Is it easier to cheat in 6.0 or CoP?

[prettykeys]

By the way, even under OBO, if there were more than two contestants, a skater who was judged much better by 4 judges might well finish ahead of a skater who was judged just a little bit better by five judges. This could happen if a thrid skater was placed between the two by the four-judge minority.

What's OBO?

That's definitely a huge problem.

 

[Daniel5555]

Mathman
Some notes on the old system...
I think we should talk about how much points they receive exactly...

I just generated some random data:
Points for Ando from fair judges:
5.7 5.7 5.7 5.7 5.7 5.8

And she really deserves 5.7
Points for Kiira:
5.6 5.6 5.7 5.5 5.6 5.7
And she really deserves 5.6

Let's assume that 3 conspirators are going to give Ando 5,6 and to Kiira 5,7 (that's for them not to be noticed).

So the mean for Ando is 5,6777(7) and for Kiira is 5,6444(4). To say the truth, that's a huge difference. In order to equal their points those 3 guys will have to give her 5,8 each and that won't go unnoticed. For her to win someone will have to give her 5,9.
That system seems to be pretty robust.

Now we should try to do the same for CoP, but it's much more difficult to simulate.
But you definitely have some good points. The major problem is that random draw. It may actually happen that some judges who gave fair, but a bit high scores, are eliminated and there remain only those with fair, but moderate scores and ones with unfair and low. That will not only affect points, it will affect the standing. And it will be impossible to notice. But they will need that 41% of luck.

 

[prettykeys]

I still like the general concept of CoP, I just think we need a smaller PCS section and a more balanced way of coming up with the Final Score.

Maybe combine Skating Skills + Transitions, Choreo + Interpretation, and Performance so we have three 10.0 PCS scales; add and then convert them into a factor (30% of total score?)

Take the TES as we have it, add a bonus for doing a complete set of triples, be a little more lax on the underrotation penalties, convert it into a factor (70%?)

Then add the factored TES and PCS together to get the Final Score.

The way CoP is set up right now is that manipulated values in especially the PCS can artificially prop up or or pull down a skater.

ETA: Continue trimming the means, and no more anonymous judging either.

 

[prettykeys]

By the way, I think this discussion shows that you can't "mathematically prove" that CoP is more "flawed" than 6.0.

There are no mathematically objective ways of saying which placements are correct.

There is no mathematically objective way of saying this kind of performance or element deserves X points.

Figure Skating judging is an art, not a science.

So someone needs to edit out that Wikipedia entry.

 

[Daniel5555]

prettykeys

By the way, I think this discussion shows that you can't "mathematically prove" that CoP is more "flawed" than 6.0.

Simulate the data, add static points for the unfair judges, get the result for 3 different cases - that will be the proof. Want the real proof - do it 10000 times and look at statistics.

But I think the point of that random draw is that even if you pay the money, there is no guarantee of the result.
In the old system if you buy 3 judges, you can't win, but at least you get the scores. Here if you do the same there are 58% of chance that you won't get what you want and you will have no your money back.
So it's more the psychological barrier than any real protection. Still it may work. No one wants to pay for nothing.

 

[prettykeys]

The scenario is that we're assuming Miki should win, and under 6.0, a mostly honest panel of judges would definitely give her the win when there are 3/9 judges who are dishonest.

I have yet to see evidence that a CoP panel would get it "wrong".

 

[prettykeys]

Simulate the data, add static points for the unfair judges, get the result for 3 different cases - that will be the proof. Want the real proof - do it 10000 times and look at statistics.

I can't see how this is possible with a lot more assumptions and variables such as "how much should/would the honest judges give Miki in terms of point-advantage over Kiira"?

For that matter, as in yours and Mathman's examples, if we are to likewise assume that the honest judges are giving Miki +1 points over Kiira, then we're still going to get the same result. (At least) 6 good judges trump (at most) 3 bad ones.

 

[Daniel5555]

prettykeys

I have yet to see evidence that a CoP panel would get it "wrong".

They wouldn't get it wrong, because in the worst case it would be 2 vs 3 judges and Miki would win. But the problem is that the fair judges are not robots and can have personal preferences and make errors. That's the reason why I generated random data - because for the fair judges it would be random (it's a normal distribution with 0.1 deviation) and for unfair static.
So if there is bad luck, Miki would lose.

I can't see how this is possible with a lot more assumptions and variables such as "how much should/would the honest judges give Miki in terms of point-advantage over Kiira"?

Already answered.

