How do you penalize falls?

Okay. The GP medalists have had something like 45 falls between them. Patrick Chan fell four times each en route to a gold and silver medal, over skaters who fell once at most (SC) or twice (CoR). There can be no argument about the fact that the high, medal-winning scores for skaters that have fallen is confusing the audience and doesn’t make for a great sporting event (and is only entertaining for those with a taste for schadenfraude). So, how do you penalize falls so that

a) Technical risk is rewarded, even when imperfect (or should it be rewarded when imperfect)?
b) Get the podiums that reward clean skating
c) Allow for program balance to be rewarded.

I’ll toss out a few ideas I’ve read here and there. I’ll include a podium analysis as well (if these rules were applied to COR, what would the scores/podiums be like, as CoR seems to be demonized for the rules, not corruption)

Reduce the element to 0
PRO: It is such a basic error to fall on any element that the element shouldn’t contribute to the overall total. We shouldn’t care whether or not someone includes an element if they don’t know if they can land it.
CON: This, by definition, will limit risk. Additionally, it argues that a fall on a higher risk element is worth a bigger penalty than a fall on a lower risk element. Forget about that: it argues that a fall on quadruple axel is worse technically than landing a single loop. With everything scrutinized within a point of its life, how many skaters will include a harder element if the risk is so great? This doesn’t penalize falls in between elements, which have happened multiple times this season (Sui/Han, Crone/Poirier, Davis/White). And what about a fall that happene d to Takahashi in his SP at Finlandia?. It’s also unfair in theory: a basic error in an element is one that doesn’t follow the definition of an element, which is not just a fall, but an edge call, a UR, etc.

COR scores/podium(does the program deduction still count?)

1. Tomas Verner: 230.31 (no falls)
2. Jeremy Abbott: 210.49 (or 212.49)

• SP Score: 77.61 (no falls)
• LP Score: 132.88 or 134.88 (2 falls)

3. Patrick Chan: 207.98 (or 211.98)

• SP Score: 76.46 OR 77.46 (1 fall)
• LP Score: 131.52 OR 134.52 (3 falls)

4. Samuel Contesti: 207.30 (no falls)

So Abbott beats Chan as long as Chan still does a fourth combo. Assuming he doesn’t and lands the 2A with 0 GOE, he takes silver.

Have a Sliding Scale for Falls: Each subsequent fall being more deleterious to the total than the last; same rules for GOE application
PRO: It recognizes that a program with one fall can still be technically/artistically superior to a clean skate and doesn’t punish the skater for going for a single high risk element and failing. It also punishes falls that happen in between the elements. But if you have three or four elements beyond your grasp that you include with the hope of making them or benefiting from the points a failed element gives you, no matter what your skills are, your program is a mess.
MIXED: It takes more out of the judge’s hands. People have complained about the technical panel having too much power. How do they feel about the judges having even less power here?
CON: This is a system that needs to apply to all skaters, whether novice, junior, senior etc. While a sliding scale makes sense for the senior skaters, I have to admit I’m not particularly enthused about the idea of a junior skater giving a program and then getting a negative score for it, regardless. A quick scan of the early JGPs suggests I’m over-reacting a little bit (the lowest score would be something ~4 points) But that also depends on just how big the deduction for the subsequent falls would be.

1st Fall: -1
2nd Fall: I’ve heard -2, -3 and -5.
3rd Fall: -3, -7 and -15
4th Fall: ????
5th Fall: ????

Now, a three fall short program would be affected by less than 10% according to the first method, 15% the second, and 29% for the third structure (pegging an SP to 72 points total. Arbitrarily). How much should a program suffer (and half that for a lp)? And what about when we get into the 4th and 5th fall?

1. Tomas Verner: 230.31 (no falls)
2. Patrick Chan: 222.21 (first system, extra -5 deduction)
3. Patrick Chan: 217.21 (second system, extra -8 deduction)
4. Jeremy Abbott: 216.21 (first system, extra -1 deduction)
5. Jeremy Abbott: 215.21: (second system, extra -2 deduction)
6. Jeremy Abbott 210.21 (third system, extra -4 deduction)
7. Patrick Chan: 210.21 (third system, extra -18)
8. Samuel Contesti: 207.30 (no falls)

Make a more Explicit Connection to PCS
PCS have the following instructions on specific components as to criteria of grading.

