Loctav wrote:
This entire discussion is bullshit. Get out.
Just ignore this if you aren't interested in the discussion, please.
Loctav wrote:
Your entire asset of "determining skill" fails to begin with, because a tournament is not only influenced by skill, but also by environment.
Do not forget that every stage has different pools of extremely different skill level that it requires to play. You are comparing apples with peaches.
This depends on how you define skill. One could define skill in a way that is completely isolated from the effects of the environment, and in that way performance would be dependent on skill+environment+chance+(other possible factors). Another possibility is considering skill as dependent on environment, in this case, "skill during a multiplayer game with the pressure of a possible prize" is something different to "skill during single play without pressure", for example. The relevant skill for the tournament would be "skill during a multiplayer game with the pressure of a possible prize".
The fact that different stages have different map pools, does make using the tournament as a mean of determining a certain kind of skill, that is independent from the tournament format itself, a harder task. In the "amount of birthmarks tournament" analogy, this is similar to changing the winning criteria from stage to stage (for example, in the first stage, it is "most amount of birthmarks in the face wins", while in the second stage it is "most amount of birthmarks in the arms wins"), which makes the relevant skill parameter of each player not be unidimensional anymore.
In practice, there is some correlation between 1 dimension of skill and another (i.e. teams that do well in a certain map pool tend to do also well in another map pool). If the correlation is strong enough, it is possible to consider a unidimensional value for the overall skill of a team; if there is no correlation, then the the overall skill of a team can't be separated from the format of the tournament itself.
Loctav wrote:
tl;dr: all your examples are highly hypothetical and not applicable to anything we do here.
The examples were based on an ideal scenario. If a certain method is not reliable on ideal conditions, then the reliability of the method in non-ideal conditions is expected to be even worse in most cases.
This is analogous to trying to find the triangle with the most area. You could try finding the answer by measuring the perimeter of the triangles, but, even if all measurements are done perfectly, it is possible that the triangle that has the biggest perimeter is not the triangle with the most area. Since the method is not reliable with perfect measurements, it is not reliable either when there is a chance that some of the measurements are wrong.
Loctav wrote:
Double Elimination perfectly works to determine a podium in a simple "who drops out first, who drops out last" format.
Whether or not the Double Elimination works perfectly depends on the objectives of the tournament. If the objective is: determine a podium in a "who drop out first, who drop out last" format, then the objective is met indeed (since that is what the tournament literally does).
But, that doesn't mean it meets other objectives, such as "determining which teams are the best at playing the game in a tournament", or "give entertainment value to the players and spectators".
Loctav wrote:
Nothing is under our control.
Because of the format of the tournament, even if the team groups are selected randomly, the overall method is biased (with respect to the result of the third place). There are 2 possibilities:
- The organizer decides to set up the groups randomly: the decision to do so introduces a probability where the third place is not given to the third best team (which is independent to the uncertainty caused by the fact that a better team doesn't always win against a worse team in a match). This is most likely the case of this tournament.
- The organizer uses a non-random criteria to set up the groups: the 3rd place in the tournament is affected by the criteria used. The third place wouldn't be determined in advance certainly, because of the probability of a better team not always winning against a worse team, but the 3rd place would incline towards a certain result.
Loctav wrote:
What you complain about is a very unlikely to happen constellation. Every tournament format has said edge cases. They are anyways super unlikely to happen. And even if it happens in 1 of 200 tournaments, people will just book it under "bad fortune" and move on.
In the ideal case where a team always wins against another team if they are more skilled (and it is possible to talk about a one-dimensional measure of skill that is relevant to the competition), and the groups are selected randomly, the probabilities are:
Chances of the Best Team entering the Double Elimination Tournament: 100%
Chances of the 2nd Best Team entering the Double Elimination Tournament: 100%
Chances of the 3rd Best Team entering the Double Elimination Tournament: 154/155 = 99.3548%
With the tournament brackets selected by seed, then:
Chance of the Best Team that entered the Double Elimination Tournament winning 1st place: 100%
Chance of the 2nd Best Team that entered the Double Elimination Tournament winning 2st place: 100%
Chance of the 3rd Best Team that entered the Double Elimination Tournament winning 3st place:
100%Edit: 100% was in the case with 8 teams, with 16 teams the probability is less than 100%. The probability is approximately 88.5831%.
So, in the ideal case, there is a chance of about 11.9884% of the 3rd best team not winning the 3rd place.
If you are curious, those are the probabilities if the tournament brackets were selected randomly instead of by seed:
Chance of the Best Team that entered the Double Elimination Tournament winning 1st place: 100%
Chance of the 2nd Best Team that entered the Double Elimination Tournament winning 2st place: 100%
Chance of the 3rd Best Team that entered the Double Elimination Tournament winning 3st place: 80%
So, the chance of the 3rd best team not winning the 3rd place is 159/775=20.5161%