When the Mighty Have Fallen

Patrick Chan's first place result in his Grand Prix Final short program, after a spectacular fall on the required jump combination, has again brought to the fore the question of whether IJS rewards failure by allowing too many points on an element (particularly a jump) with a fall.  This has been the subject of debate since the inception of IJS with critics saying IJS has made the penalty for a fall just a slap on the wrist compared to the 6.0 system, while the IJS gods proclaim that harsher penalties would stifle the skaters' motivation to attempt the difficult triple and quadruple jumps. While we have generally held the rewarding-failure-is-a-bad-thing point of view, the most recent discussion of this entertaining subject got us wondering again, exactly how does the penalty for a fall today really compare with the "before time?"  Before IJS, that is. Falls Under the 6.0 System Under 6.0, falls on elements in the short program were serious business.  In the jump combination, for example, a fall on the second jump was a 0.3 point deduction (with no cheat on the rotation), compared with a 0.4 deduction for a fall on the first jump and 0.5 for omitting the element entirely.  So for Chan's combination, one might say in one sense he would have kept 0.2 points out of 0.5 points, or 40%, for the element under 6.0. However, this is not strictly correct since the elements under 6.0 didn't really have a specific point value, and the judges would also account for the difficulty in what was attempted through a base value for the first mark if the program had been executed without errors.  More strictly correct, one would say that a fall on the second jump resulted in 60% of the maximum possible penalty in the execution of the element.  But even that does not tell us much.  The bottom line really is how much would the fall impact the the placement of the skater. Under 6.0 judging, the judges typically separated each place by 0.1 point or less in total mark for a high level or championship competition, resulting in a typical drop in placement of 2-4 places for a 0.3 deduction.  This of course would also be affected by how high a base mark the judges gave the performance and whether they decided to "hold the skater up" in the second mark.  But in general, a fall in the short program resulted in the skater dropping a few places, which is why it was all important in the short program to skate clean.  If an error dropped a skater out of the top three places, it was impossible for the skater to win the event by winning the free skate without the "help" of other skaters. In the free skate the situation was different.  There was no required penalty for a fall (except for a few seasons when the ISU played around with that) and the rules specifically stated that a fall was no bar to winning.  If a skater fell on a jump it got, in effect, no credit.  Attempting the failed jump did not help the skater, but it also did not cause a specific reduction in score.  If the remaining content was sufficiently impressive the skater could still receive high marks and still win.  As long as the skater did not wallow around on the ice, but got quickly back into the program, it might not even affect the second mark.  This idea that a jump with a fall got no credit in the free skate but receives partial credit (points) under IJS rubs some people the wrong way. But it isn't that simple. While a jump with a fall got no credit, it also did not necessarily prevent the skater from winning in a 6.0 free skate.  Of course, if failing to execute the jump caused a skater to have less total content that another, the skater would end up penalized, but they were not crucified. We estimate one quick fall on a jump element typically might have cost a skater up to 2 places at the most.  This is because under 6.0 the contribution each element made to the judges' placements was not as sensitive to the difficulty of the elements as it is under IJS.  For example, if one skater landed seven triples and quads and another landed only six due to a fall, that difference of one jump could potentially be made up in spins, steps and/or presentation, even for a failed quad. Falls Under IJS Falls under IJS are handled the same way in both the short program and the free skate.  There is a reduction in the GoE of 3, and there is a 1.0 point deduction for the fall.  In the case of Chan's jump combination the base value of the combination was 14.40 and he lost 3.29 points on the GoE reduction plus the fall deduction.  As a result he kept 84% of the value of the jump compared to a clean element with 0 GoEs.  He failed the element (did not complete the required combination) and kept 84% of the points vs. 40% (sort of) for the 6.0 system.  Which seems like a major reduction in consequences for a fall; however, this may not the best comparison to make,  The more important question, in our view, is how many places do the points lost on a failed jump element correspond to; and also, how difficult is it to overcome the loss of points on a fall and still win the segment.  Under IJS, the average point difference between places in an elite or championship competition is typically about 2-4 points.  This varies somewhat from first to last in a competition, and in some cases the point difference between places can be small, or even zero.  Other times it can be huge  But 2-4 points is a good rule of thumb for making some simple comparisons. Under IJS, a fall on a triple or quad jump typically results in a loss of up to three points in GoE and one point for the fall deduction, for a total penalty of up to four points.  Using the 2-4 points per place rule of thumb, then, a fall typically (averaged over many skaters and competitions) can cost a skater up to 2 places, (and that's typically, for comparison purposes, not a prediction for any specific skater or competition) which is the same net effect of a fall in the free skate under 6.0, but about half what it was in the short program under 6.0.  Further, with a loss of 4 points on the element, the skater has a chance of making up for the loss from points earned on spins, steps and components and still win, just as was the case under 6.0. But this is only true because the jump attempt has gotten some partial credit.  If the skater lost all points on the attempt, for a high value triple or quad they would be much worse off than for the free skate under 6.0.  With a loss of all points on a high value triple or quad the error would typically cost the skater 2-4 places, and making up the lost 8.5 to 13.6 points would be extraordinarily difficult since jumps make up the vast majority of the points scored. It would certainly be logically more satisfying if a fall on a jump resulted in no points (since the jump is failed), so long as the skater still had a chance to win through strengths elsewhere in the program, as was the case under 6.0.  But the current structure of the IJS points system does not allow that.  By using a point model where just one element can be up to 8% of the free skate score, failure of one such element makes it nearly impossible for a skater to fail that element and still win due to other strengths.  The less than satisfying approach IJS has taken is to give partial credit for the jump, so that the skater is both penalized for the error, but not so much so that they do not have a chance to win if they have strengths elsewhere in the program. Bottom line:  Under IJS, the penalty for a fall, and the difficulty of overcoming that fall in the short program and win the segment, is about half what it was under 6.0.  The penalty for a fall, and the difficulty of overcoming that fall in the free skate and win the segment, is about the same as it was under 6.0.  Giving a jump with a fall no points would double the effective penalty for the fall and at least double the current difficulty of overcoming the fall and still win the segment under IJS. Which brings to mind the following thought.  If a jump with a fall received zero points in the short program the impact on the results would typically be the same as it was under 6.0.  This modest change would bring the penalty for a fall in the short program under IJS (in term of places and difficulty of overcoming the error) back in line with what it was under 6.0.  It would also return the short program to its original purpose, to have the skaters demonstrate they can cleanly execute certain required elements, and not be just the shortened free skate of no other distinction that it has become.  Loose Ends The above discussion relates specifically to high value triples and quads in Junior and Senior events.  It does not apply to singles and doubles in Juvenile through Novice events, where the penalty for a jump with a fall is a net loss of points.  This aspect of IJS offends us far more that whether Junior and Senior skaters are getting too much partial credit on a fall.  It is obscene, in our opinion, that if a skater falls on a single or double they not only get no points, they lose some of the points they earned on other elements.  No other sport does this.  It is like a touchdown being called back on a penalty and the team not only loses the potential points for the touchdown, the lose the points for a previously scored touchdown.  Just nuts! Revised 31 December 2011