•   days until the Eurovision Song Contest Final in Rotterdam!

Resolving the combined ties


Which tiebreak system do you prefer to resolve ties resulting from the combined votes?

System A 9 vote(s) 40.9%
System B 12 vote(s) 54.5%
System C 3 vote(s) 13.6%
Multiple votes are allowed.

    How is it double work?
  2. Wiggles Administrator

    It's not that much extra work, I think. Let's look at the scenario:
    a 8
    b 7
    c 6
    d 5

    a 5
    b 6
    c 7
    d 8

    So all songs get 13 points, but the pairs are
    a 8, 5
    d 5, 8
    b 7, 6
    c 6, 7

    So first ranked on high point, then ranked on leader's order. Yes, it's one step more than option A, but it wouldn't be that difficult.
  3. Julio STRANGLES

    to implement double rules? one if for the normal tie break other for the unsolved tie breaks, having to calculate all of them? cause at the end of the say the same thing that happens on option A can happen on option B. the leader might give a song 10pts while the other give it 0 and on option B the song that got 10pts would go ahead still. so i see no point on complicating things more than what they are necessary and make life harder.
  4. Julio STRANGLES

    Not wanting to be offensive , but what might be easy for you might not be as easy for others

    I genuinely don't see it as making it harder and i also don't think the problem is with a
    10-0 or 0-10 split. The leader's vote takes precedence in either. That is established. I think the problem would be with the 5-5 split and then, whichever way this was resolved, it would be unfair to one person. This is something that we just have to accept. That is my point.
  6. Julio STRANGLES

    Someone would always find it unfair at some point, but thats how 50/50 works, if people didnt like the idea they should have voted for separated votes rather than the combined, now they simply have to deal with the fact someone will always have his votes changed due to the combination of votes, once again, its how 50/50 works

    And that is exactly my point. You see, we are in agreement, partner <<
  8. Fat Sandera Member

    I wouldnt like to see how a song i voted high is ignored just bcos the leader didnt like it. Either i'd like to see how a song i voted poorly wins over a song i voted with 12 or 10 cos we both gave it 4 points. To me System B is quite fair, it doesnt void non leader votes completely and keeps giving the leader priority which is his/her purpose.

    To explain why I voted B: I feel that the Leader of a team shouldn't be the Dictator of the team. Yes (in my example), CTP's votes will be dominant, but that doesn't mean that my votes should hardly be worth anything. And that's what I feel is happening with System A. Cause basically, there you're gonna say that first we're putting the song the leader liked best, secondly the song he sort of likes and third the song I really liked. See the order? That would be him-him-me.
    So even though I think the leader system is good, System A would make it an absolute bonkers idea, cause it really is not 50/50 anymore. CTP's points would be much much more dominant than mine.

    And basically, I also don't see how System B is 'much more complicated' than System A. I really don't, so for me, that argument can be forgotten.


    I voted for System B as well. The reasons have been said, we made the decision to have a leader in order to solve tie problems, but when big differences in the votes appear, as in that scenario, IMO it doesn't make sense to solve the tie only with the leader mastership anymore. System B is a system for a team, System A is a system for a leader with a helper. Prevalence of the votes of the leader? OFC, that's why we chose it. But I don't think that is what we are applying with System A.
    And I don't think you should be so belligerant with this, it's a matter of opinions. And if JW has asked for our opinions it's because he wants them and he's able to implement any of the systems, tougher stuff has been made with a huge success-and thankfulness from all of us, and we should vote between the options that we have been given!
    Jw, in the case you cannot implement one of the options you've given to us, let us know and we will close this noisy coven and take the option that you can implement.
  11. Curtis Belgium is a non country.

    I voted C because I can.
    Nessie likes this.
  12. adamacs RUTH DAVIDSON FOR PM.

    I support B simply because it is the same format that had been used in esc until last year. It is the most fair so that a son liked by both is not rewarded simply by being liked by both with the leaders votes being a last resort
  13. DenDutch Well-Known Member

    Oh great, if the votes stay this way, we have to find a tiebreaker to break the poll tie :D
  14. jw HI. I'M ADAM SCOTT.

    I would use my discretion to implement my own preferred system (out of the tied options) in the event of a tie :P
  15. Genesis2703 Pointless

    I went with B. Did some maths which i'll explain below. But I found the conclusions surprising. Before that though, people who say option B is complicated are really are foolish (or have not seem many things that are actually complex in their life...)

    Yeah as Wiggly said, this decision is fairly important. Now onto the maths:

    I calculated Spearman's Rank Correlation Coefficient for two randomly generated rankings (i.e. of both partners rankings). This number is between -1 and 1 and decides how much rankings agree. for example me and Sokko would be closer to -1 as we disagree loads, while I imagine for example TWO and Rua have a higher number. The closer it is to 0, the less strong the correlation (and is +/- is whether it agree or disagree).

    The two rankings I generated were 7,10,9,1,5,2,8,4,6,11,3 for the Leader and 5,1,7,2,10,4,9,11,3,6,8 for the non-leader.

    Sadly this still isn't too simple, in the 1st list. The 7 means song 1 was put 7th, song 2 10th, .... song 11 3rd. The correlation coefficient of the partners is -0.036 so basically there is nothing mathematically in common overall. Now under system A the end ranking is 6,4,9,1,8,2,10,7,3,11,5 and system B is 6,3,9,1,8,2,11,7,4,10,5. So for both song 4 got 12 points (22/24 scored if you do the maths) while the ties are with songs 2/9 7/10 and 5/8 to which the 1st two pairs did indeed swap ranking. Now the coefficients are:

    Leader with A: 0.656, Leader with B: 0.609, Non-Leader with A = 0.664, Non-Leader with B = 0.709.

    The drop for the Leader and rise for the Non-Leader are obvious as section B slightly hinders the leader. The ones among you thinking ahead may even notice the average of the coefficients is slightly higher with A than it is with B. So you may wonder why I voted for B, as maths doesn't lie does it? Well:

    1) This is only one simulation, someone with more time (and programming knowledge as online calculators are slow as fudge). Everyone with any knowledge of statistics knows that 1 random simulation is not that accurate, what would need to be done is a loop with 1000+ simulations for sure. And I do not have time to do such programming at the moment.

    2) Notice that in this unusual case the Non-leader was more happy than the Leader from option A already. (This case is clearly unusual by the definition of what option A does). so when we apply option B to the ranking, the leaders number will be more sensitive than normal, while the non-leaders number will be less sensitive than normal. (i.e. 0.6->0.5 and 0.2->0.5 is more likely than 0.5-->0.4 and 0.6->0.9)

    As a fair mathematician I will say the one point that is definitely FOR idea A: We need to consider the points jump between 1st-2nd-3rd. 5th and 2nd = 3rd and 3rd (and lets say for example they are fighting over 2nd/3rd place) so mathematically being biased towards the Leader (so letting 3rd 3rd get 10 points) is BETTER for the total average of the SRCC (correlation coefficient). This is based off the mathematical equation for it which I will NOT explain.

    The best way to resolve all these issues is to run loads of simulationshe like I said, BUT sadly my case was weird in the regard of point 2) but on the whole the average "happiness" is going to benefit more with B.

    There are a couple ethical concerns as well, and I apologise for the very technical post. I do not mind clearing up details that have caused confusion however :)

    Well thanks for clearing that up, Gen :P
    Genesis2703 likes this.
  17. jw HI. I'M ADAM SCOTT.

    I understood nothing. So DO clear it up, as you promised, won't you miltz.
  18. Wiggles Administrator

    My excel work has allowed me to show the different implications in the leader influence graphically, not as mathematically as genesis has done. I hope this is clear, but I'm not a great instructor so maybe it isn't.

    This should be clear enough, leader is to the left, non-leader on top, and this is the sum of all their possible votes combinations.

    Now, the combinations are 121, so they need to be ranked by importance. With no tie breaker there are lots of ties. (I use rank average and it's rounded, if you're scrutinising the actual numbers in the table)

    Under system A, the rank of the combinations looks like this:

    Under B, like this:

    And C like this:

    Now there isn't much difference to be spotted by the naked eye there, which makes it useful to take the difference between the systems.

    This shows no-tie-breakers System subtracted from System A. You can see in the green areas where ranking goes up (to a lower number) and the red where it goes down. This image shows how the greatest skewness is with high leader-points and low-non-leader points and vice versa, showing the leader dominance the greatest in those corners.

    System B minus no-tie-breaker shows a greater dominance of high-pointers by either come at the cost of mid-points by both, like julio has said, but the dominance of the leader has been lessened severely, as the difference in the corners is only 1 place.

    And system C minus no-tiebreakers show a picture similar to a with a leader dominance but also a severely decreased influence by 0 pointers. Compared to system A, leader's 0 pointers are unchanged, but non-leader's 0 pointers lose a lot in ranking.

    Now as to my personal opinion, if there were more songs and therefore more 0 pointers by either voter, I would favour C, or a combination of C and B. In my opinion there is too much leader influence in A and C, and the fairest is B.

Like this page on Facebook

Comments posted on the forum and in the chat room are the sole opinion of the respective author and not escChat.com.
escChat.com © 2009 – 2021