There is a large difference between race usage between Korea and everybody else. Koreans love Terran, and far more Terrans compete at a high level there. I’ve been checking out the start of the games history. From 2010 - 2013, the devs slowly nerfed Terran, and buffed Zerg and Protoss. While this looked good for the fans watching tournaments in Korea, the 2012 World Championships had much less Terrans than either other race! This is with a sample size of about 400:
h ttps://imgur.com/a/7KAlvS5
Terran definitely became the weakest race worldwide after the nerfs. Maybe they should have only ‘balanced’ the game in Korea?
I think koreans had fewer opportunities due to the region lock and that negatively skewed terran’s odds to win premiers given the best terrans were in korea. That’s one of the reasons historical tournament data is very misleading and not likely to be correlated with balance. This issue continues today because the korean scene downsized to probably 1/5th what it was. There are probably 5x fewer champion tier terrans as a result of the disastrous region lock. We used to have Flash, Bbyong, Taeja, Bomber, MMA, MVP, and the list goes on and on. Now we have Byun and Maru.
As it stands, terran seems to be in a good spot:
https://i.imgur.com/bWmUamU.png
The top 10 of the ladder is fairly predictive of the semi finals for premiers. There are 3 protoss in the top 10 but they all cluster near the bottom. The difference between rank #1 and rank #10 is ~500 mmr which is a crazy spread in performance. The highest protoss vs the highest player is serral vs maxpax, but maxpax won’t attend offline events. The next closest after that is skillous with 6700 mmr. We have a spread of 491 mmr between the peak talent for protoss and zerg.
The elo equation says the win-rate between these players is 70/30% win-rate split. That means Serral is favored at 9 to 1 odds to win a best of 9 (finals). Double elimination brackets are common with best of 5’s. What about that? He has a 97% chance to advance.
Serral is going to dominate every premier tournament in 2025 unless they change the tournament rules to limit players to 1 premier win per year per race. Then Serral can only win 1 with Zerg. After that, he has to play terran or protoss the rest of the year if he wants to win another one. If that were done, we could very likely see a skillous or hero premier champion. Without that, Serral is going to win with maybe an odd exception to losing to Reynor or Clem.
This is why the region lock was so disastrous. Korea was sc2’s #1 fan, and they killed the korean pro scene and forced players back to broodwar with the region lock. They did this to increase the diversity of people winning tournaments. But now Serral wins them all. So they ironically reduced the diversity by easily a factor of 10. They pushed out all the players capable of beating serral and then went SurprisedPikachu.jpg “why does serral win everything.”
It’s hilarious that they thought that downsizing the size of the pro scene was going to make it more diverse. It’s absolutely comical. They needed policies that made it harder for repeat winners without excluding players from the process. Race switching incentives would’ve accomplished what they needed without any of the negative side effects. So Innovation wins a GSL and now he has to play protoss or zerg. This increases the racial diversity because it reduces repeat mono racial victories. It also increases the diversity of players because a player takes a small nerf after winning since they have to offrace.
This is off-topic, but I also think that the only match ups should be PvT, and ZvT. It is far more thematic to see the humans versing the aliens or the bugs.
I would disagree. You have to balance the game for the high end, although there are probably ways to make low-level play more even while achieving that goal. I do think things like giving all casters an auto-attack (something weak to prevent them from rushing forward) would certainly help.
There is no distinction between the two. If your theory is that performance flip flops with skill then that is a theory that skill is the cause. You aren’t describing balance, you are describing skill. If balance is the cause, it must affect the entire racial faction excluding outliers.
What you are saying would be like eating cake to treat a headache then concluding cake treats headaches without mentioning the people you gave the cake to didn’t have a headache to treat.
When the dependent variable changes separately from the independent variable, the dependent variable is not dependent. Headaches are not dependent on cake and pro level performance is not dependent on balance.
It’s complete nonsense. It’s equivalent to the cake treats headaches example. A logically correct statement would be that protoss is favored at all levels in broodwar, but at the highest levels the skill differences overcome the balance advantage.
If the performance of protoss changes within the protoss grouping, protoss isn’t the cause of that change. If performance changes between races then race is the cause. A change in performance between top protoss and bottom protoss is a change within protoss and protoss isn’t the cause.
We’ve been through this, there are obviously units that scale better with skill than others, due to a skill ceiling. You can’t get much better with Carriers, but you can with Blink Stalkers.
Well you’ve said it before, there are some strategies that are really broken below pro level, the most dominant being air toss. I’m not sure how true that is after the patch, but you get the point.
Conclusion: Eating cake is overpowered at causing headaches when hit on the head.
vs
Number of players
6800+ protoss
0%
6800+ zerg
40%
5000 - 6800 protoss
45%
5000 - 6800 zerg
23%
Conclusion: protoss is overpowered at winning when 5000 < mmr < 6800.
This is an absurd argument. The proper way to interpret this data is that no change in performance is observed in the protoss group. Change in performance is observed in the skill group. Therefore skill is the cause and protoss is not. Likewise, no change is observed in the cake group but change is observed in the “took blow to the head group”, therefore cake is not the cause of the headaches but being hit on the head is.
To measure protoss performance, you measure “true or false, is this player a protoss” and no other variable is relevant. If protoss is the cause, then the performance will change between “true, this is a protoss” and “false, this is not a protoss”.
If you did that, you’d see protoss has an ~300 mmr advantage. They push zergs down 150 and move up 150 mmr themselves. Why don’t they win premiers? Because 150 mmr isn’t going to bridge the gap between 6600 Hero and 7200 Serral. Premier tournaments are caused by skill; ladder is caused by balance.
You can take out the numbers and just use intuition here. You play the game enough. Those Zealot rushes and Cannon rushes you are talking about in the other thread are probably too strong at the hobbyist level. That’s it.
The problem with alot of your analysis is that you are correlating skill with mmr. If a player is using unbalanced strategies to gain mmr, then you can’t use mmr to judge their skill level. That’s the absurd argument you are providing. You are trying to justify using low skill strategies to climb the ladder as more skilled, due to your mmr that you gained using those low skill strategies. It’s ridiculous.
It’s one of the challenges to using mmr to measure balance, yes. But it’s more to point out the way correlations are identified is by grouping items into bins and seeing how the variable changes depending on the grouping. If you observe a change in grouping A but not grouping B, then grouping A is the cause.
When it comes to isolating the impact of balance and skill on a player’s mmr ranking there are a number of methods that can be used. For example, if protoss has an advantage in PvZ then we’d expect him to have high PvZ win-rates and lower PvT and PvP win-rates. In fact, we’d expect the difference to be split between these matchups according to how frequently they are played.
So if you have a 10% balance advantage in PvZ (50% → 60%) then we’d expect a 50% → 45% reduction of both PvT and PvP if the same number of games are played in each. If there are a different number of games, it becomes the ratio. So if there are 100 PvP and 200 PvT, then we’d expect a 50% → 47% winrate reduction in PvP and a 50% → 43% in PvT.
That’s because if you play 100 games and win 60 of them vs zerg, the mmr system tries to equalize this to 50/50, and so adding 100 more games vs terran at 45% and another at 100 games vs protoss at 45% brings the total win-rate to (60+45+45)/3 = 50%. Due to behavior of the matchmaker, if a balance advantage exists in a matchup then the win-rates will display this behavior. The problem is that there is a lot of randomness and noise, and you won’t notice it unless you average for a thousand or so players.
The bridge between KR and everyone else closed pretty handily over the last few years of LOTV. It really started with Serral though.
So to me the question shouldn’t be about balancing for KR or the world, it’s whether we balance around statistical outliers or reasonably good players in Diamond/Masters/GM?
And if we do balance around one or the other, then we need to have that philosophy of balance applied equally to all races…which is part of what makes this Balance Council absolutely suck. They balance Protoss around ‘low’ level ladder while balancing Terran and Zerg around pro play.
Overall this is what makes the balance council suck. They don’t apply their own internal philosophies evenly to each race which in turn creates dumb crap like the observer nerf.
That’s why companies pay me the big bucks and no one else. It’s because the ability to understand complex systems is rare. You can see that here as people struggle to understand what a correlation is. They can’t separate player skill from balance and equivocate the two in ways that violate the correlations. They say balance is the cause of a performance difference for a sample of maybe 10 people when if you calculate the correlations it would correspond with player and not race, meaning the correlation is race-invariant. Translation, zero correlation with race. Then they explain performance differences at a GM level with skill, even though the definition of GM is to enforce a uniform skill level which means skill is invariant in their sample. If you calculate a correlation between skill and performance in a a sample that is skill invariant your correlation will be zero. How do you explain a change in performance with a correlation of zero? It’s all absolute nonsense. Meanwhile you can actually get a good correlation between skill and performance in the top 10, and a good correlation between race and performance in GM. They simply don’t understand the very basics of correlations.
Also, you are on a public board. Anyone who follows the rules gets to post here. If you want to exclude people you personally don’t like, I suggest you use discord.
Lastly, whining that I am wrong does not in fact make me wrong. The way to show that I am wrong is by attacking my arguments, and not me. Your obsession with attacking me is complete and utter proof you can’t compete with my arguments. A common mode of attack is to look for assumptions and ask if the assumptions are justified and how it changes the logic if not. You can do it! I believe in you!
The reason you don’t try is because my arguments are iron clad. They are correct. They just find a conclusion you don’t like, and you for some reason aren’t allowed to think something is true unless you like it. That’s the vibe I get from your behavior. That’s not how science works. Science works independent of emotion and you accept the conclusion no matter what the conclusion is.
The fact of the matter is that pro play is not relevant to balance. I can quite literally quote the textbook on this one:
It’s funny how people on sc2 social media become so upset by this fact. The player base themselves would benefit from a game designed for them. The only people who stand to lose something if these facts get out are the esports people. Why? Because they want to rig game outcomes like the entertainment companies do in the US do. The competitions aren’t real and they pick the winner based on what’s best for their viewership numbers. This happens everywhere from football to cooking competitions. Aka totally staged.
That’s what they want to do with sc2 esports and it’s going to kill the game in the process but they don’t care as long as they make a buck. And that’s why people like you are upset that I share information like this. It makes people aware of the fact that they want to rig esports to produce preset outcomes.
You work for the esports industry somehow and I don’t know how but your sensitivity tells me I am on the right track.