 

[prettykeys]

They wouldn't get it wrong, because in the worst case it would be 2 vs 3 judges and Miki would win. But the problem is that the fair judges are not robots and can have personal preferences and make errors. That's the reason why I generated random data - because for the fair judges it would be random (it's a normal distribution with 0.1 deviation) and for unfair static.
So if there is bad luck, Miki would lose.

But that's how it is in 6.0, too. In a close competition, aren't we idealizing the notion that Miki will always be given the win? We're thinking of the perfect, ideal judges. And STILL, you cannot quantify what the difference will be, or should be.

Already answered.

Oh yeah? Quote yourself, because I don't see the answer.

 

[Mathman]

Mathman
Some notes on the old system...
I think we should talk about how much points they receive exactly...

I just generated some random data:
Points for Ando from fair judges:
5.7 5.7 5.7 5.7 5.7 5.8

And she really deserves 5.7
Points for Kiira:
5.6 5.6 5.7 5.5 5.6 5.7
And she really deserves 5
Let's assume that 3 conspirators are going to give Ando 5,6 and to Kiira 5,7 (that's for them not to be noticed).

So the mean for Ando is 5,6777(7) and for Kiira is 5,6444(4). To say the truth, that's a huge difference. In order to equal their points those 3 guys will have to give her 5,8 each and that won't go unnoticed. For her to win someone will have to give her 5,9.

That system seems to be pretty robust.

It is even more robust than your analysis shows. Under ordinal judging there is no averaging of the marks. Only the ordinal placements count.

In your example, Ando has first place ordinals from judges #1, 2, 4, 5 and 6. No matter what the three conspirators do -- even if they give Korpi 6.0 and Ando 0 -- Ando wins with a majority of first place ordinals.

By the way, I think this discussion shows that you can't "mathematically prove" that CoP is more "flawed" than 6.0.

Worse than that. You can mathematically prove (in the topic of "social choice theory") that there does not exist any voting system that always satifies a few mild and desireble conditions, such as the condition of independence of irrelevant alternatives: "if every voter prefers X to Y then the entry of Z into the contest will not affect the voters preference of X over Y."

The most famous of these theorems is Arrow's Impossibility Theorem, after economist Kenneth Arrow.

 

[janetfan]

What if one was to consider Occam's razor?

"The simplest explanation is usually the best."

Would it apply to 6.0 vs CoP ?

Could this also mean "the simplest system is usually the best."

Is 6.0 a simpler system than CoP?

And what of Newton's take on Occam's razor:

"We are to admit no more causes of natural things than such as are both true and sufficient to explain their appearances."

If this is applicable then Occam's razor seems to suggest that Cop with all of it's extra means and categories of marking would allow more room for error or for cheating.

Yes or no?

 

[prettykeys]

It would be interesting (and mathematically possible), to quantify how large the degree of cheating has to be to overcome modest variations in honest judging that can happen in two skaters considered reasonably close in ability.

Something like the probability that three cheating judges can establish their favourite as the winner if they cheat by X amount when the honest judges ideally give the better skater a higher score by Y.

 

[Daniel5555]

prettykeys

But that's how it is in 6.0, too. In a close competition, aren't we idealizing the notion that Miki will always be given the win? We're thinking of the perfect, ideal judges. And STILL, you cannot quantify what the difference will be, or should be.

That's why there is a simulation.
If both Miki and Kiira are equal in skating and ALL judges fair, then the result would be random with equal probability for them to win. At least statistically after making a lot of simulations there would no clear winner.

If Miki is slightly better than Kiira and ALL judges fair, then the result still would be random, but after a huge number of simulations Miki would be clear winner.

The question here is that if Miki is slightly better than Kiira, would a certain number of unfair judges definitely make her a clear winner after a huge number of simulations (or the same, would her chances to win be higher than Miki's, even you can't see it clearly).

And it's not the same as with 6.0 system for sure.

Oh yeah? Quote yourself, because I don't see the answer.

What's going on with you? No offence, but lately you are strange.
My post number 48 is all about that.

 

[Mathman]

I have yet to see evidence that a CoP panel would get it "wrong".

Easy as pie.

After the draw there are two conspirators (judges 6 and 7) left and 5 honest judges. In choreography the marks go

Miki: 7.00 7.00 7.00 7.00 7.00 6.00 6.00

Kiira: 6.75 6.75 6.75 6.75 6.75 7.75 7.75

Throw out highest and lowest.

Miki: 7.00 7.00 7.00 7.00 6.00. Average 6.80

Kiira: 6.75 6.75 6.75 6.75 7.75. Average 6.95

Kiira wins even though Miki is favored by four of the five surviving judges. (Notice that judge number eight did have influence on the success of the conspiracy. By throwing himself on the sword, he made sure that judge number seven would survive to carry the day.)

Now multiply this by all the PCSs and all the GOEs and you can see how Kiira could easily come out ahead overall, even overcoming a deficit in TES.

Edited to add You (prettykeys) are right, though. In my original example i did not take into account the fact that Miki started out ahead, so the pro-Kiira copnspirators would have a deficit to make up first before they surged into the lead.

 

[Daniel5555]

Mathman

It is even more robust than your analysis shows. Under ordinal judging there is no averaging of the marks. Only the ordinal placements count.

In your example, Ando has first place ordinals from judges #1, 2, 4, 5 and 6. No matter what the three conspirators do -- even if they give Korpi 6.0 and Ando 0 -- Ando wins with a majority of first place ordinals.

Pff, forgot about it.
But that is not an advantage, really. It makes it almost impossible to cheat, but the marks are not showing much thing really. With the 6.0 mark you already make someone a winner. So the favoritism really matters in that system under some circumstances.
That system is too rough.

janetfan

What if one was to consider Occam's razor?

"The simplest explanation is usually the best."

Would it apply to 6.0 vs CoP ?

Could this also mean "the simplest system is usually the best."

Not at all.
CoP is much better.

If this is applicable then Occam's razor seems to suggest that Cop with all of it's extra means and categories of judging would allow more room for cheating.

It would allow more room for cheating, but it has more advantages anyway.

prettykeys

You're still not able to tell me how a "close" competition between Kiira and Miki automatically convert to a number of points that would, sometimes, in error, give Kiira the win.

I need a simulation for that, but it's pretty obvious with everything that Mathman said.

What is "strange" about me "lately"??

You seem to me more rude than usually.

 

[prettykeys]

Easy as pie.

After the draw there are two conspirators (judges 6 and 7) left and 5 honest judges. In choreography the marks go

Miki: 7.00 7.00 7.00 7.00 7.00 6.00 6.00

Kiira: 6.75 6.75 6.75 6.75 6.75 7.75 7.75

Throw out highest and lowest.

Miki: 7.00 7.00 7.00 7.00 6.00. Average 6.80

Kiira: 6.75 6.75 6.75 6.75 7.75. Average 6.95

Kiira wins even though Miki is favored by four of the five surviving judges. (Notice that judge number eight did have influence on the success of the consiracy. By throwing himself on the sword, he made sure that judge number seven would survive to carry the day.)

Now multiply this by all the PCSs and all the GOEs and you can see how Kiira could easily come out ahead overall, even overcoming a deficit in TES.

But why does "close competition" convert to a mere difference of 0.25 by the honest judges, and a whole 1.0 by the crooked ones?

I mean yeah, that's how it goes because in 6.0 the smallest difference up or down is 0.1 of an ordinal, and the crooked AND the honest judges marked the two skaters up or down by 0.1.

But in CoP, the close ones have to give Miki only a 0.25 advantage by points, and the cheaters are by 1.0 point? That's 4x the difference.

That's not objective, mathematical "proof". You're saying that that's how it would go (or assuming that's how it would go.)

 

[janetfan]

[B[/B]

CoP is much better.
It would allow more room for cheating, but it has more advantages anyway.

.

I think Einstein, da Vinci and Newton, all believers in the meaning of Occam's razor would conclude that CoP has more possibilities for cheating.

And they would not crunch any numbers to reach such a conclusion but would rely on logic and the natural laws of the universe.

 

[prettykeys]

Occam's Razor has to do with numbers of assumptions, not variables.

I think.

 

[shallwedansu]

Maybe your English is worse than normal. :think:

:disagree:

:no:

 

[Daniel5555]

Mathman
I think 2 conspirators is not very dangerous. 3 of them yes, but 2 is still ok.

prettykeys

Maybe your English is worse than normal.

You just proved what I said about you.

janetfan

I think Einstein, da Vinci and Newton, all belivers in Occam's razor would conclude that CoP has more possibilities for cheating.

And they would not crunch any numbers to reach such a conclusion but would rely on logic.

If you would read my posts, you would know that I didn't say anywhere that CoP has less possibilities for cheating.
I said that it's better (generally) than 6.0 in response to your phrase:

"the simplest system is usually the best."