by BigBallinStalin on Sun Mar 10, 2013 2:00 am
SaMejoHn wrote:BigBallinStalin wrote:xiangwang wrote:Chariot of Fire wrote:You can repeat all you like about the stats not being favourable for either of you as individuals when you play together as opposed to when you play alone. But the higher win % when you are together (in a very convenient silent agreement to share the spoils of victory) is certainly not favourable for the other 6 poor sods who have joined your games. So whether you stand to benefit individually is irrelevant. As a 'team' you do, and this is unfair and detrimental to the chances of the other players.
I believe it's stated that the win % is lower as a "team" (or whatever you allege) than playing individually. I have already calculated the numbers and have shown that there is NO higher win % for 8 player freestyle playing together or when we play as a team. The OP already agreed with me thus refuting your largest allegation. When most people cheat or play multi, their win % as a team increases by at least 10-20%, for us it doesn't change or is slightly reduced. We both play for our own win. If we intentionally wanted to collude, i think we can hit 80-90% win rate together.
Nevertheless,
assuming that you two are colluding,
then this could boost your win rate by 10-20%.
Assuming that you two would stop colluding,
then your "natural" win rate would be 10-20% lower than the current win rate.
So, with this in mind, it could be the case that y'all two aren't that great, so y'all collude in order to attain a 'normal' win-rate of 50% or whatever seems normal--compared to better plays who don't collude yet do attain normal win rates. Therefore, comparisons of your current win-rate do not help us determine if collusion has or has not occurred.
But if the win rate is the same when they are without each other (we know that it is), doesn't that contradict what you are saying?
No, but I probably wasn't clear enough. I'm not comparing their win-rates with each other and without each other. I'm analyzing K&X's win-rate with collusion, and K&X's winrate without collusion in order to show that X's #2 argument fails to support his position that they were not colluding.
X basically makes two arguments:(1) "Hey, if we were colluding, then our win rate would be higher (90%). Since it's 50%, then we aren't colluding."
Obviously, this doesn't settle the issue because if they're smart, they can collude enough so that they do not attain a 90% win rate. Instead, they could opt for a 50% winrate.
(2) "Our current winrate is 50%. If were were colluding, then our winrate would be 20% higher. Since it's not 20% higher, then we aren't colluding."
I wondered: this proves nothing. If they weren't colluding, their winrate could drop to 20%, thus it would be 30%. Since the counterfactual is unknown, i.e. since we can't determine what would have happened, then we cannot conclude if collusion occurred or didn't occur. How do demonstrate my case?
Thus,
BBS basically makes the argumentImagine two worlds. World Red is where K&X collude. World Blue is where K&X aren't colluding.
In world Red (with collusion), K&X attain a 50% win rate.
In world Blue (without collusion), K&X attain a 30% win rate--because they aren't that good without collusion.
Okay, back to the RL.
Therefore, even if their winrate is 50%, and that this does not allegedly constitute as colluding in X's opinion, it still is not clear if they colluded or not. It could be the case that given their best ability in collusion, they can only attain a 50% winrate. Without collusion it would drop to 30%. But I made up these numbers of the counterfactual for the sake of that example. In other words, we have no idea what their winrate would've been without collusion, and we cannot from the beginning know if they colluded or not.
Which world is the real one? Red or Blue? We can't know lol.
In short,
1. since the counterfactual cannot be demonstrated,
2. and since we do have
a priori knowledge of their collusion or non-collusion,
3. then no one can conclude that collusion did (or did not) occur--given X's arguments and anyone else's who uses similar reasoning.
Big problem.