Transitions:
Quality

Performance:
In all skating disciplines each skater must be physically committed, sincere in emotion,
and equal in comprehension of the music and in execution of all movement.
Clarity is characterized by the refined lines of the body and limbs, as well as the precise
execution of any movement

PRO: To me, these are criteria that, as written, are affected by falls. But arguably, don’t falls demonstrate poor skating skills and interpretation as well (potentially). More specifically, your perception of a program on the whole IS affected by visible errors more so than by invisible ones – regardless of level of expertise. So it almost makes sense to have falls on element affect PCS to the same extent that we see it affecting TES score (~4 points per fall)
CON: Doesn’t this link PCS closer and closer to the elements? One of goals of the IJS/COP is to have a way of judging the non-element construction of a program. Closely linking the PCS to the elements has the effect of essentially weighing the TES higher.
MIXED: How else do you have a system that privileges SKATE above SKATER and PROGRAM.

Have a higher penalty in terms of GOEs for Falls

Currently, there are a handful of circumstances where a -3 GOE is mandatory
a) No combo jump in the SP
b) Less revolutions than required in SP (so if you do a 3-1 combo, you get a -3)
c) The dismount from a lift is done on two feet.

PRO: Having a specifically higher penalty on elements with falls (say -4) has the consequence that a fall doesn’t negate the element in its entirety but unless everything else is really high quality, you can’t make it up. Additionally a specific fall –GOE allows judges to more easily judge elements with multiple errors without falls. For example, Mroz scored a -3 from one of the CoC judges for his spin. But he didn’t fall, so it does seem that by definition, his spin should be better than if he had fallen on it. A triple axel is worth 8.5 fully rotated. 3A with a fall is 5.5, 4.5 with the program deduction (which doesn’t apply to the jump, so the jump still contributed 5.5). This would mean that the triple axel would now contribute 4.5 (3.5 w/ deduction), which truthfully, still seems rather high.
CON: The element still contributes something, which for some people is definitely a con.

COR Scores

1. Tomas Verner: 230.31 (no falls)
2. Patrick Chan: 223.51

• SP Score: 80.96 (1 fall)
• LP Score: 142.55 (3 falls)

3. Jeremy Abbott: 215.51

• SP Score: 77.61 (no falls)
• LP Score: 137.90 (2 falls)

4. Samuel Contesti: 207.30 (no falls)

None of the above solutions. Negative GOE should reduce the base value of the element to a certain set percentage, after it's already scaled for any downgrades or being a sequence. Maybe to start out -1 GOE would be worth 90% of base value, -2 would be 80%, -3 40%. That would make negative scores very unlikely. Value %s could be tweaked as necessary, and don't need to be the same across all elements, and the steps between levels of negative GOE shouldn't be uniform anyway. A shaky landing on an otherwise clean jump should be worth a lot more than a complete wipe-out, but that's not the case under the current system.

Or we could go back to a system that gives the judges the benefit of the doubt that they know how to do their job, which is rank skating performances, instead of micro-managing them. At the same time, get rid of the anonymity and post everyone's scores with names clearly attached, so if there's abuse of the system it's out in the open and can be challenged. Trust but verify. There's never going to be a system that works perfectly to value exactly needs to be valued but penalizes only what needs to be penalized. It's a subjective sport, and it needs to go back to a system of judging that acknowledges that subjectivity, instead of trying to legislate it out of existence every single season.

Love this thread!:thumbsup:

Going back to study your post details.

Go with the same methods that are used in Ice Dance (ISU 1567) p. 13:

They dock the PCS scores for falls and stumbles. For singles, you might want to eliminate the part about stumbles?

