ManiaExchange
  • SpecialStr
  •  Search tracks for ""
  •  Search tracks built by ""
  •  Search mappacks for ""
  •  Search mappacks created by ""
  •  Search users named ""
Location: Home  Community Forums  Competitions  AuthorMania - Event Details 
AuthorMania - Event Details
Page: 5 of 6 Post Reply
22 December 2014 16:48:00  
lol wtf this doesn't make sense at all the results...

zengo: Karma 77%
mlp (who wins): karma 75%....
3rd place RV: 72%
while on spot 4, 7, 8, 10 and 11 are tracks with an higher score (karma).
(karma is an inbuild voting system that calculates the results so far...)

I suggest that you should not take the total of votes on every map but take the avarage number of votes on all tracks and calculate the score on that.

also I could remember the server having more than 2 vote options If I'm correct: "--", "---", "++" ,"+++", etc. so how do you calculate these?

this seems to be dishonest to be honnest.
Last edited by FT»lolig, 22 December 2014 16:48:57
Awesome Dude!
Location: BE
22 December 2014 17:48:37  
Zengo says:
... some would say that the answer to an unequal amount of votes is normalizing the results by dividing the score by the amount of votes each map received.

It's even mathematically proven. Normalization is indeed the keyword. The absolute points are not really good, relative points are a lot better 8-|

Good Job for the organisation. I'm amazed by how many authors delivered in time.
But I have to agree, the voting system is really unfair;)
Last edited by haenry, 22 December 2014 17:49:59
gado is a
Location: DE
Another explaination :) 23 December 2014 02:50:04  
Zengo says:
... some would say that the answer to an unequal amount of votes is normalizing the results by dividing the score by the amount of votes each map received. This system you devised to account for different amount of votes only causes the total number of votes to have more impact (e.g 100-50 would beat 49-0).

Congratz to poivrot, but i feel a little cheated by the system.


Zengo, I assure you, you were not cheated and I am sorry you feel that way. I am not dishonest as Lolig as labeled me :) . I will normalize the results for you, as I do not want anyone to feel as if I chose a quick and easy method for calculating results. However, I have done this many times before and have tried various methods in the past. I know the "normalized" results will reflect in the same manner that I have presented the AuthorMania results. The main idea is to always keep in mind how many. A simple percentage or average of the like/dislike only gives a grading and does not reflect the aspect of how many. Think of this... If I were to say that I would buy a pizza for you if you could eat 75% of it, would you do it? If you can't eat 75% percent of it, you pay for the pizza and I get to take home the rest ....:) .... I am sure most people would say yes. Now... Let's say you've got that pizza in front of you... and it turns out to be the size of a small parking lot. Would it have helped to know how much (or how many slices :p) pizza would equal 75%? Now I want pizza(brb) ....



Lolig, Haenry - Local Karma average is essentially a percent that gives a track a grade like C-, B+, or A-. By looking at the percent you do not know how many votes. For example, if 1 person votes +++ on a map and no one votes negative, that track is at 100%. It's important to know how many voted otherwise scenarios similar to the one vote example would occur.

Taking the ---, --, -, +, ++, and +++ out and calculating the votes as either like or dislike removes the factor of someone liking or disliking a track to a specific degree. For the majority of the tracks, there were so few negative votes that the positive votes would be greatly inflated if we were to use the ---,--,-,+,++,+++ system and give a value to +++ or ++. A simple yes or no was used in this situation. If we relate the score to the number of votes for that track, we would simply get another percentage that merely gives the author another grade like the Local Karma average, which doesn't convey how many.

Percentages and averages can be misleading. I would consider my method as dealing with "absolutes", as Haenry has stated, which presents the raw data. I only placed the Karma Average in the results so the authors knew what their average rating was for the Local Karma Average.


To normalize the votes as Zengo suggested, I have calculated the number of likes per map, and the number of dislikes per map, in proportion to the total sum of votes. This means everyone is given a score based on the number of likes and dislikes they received in relation to the total number of votes.

