I just wanted to throw this out there because I've had some discussions about it with friends and it seems a worthwhile topic for further discussion.

Let me start explaining my point of view by the following: A correlation does not imply a causation. It seems a fact that is too often ignored. I've heard some people say, for example, that they've read studies that show that people with more than 2 hours of screen time (TV, computer) per day have a larger chance to be obese and have heart problems, etc. This led some to believe that reducing your daily screen time to less than 2 hours would decrease your chances of obesity and heart problems.

I think it's clear that this line of reasoning doesn't work - there are simpler counter-examples in which a correlation of two completely irrelevant variables is compared, but I think this one shows that even if the variables are connected, they don't actually imply much. A correlation does not imply a causation.

Let's get back on the subject of StarCraft II. Many people like to use the argument of "In GSL, X percent of ZvP are won by zerg, hence imba!" Now, it's necessary to point out that though a correlation does not imply a causation, it does not deny it either. That said, the win rate argument seems quite silly to me. While it can certainly support an argument in which one side claims imbalance, it cannot be the basis for an argument. It can certainly be a cause for investigation as to the cause of potential imbalance, but one cannot say something like "ZvP winrate is 65%, therefore zerg is imbalanced" or "ZvP winrate is 65%, therefore infestor is imbalanced".

It might seem obvious that this is true, but every time you hear (or read ) players using statistics to argue balance, ask yourself if they are using the statistics as a basis for their argument or whether they have an actual gameplay-based argument that they use statistics to support. It's surprising to me how many people actually fall into the former category.

The main counter-argument I've heard is that the winrate statistic is not a correlation, merely a statistic. If you are talking about balance, however, then you are essentially projecting the past winrate to the present and future, thus adding another variable, time. You are essentially saying that there is a correlation between winrate and time. After this, someone might argue that the game has not changed since the time the winrate was recorded (assuming no patch, obviously). However, that is not taking into account the metagame state.

Imagine a world in which MLG Anaheim never happened and the SlayerS team didn't come up with various new ways to abuse the blue flame hellion. Imagine that it stayed in obscurity. Would it have been patched? I tend to think that it would not. What happened with blue-flame hellions is an example of a metagame shift. Nothing changed balance-wise, but Terrans discovered the potency of the blue-flame hellion. This sort of occurrence means that even if no patch is released, the winrates of the various matchups will vary due to metagame shifts.

Until we are absolutely certain that every single possibility in the game given a certain patch has been explored, the metagame is bound to shift around. Eventually, once all possibilities are uncovered, the winrates will be steady and more reliable as a source of discerning balance. The false assumption that many who argue balance using winrates make is that the winrates are already steady, thus implying that every possibility has already been explored. Personally, I believe that it would take years to explore every possibility, but this point could be argued.

Thoughts?

tl;dr - Apply the rule of "Correlation does not imply causation to StarCraft 2 statistics as a method of arguing balance. Use the actual game, not the numbers."