number of previous ones - for example the last 3 (having a few will make it easier to compare as you could place skater C between skater A and B, instead of just saying C was better than B). And then just scale it up giving the best skater a 10 and the worst skater a 0, something like that?
Not sure how that would work.
Do you want to try applying your concept to a JGP ladies' short program with 30+ unseeded skaters? Or even an ISU championship SP with that many or more seeded ones?
Might be interesting to choose a season where a JGP winner also competed at Euros/4Cs/Worlds, to see how the scores for similar(?) performances come out differently if she places 1st or ~10th in different fields.
Do you want to try applying your concept to a JGP ladies' short program with 30+ unseeded skaters? Or even an ISU championship SP with that many or more seeded ones?
Might be interesting to choose a season where a JGP winner also competed at Euros/4Cs/Worlds, to see how the scores for similar(?) performances come out differently if she places 1st or ~10th in different fields.
actually as you probably noticed i am a scoring nub =)
but in general, i guess, the final goal of the scores is to rank the skaters - as we want to know who is the best, second best etc...
As for the skaters numbers - this is why i suggested taking the last 3 skaters instead of comparing the skater with each and every one. Actually, not sure if that would work =D
If the goal is just to rank the skaters, we can go back to 6.0 scoring.
Although even there, though the actual scores could vary noticeably based on the size and strength of the field, and on skate order, there was some rough consensus in judges' minds about what constituted 5.0 skating vs. 4.5 or 5.5, for example.
It sounds like you're proposing to remove all correspondence between the scores awarded and an external standard of quality, with the only point of comparison being the other skaters who have already/recently skated in this event.
That makes it almost impossible for the sizes of the differences between the numbers to correspond meaningfully with the size of the differences between the quality of skating, in each judge's mind. So I don't see how such comparative numbers could be combined meaningfully with the element scores, which would become even more objective in comparison with technological assistance.
I don't think it would work well at all, as I'm imagining it. But if you have worked out the details and want to make a case that it would work, I'd be interested to discuss.
ETA:
Also, with IJS, it is possible to score a program in isolation, with no other skaters in the field. Of course a skater in a one-person event will finish first (and also last
), but the scores give feedback on what they did, what they got credit for, what the judges thought of it. Input from technology about their speed or jump height or jump rotation, etc., would also be of interest if available.With ordinal judging under 6.0, skaters might get to see the marks the judges gave them, but essentially their scores are a row of straight first-place ordinals. The judges can give those without watching the program. If they know the general skill level of the skater, they could give 6.0 scores before watching the program and might see little need to adjust them after watching it, if they actually do watch.
These facts make little or no difference to international-level events with plenty of entries.
They make a difference at club competitions, or national championships in countries with few skaters, especially for events other than ladies' singles.
Similarly, in actual contested events with at least two skaters, but not many more than two, sometimes there can be large obvious gaps in quality between the skaters. So the rankings might be unanimous that skater A was best, skater C worst, skater B somewhere in between. A scoring system that forces the numerical difference between A and B to be identical to the difference between B and C doesn't tell skater B, or interested members of the public, anything about how skater B actually did. Was she almost but not quite as good as A? Or barely better than C?
And what if skater D skates much better in the free skate than in the short program, and A skates much worse? Should the "how much better" in each program figure into the decision as to who wins overall? Or only the number of other skaters between them in each program?
There are arguments both for and against preserving margin of victory in each program. Personally, I lean in favor, especially if the sizes of the point differences are meaningful and if there are more opportunities in the freeskate to make up deficits than there are opportunities to incur them in the short program.
Last edited:
- Joined
- Feb 21, 2006
Some skaters may not take the speed into the jumps that others do.....because their rotational speed in the air is quicker, and they can successfully land the triples or quads with less speed going down the ice. All physics.
My issue with all of this is multi-faceted.