Skating Skills
Outside of Required Elements:
- Skating with hand(s) on ice at any
time including during introduction
and/or conclusion
by one - 0.5 per each
by both - 1.0 per each
- Loss of balance or Stumble:
by one - 0.5 per each
by both - 1.0 per each

Click to expand...

Additionallly

Technical panel takes automatic deduction from total score: - 1.0 for every fall of one and - 2.0 for every fall by both partners;
If the fall causes interruptions to the program that exceed 5 seconds and part of the program was missed, the Referee additionally applies the
following deductions: - 1. 0 for 6-15 seconds interruption, - 2.0 for 16-30 seconds interruption etc.
In addition with falls – judge’s scores in some or all Components may also need to be reduced as well if a fall affects the rest of the program or part of the program.

Click to expand...

That way, the tech elements are correctly represented, as they are today, by negative GOE and the -1 penalty per fall, but the interruption to the program, whether in elements or transitional moves, is explicitly reflected in the GOE.

doubleflutz said:

A shaky landing on an otherwise clean jump should be worth a lot more than a complete wipe-out, but that's not the case under the current system.

Click to expand...

How is that not the case?

A shaky landing on an otherwise clean and rotated jump might earn -1 GOE.

A complete wipeout on a jump with the same takeoff and same amount of rotation should earn -3 GOE and a fall deduction.

Now, if the shaky landed jump was downgraded and the wipeout jump was not, that might cancel out the difference. Similarly if the wipeout jump was significantly more difficult to begin with.

But even under the old system judges always had the discretion to reward attempts at difficult jumps even if they ended in falls more than shakily landed easier or badly cheated jump.

Now it's just spelled out exactly how much each element is worth based on takeoff and rotation and judges' average evaluation of success/quality.

Personally, in answer to the original question, I'd like to see the negative GOEs for difficult jumps be more like what they were last year, so there's less positive value left over after -3 and fall deduction.

I also wouldn't mind a mild increase in deduction penalty for subsequent falls, maybe -1 for the first fall, -2 for the second, -3 for the third. These deductions would need to be smaller at lower levels. Already I think -1 is too large a deduction below novice level.

And encourage judges to penalize for disruptive errors (not only those that qualify for the fall deduction) in PCS, especially Performance/Execution, maybe write something explicitly into the P/E criteria to encourage penalizing more than one such error. But don't require specific reductions in the PCS or specific caps on component scores for programs with falls. Not all falls are equally disruptive, and some programs start out at such a high level that even after reductions for falls they still deserve to score higher than other skaters' programs with no falls.

I don't see the need to make it complicated. Rather than say a skater is "penalized" for such-and-such. Let's say a skater gets credit for ....

1) full rotation
2) one foot landing
3) no hand-down
4) no fall
5) smooth flow in and out

If something in this list does not happen, less credit. If a fall interferes with the choreography, then lower scores result.

Thank you Imaginary Pogue for the work you put into that topic!

Can you imagine a skater being announced for a Gala as the one who fell 5 times and won gold?

I believe the standard penalty is a -1 for the overall score of a contestant for each Fall. So if a skater is granted highest scores for Skating Skills in the PC portion of his total score, it may well negate the 5points for Falling. Apparently Falls do not show poor skating skills during that competition.

Falls definitely show an inability to complete an element, be it jumps, spins, footwork, in Singles. The Tech is marred. No?
Falls also show a disruption the Program when there is a loss of musicality (interpretation, if you wish), skating ability, Flow and Time to recoup. The PC is marred. No?

However, there is the penalty of -1 for each Fall. Is it sufficient given the inability to complete an element and the resulting performance of the above. The UR has a much harsher Penalty for not completing an element, and in most cases does not disrupt the program.

Falls in Pairs and Ice Dance are treated much harsher in PC for disrupting the program.

The ISU should set up a Working Committee to Review the Penalties.

The official penalty is -1 point, but we all know in practice it is much more.

The only probem, in terms of the system with regards to this topic, is that -GOE needs to be greater.

gkelly said:

How is that not the case?

Click to expand...

The way negative GOE works. They take off flat amounts, not a percentage, and the scale between levels of GOE on most elements is linear. Right now, on a 4T a -3 is not harsh enough, but depending on what -1 means, it's maybe even too harsh. There's not enough room between levels of GOE, or rather, they aren't scaled in a way that relates properly to the severity of the mistakes.

The penalty for -1 3A is -1, -2 is -2, -3 is -3, all off of a base value of 8.6. That means a completely botched triple axel is worth roughly the same as a good triple lutz with a shaky landing. That is stupid. Actually, an even better example of the absurdity: a completely botched 4A with -3 across the board (10.4 counting fall deduction) is worth almost as much as a 4S without any negative GOE (10.5), presuming only that both are fully rotated. That is just moronic. There's YouTube video of Igor Macypura getting all the way around on 4A before he wipes out, he should put that in his SP, think of the points! Or maybe Kevin Reynolds should give up trying for clean 3A and just work on rotating 4A, it would actually give him better points if he could do it.

The system doesn't do enough to differentiate between "accomplished, but flawed" and "not accomplished at all".

dorispulaski said:

Go with the same methods that are used in Ice Dance (ISU 1567) p. 13:

They dock the PCS scores for falls and stumbles. For singles, you might want to eliminate the part about stumbles?

Click to expand...

Why? It's a lesser penalty than for falls, but it's still a disruption.

Is stepping out of five jumps in a long program with two clean landings less disruptive than falling on two jumps and landing five cleanly?

I think the ice dance penalties are too large for singles and pairs, which include a lot more risky elements that are likely to involve significant losses of balance (with or without complete falls).

So I could see suggesting guidelines for judges to reduce PCS in the case of breaks in the program, falls, or significant or multiple stumbles. Maybe recommend 0.25 off Performance/Execution for each fall or moderate-to-severe stumble after the first, 0.5 off P/E and/or Choreography for each break of 5 seconds or more to recover from a mistake.*

*So if a skater falls but doesn't break the flow of the performance or break character, the judges are welcome not to penalize.

Of course the judges aren't timing the breaks -- the referee/timekeeper take deductions for significant breaks.

Maybe recommend 0.25 off Skating Skills for each two moderate or one severe stumbles that occur during step or spiral sequences or between elements.

Some judges are probably doing something along those lines already. Would writing them into the guidelines make it more consistent without taking away the judges' discretion to exercise their judgment?

(Keep in mind that in long programs the PCS are doubled. Does that mean that the penalties should be larger in short programs?)

doubleflutz said:

That means a completely botched triple axel is worth roughly the same as a good triple lutz with a shaky landing.

Click to expand...

So you're comparing different jumps, which was not specified in your original post.

Yes, I agree that the -GOE penalties on the harder jumps should be larger than they are this year.

But don't we still want a fall on a rotated very difficult jump (say 4T) to be worth more than a shaky landing on a very easy jump (say 1T)?

Now, when it comes to comparing very difficult jumps (triple axels and quads) with moderately difficult jumps (double axels and triples), then we need to fine tune the specifics to balance out the risk of attempting something only a handful of skaters in the world can even attempt vs. the aesthetic value of a clean landing.

Maybe the -GOEs and fall deductions should be calculated so that falling on a rotated difficult jump will always be worth the same as a -1 GOE shaky or 0 GOE just adequate landing on a jump from the same takeoff with one less rotation, or some system with similar principle.

Most people aren't taking into account that there is much more to a jump that the landing. There is the speed, entrance, height, rotation, ice coverage then landing. So if everything is perfect but on the landing a ankle rolls and the skater falls, the skater recives no credit for the entire jump? That is really harsh.

Here are some examples:

At 0:28 Thomas who messes up the landing vs. Mao (0:36) who completly misses a jump.

One should count for something and the other one not at all. More examples of a jump that shouldn't count is at 1:11

Skater falls

The same thing could be said for the twist in pairs. Just because it's a crashy catch mean the judges should not count the difficult entrance, split, hight, rotation and landing on one foot by both partners just because the catch was less than ideal?

here is another example of a triple thow where she fell but the throw was beautiful, should the whole throw not have counted?

2:09

gkelly said:

So you're comparing different jumps, which was not specified in your original post.

Click to expand...

No, it applies within the same jumps, too. There isn't a big enough gap between a -1 3Lz (5.3) and a -3 3Lz (3.9). The stupidity is just easier to illustrate with cross-jump examples.

But don't we still want a fall on a rotated very difficult jump (say 4T) to be worth more than a shaky landing on a very easy jump (say 1T)?

Click to expand...

With reasonable base values, that's not going to be a problem. A shaky landing on a 1T is worth a whopping 0.1 - a tenth of a point. Even something as small as 10% of current 4T value (10.3) is worth quite a bit more than that.

Now, when it comes to comparing very difficult jumps (triple axels and quads) with moderately difficult jumps (double axels and triples), then we need to fine tune the specifics to balance out the risk of attempting something only a handful of skaters in the world can even attempt vs. the aesthetic value of a clean landing.

Click to expand...

It's not the aesthetic value of a clean landing, it's the athletic value. If a skater doesn't at least land on her blades on something vaguely resembling an edge, even if she two-foots it and generates a ton of snow with a scratchy landing that goes nowhere, in terms of skating she has not really done diddly-squat. The point of jumping in skating from an athletic perspective is not the acrobatics of the rotations in the air, it's that doing more rotations (and bigger jumps) makes the skating into and out of the jump harder. Same principle by which smooth, flowing landings are worth more than weak, scratchy ones. It's not aesthetics, it's technical performance. If there's no skating out of the jump, because butt has met ice, it has failed as a technical element.

here is another example of a triple thow where she fell but the throw was beautiful, should the whole throw not have counted?

2:09

Click to expand...

Actually that was a quad throw.

Which only enhances your point -- should an envelope-pushing element that actually had a one-foot landing on an edge with speed and flow for several feet before the fall really end up counting for no more than if the skaters had just skipped the element and continued the program without attempting it? That's what the 0 points for falls approach would result in.

mousepotato said:

Most people aren't taking into account that there is much more to a jump that the landing. There is the speed, entrance, height, rotation, ice coverage then landing. So if everything is perfect but on the landing a ankle rolls and the skater falls, the skater recives no credit for the entire jump? That is really harsh.

Click to expand...

Actually, if the system was tweaked to be a bit more like 6.0, that could be less harsh. If a skater made a mistake that completely voids the jump as an element, they should take a hit on PCS depending on how disruptive it was and the mandatory fall deduction, but they would still have that jumping pass free and would be able to add it in again and get full credit if they land it on the second try, if they're quick-thinking enough to do so and still have the stamina. I liked that was an option under 6.0.

Blades of Passion said:

The only probem, in terms of the system with regards to this topic, is that -GOE needs to be greater.

Click to expand...

This works for me. I see a fall is a more complete failure of an element then other mistakes - ur, wrong edge, two footed landing. I went back and looked at the score sheets for TEB, if a skater fell on a jump they recieved -3 GOEs across the board - thats fine - but then -3s are given on other, less completely failed jumps. It just doesn't seem right. A -4 GOE would give the judges the ability to more clearly distinguish between an element not completed because of a fall versus a seriously flawed element.

BTW count me as one who hasn't quite figured out when -2 GOEs really loses the skater 2 points and when it doesn't

Glad to see the Forum is not ok with pampering the Falls and other elements.

If in the unlikely event the ISU takes up a review of the penalties, they should also take up the business of partial credits. The rotations in a Flutz do not carry the counter rotation of a Lutz. There just was no counter rotation jump. Therefore there was no Lutz. It should get a 0 score for that jump. Most skaters do have a genuine Lutz. They worked hard to get that jump. So can the flutzers.

I would rather not get too carried away with this. Falls are ugly and should be deducted, but I have never felt they should negate everything else a skater does correctly.

I think - you fall, no points. Is that harsh? But I still don't understand all the nuances of CoP...