# of likes / 818 total votes
minus
# of dislikes / 818 total votes




----------- Results -----------
*** 5 decimal places is enough, (y) haha

(.05501) Poivrot [.05867 - .00366]
(.05012) Zengo [.05134 - .00122]
(.04768) RachoVolker [.05134 - .00366]
(.04646) S'habba.hu [.05012 - .00366] <-- Karma used as tie breaker
(.04645) Keksbude [.05378 - .00733] <-- although I didn't need to
(.04401) T!K [.05256 - .00855]
(.04278) Zack11 [.04889 - .00611]
(.03790) Sic [.04767 - .00977]
(.03667) Blinnemans [.05378 - .01711]
(.03423) MirExpress [.04034 - .00611]
(.03301) Alex bf [.03545 - .00244]
(.03057) Sayo [.04523 - .01466]
(.02812) Niborz [.03789 - .00977]
(.02323) Keirabxtch [.03178 - .00855]
(.02200) Adsun [.03422 - .01222]
(.01712) Ad'Gado [.02689 - .00977] <-- Karma used as tie breaker
(.01712) Danneboy [.04034 - .02322] <--
(.01589) Kamakaze [.02689 - .01100]
(-.02200) PapyChampy [.01589 - .03789]



As you can see, the results of normalizing this information reflects the same results presented for AuthorMania. The only difference is what it looks like. I prefer seeing a simple representation, such as the results on the ARK site, rather than having to decipher numbers with decimal points. I am sure others would agree with that8-|

It's only natural for some to disagree. Everyone has an opinion of how they would like things done, and should properly do so when they host events. With that said, If we do another AuthorMania, as some are requesting, it will be different in some ways based on the information I have gathered from forums and talking in-game with authors and racers.


^^^
I just read that....:wait::wait: .... quite long:p .
Old Age Caravanner
Location: US
and.... 23 December 2014 02:57:03  
Maybe next time, I'll post normalized results rather than a simplified representation.:s
Last edited by RD, 23 December 2014 03:27:25
Old Age Caravanner
Location: US
23 December 2014 10:39:09  
I have to agree with you on the point that relative results in percent are very inaccurate for low amount of voters. It is obvious that the results are inaccurate when only one guy votes.
However in this case every track received at least 30 votes, so I think it's fine to use the relative system.

The problem with the absolute voting here is that you have different amount of votes for every map. What can alex BF do in his case because his map only received 29 upvotes and 2 downvotes, while others in front of him received a lot more downvotes (e.g. blinnemans). But AlexBF is still further down in the ranking than blinnemans, as there were simply less player trying his map (for whatever reason).
It's the different amount of voters that somehow creates the unfairness.

I see what you understood by normalization, but just dividing the first results by 818 doesn't help. You have do normalize for each map ;)

I quickly calculated the relative results here. I'm not saying that this is the perfect way to it, but in my opinion it's more fair, due to the different amount of votes.
(sadly there is not the perfect voting system, but as stated above: the relative system works well as long as you have enough votes for each map ;) )



PS: I'm not asking you to change the results, but I want you to know the difference between absolute and relative votes for all kinds of further contests :)
Last edited by haenry, 23 December 2014 10:40:55
gado is a
Location: DE
23 December 2014 13:23:07  
Awards won't be counted:'( ??
(it's a joke to add mess around this topic) ;)

All joking aside, congratulation to the winner and thank you RD for this wonderful event.
That was really fun!
Last edited by MirExpress, 23 December 2014 17:26:50
Learner Driver
Location: FR
23 December 2014 17:30:52  
well the normalized results and after reading the explenation: it makes sense :)

btw I never labeled you as dishonnest :p

I just told this system felt dishonnest imo. I never labeled a person as it ;)

lolig
btw gz to everyone :)
Awesome Dude!
Location: BE
23 December 2014 19:54:26  
haenry says:
What can alex BF do in his case because his map only received 29 upvotes and 2 downvotes, while others in front of him received a lot more downvotes (e.g. blinnemans).

--- The other authors received more Likes, which compensated for their dislikes.


haenry says:
But AlexBF is still further down in the ranking than blinnemans, as there were simply less player trying his map (for whatever reason). It's the different amount of voters that somehow creates the unfairness.

haenry says:
The problem with the absolute voting here is that you have different amount of votes for every map.

There are many reasons why a person does or does not vote and we can never understand every situation why voter X only voted on 4 maps, while voter Y voted on 10, and voter Z voted on 19. The fact that player X decided to vote, regardless of the situation, is important. It may be possible that player X had played on all 19 maps, but felt that only 4 maps deserved a + or a -. In the end, shouldn't all votes count? I feel that every vote is important and limiting it to a max of 30 votes only discredits the player votes over 30 and negates potential positive or negative votes that could have an impact on the score. It is reasonable for you to feel that unequal votes creates unfairness, and yes, to an extent I agree with you. However, in an ideal world every racer would have played and voted on all maps, but of course this isn't the case and we shouldn't expect that to ever happen, haha. Ultimately, all votes should count, regardless if one map received more than another.

Haenry, I wish to understand your method. I have a few questions about your results. Well, many questions, but it may potentially help me to reconsider my scoring method for other events :p

- Is it only based on the like results?
- How are the dislike results factored in?
- Take Sic's score, what is the 83% out of?
- Is there a limit on how many Like votes count?
- Is there a limit on how many dislike votes count?
- Are the total votes, 818, taken into account?
- What formula did you use to calculate your scores?


haenry says:
However in this case every track received at least 30 votes, so I think it's fine to use the relative system.

If I am understanding your relative figures correctly, you are placing a limit on how many Like votes count but not placing a limit on how many dislike votes count, correct? This means that Blinnemans is graded harshly and moves from 9th to 13th. Sayo would go from 12th to 14th and Sic from 8th to 10th. They would be going lower in the ranks due to receiving more dislike votes, which do not have a limit such as the limit on the like votes. It makes me think of this: X map receives 300 likes and 29 dislikes. If all maps received at least 30 Likes and there is a limit of 30 Likes, and the 29 Dislikes are taken into account, you're left with 1 Like. Similar to your limit on Likes, if we place a limit on Dislike votes, then it would be with Zengos map. Since he only received 1 dislike, then all maps would have a limit of 1 Dislike because all have received at least 1 dislike. This of course, would have strange results.


Haenry - I took a considerable amount of time to decide what may be appropriate for scoring the results as I prepared this event over 2 months, this is why I have a lot to say :) . You can see that I obviously disagree with using relative figures in this situation, as it only conveys a grading of a limited amount of Likes and does not describe how many. For other events, relative figures are acceptable, but not for the results of AuthorMania. I still wish to understand how you decided to calculate relative figures, I may apply this to other events in the future.(y)

- Mir - You had a large amount of TMX awards, which is awesome :p . Maybe next time we'll have a top awarded TMX map. That would be a good idea. (y)

[Edit]
- Lolig - Thanks for the clarification :)
Last edited by RD, 23 December 2014 19:56:27
Old Age Caravanner
Location: US
23 December 2014 20:14:51  
haenry says:

PS: I'm not asking you to change the results, but I want you to know the difference between absolute and relative votes for all kinds of further contests :)


change it ,change it:p:p
Seriously... congratulation to the winner (l)
From my side i really sorry that i couldn't play online ,was really impossible
No matter the results... Projects like this one keep the game alive
Bravo ARKs
Old Age Caravanner
Location: GR
24 December 2014 00:49:36  
RD, it's nice to see you investing so much time.

My calculation is quite simple. It falls under the category "rule of three".

  • Here's a quick explanation.
    We calculate the result for every map independently, thus the overall amount of votes on all maps is not considered.
    We start with one map that has T = number of total votes, D = number of dislikes and L = number of likes.
    Obviously, you can see that L+D=T is true.
    We set T = 100%. => If the number of dislikes D = 0, then L = T = 100% and the map receives a voting of 100%.
    This is usually not the case and in these other cases we have to calculate the relative part of the total of likes to the amount of total votes.
  • It's easiest to show it on an example:
    D=3 and L = 47 => T=50.
    Let's find out how much 1% is. T/100 = 50 votes /100 = 0.5 votes. Thus 1% = 0.5 votes or 1 vote = 2% in our example.
    We can now deduce that 47 votes = 47 *2 = 96%.
    So we get a score of 96% of likes for the map.

  • But there is an easier and shorter way to calculate the exact same thing:
    First up: "per cent" stands for "per hundred" i.e. the value divided into 100 parts. 50% = 50/100 = 0.5, 25%=15/100 = 0.25 . 100% = 100/100 = 1. Let's call the result of this calculation value/100 = x
    In our case the value is the amount of likes = L and the total amount of votes is T.
    As we set T = 100% (examples above), we can just calculate the Ratio: L/T = x
    x now represents the result not in percent, but as a ratio. A relative number between 0 and 1.
    By simply Multiplicating with 100% we get:
    0=0% and 1=100%

  • x is the important final score, which I used in my calculations. You can multiply it by 100 to get percent, or even by 1000 to get per mill (as you know from the alcohol labels). There are even things like ppm = parts per million.
    If you wish you can even multiply it 1337 to get the result between 0 and 1337. Let's call it ppe (Parts per elite :cool: )

    This is basic school math and my explanation turned out to be longer as it actually is. (I added too many examples and an alternative easier way :d)

    I hope that this explains all your questions! :)
    Last edited by haenry, 24 December 2014 00:52:43
  • gado is a
    Location: DE
    Page: 5 of 6 Post Reply
    © ManiaExchange (mania-exchange.com, mania.exchange) 2020. • Terms and ConditionsPrivacy PolicyTop  •  Report a Problem