The ISU has their collective heads in the sand. They have to change and embrace technology. It is one of the facets that has moved the X-Games forward as well as many other sports. Fans/enthusiasts want, no strike that...DEMAND that they "are part of the action" and that is what tennis has done through Hawk-Eye and player challenges, Phantom camera slo-motion, or the NFL has achieved through multiple camera angles on plays. MLB will eventually allow managers to challenge calls by the umpires (Limiting the number of reviews on pitches/game). The ISU needs to work with broadcast companies so that standardized multiple camera angles are available at the SGP/Championship events so that everyone can see the same 3 or 4 views ...ie the tech panels will have some of they power diminished, but the reality is...embrace new technology or risk loosing most of your fan base, i.e. advertising $, etc.
- Joined
- Jan 1, 2013
Once again, if you start introducing three or four camera angles you are massively slowing down the process. Fans want to "be involved", well...how many fans are going to sit through a competition where a single warmup group for a short program takes an hour and a half to get through?
I'd also like to insert a friendly reminder here that the technical panel, as with all judges, are volunteers.
I'd also like to insert a friendly reminder here that the technical panel, as with all judges, are volunteers.
A large part of figure skating's fan base has been people who are arts fans and may not be interested in other sports.
Some of those may have already lost interest in skating because they feel the emphasis has become too technical and the programs are less artistic than their favorite era.
So putting even more emphasis on technical details in ways that discourage skaters to focus on performing coherent programs might lose even more of those fans, or at least not get back the ones who have been lost.
Will it attract more sports fans, e.g., X-Games fans, who have not been much interested in figure skating before?
For people who are currently already skating fans, what will make them even more engaged? Golden Skate provides a cross section of avid fans, older and newer/younger, tech vs. arts focused, etc. -- not sure how representative it is of skating fandom in general or the English-speaking contingent thereof.
What do Golden Skate fans think? Will we lose interest in skating if technology to make them "part of the action" doesn't get adopted ASAP?
Some of those may have already lost interest in skating because they feel the emphasis has become too technical and the programs are less artistic than their favorite era.
So putting even more emphasis on technical details in ways that discourage skaters to focus on performing coherent programs might lose even more of those fans, or at least not get back the ones who have been lost.
Will it attract more sports fans, e.g., X-Games fans, who have not been much interested in figure skating before?
For people who are currently already skating fans, what will make them even more engaged? Golden Skate provides a cross section of avid fans, older and newer/younger, tech vs. arts focused, etc. -- not sure how representative it is of skating fandom in general or the English-speaking contingent thereof.
What do Golden Skate fans think? Will we lose interest in skating if technology to make them "part of the action" doesn't get adopted ASAP?
Nah, I don't think the lack of tech will scare the fans.
They seem more upset with judging issue and thisis the reason to bring the technology in, not the technology itself.
I believe Nigel is wrong about reality - new technologies are only worth it if they bring some improvements to the sport, which is not the case in FS.
They seem more upset with judging issue and thisis the reason to bring the technology in, not the technology itself.
I believe Nigel is wrong about reality - new technologies are only worth it if they bring some improvements to the sport, which is not the case in FS.
- Joined
- Jan 13, 2014
IMO as a male and relatively new to figure skating, I would love to see more hard physical stats. Tired of people thinking that figure skating "looks so easy, its just some jumps and spins!". Instead, what if they could measure rotation speed, or average speed in a program etc. (Wow, xyz skater has an average speed of 20mph, while the other skater averaged 15 mph).
Just an article I ran into, if they can measure forces, I'm pretty sure they can measure speed.
http://gizmodo.com/sensors-show-figure-skaters-absorb-8x-their-own-body-we-1526842854
Just an article I ran into, if they can measure forces, I'm pretty sure they can measure speed.
http://gizmodo.com/sensors-show-figure-skaters-absorb-8x-their-own-body-we-1526842854
- Joined
- Feb 25, 2014
@jinhamasaki, then you will like the Data Man documentary. Sharing the below from Yuzu's thread.
Last edited:
