Official State of the Game Podcast Thread - Page 949

Stop linking that fucking stupid chart. You have no fucking clue how to read or interpret it and have frequently spouted objectively incorrect bullshit in just this thread alone. Several pages ago you couldn't even read off the X axis and stated that Zergs were beating Protosses for the past two months even though the tipping point was around 5 weeks ago. You're going to statistician hell for your crimes against numbers. You don't make any case or even state a reasonable tolerance for what difference in win rate would still constitute balance and don't even mention that the at its worst Protoss were 18.8% more likely to win any given match against a Zerg and at Zergs best they are only 2.8% more likely to win, and even that is an entirely unsupportable claim because the real figures are entirely obscured by the smoothing function. I assure you, people aren't not responding to it because they don't understand it or don't want to face the implications of it, the lack of understanding is on your part.

--------------

As for Day9's argument, ultimately it isn't strictly wrong, but it's also completely fucking obtuse and worthless in context of any useful defining of the goal of balancing an E-Sport. I love you Day9 but you're being really silly.

Ultimately you could argue whether or not Chess is a balanced game, if I recall correctly given how much we have determined(chess still is not a solved game) White has a slight advantage. You can claim all you want that because not every permutation of the game has been played out that we cannot know that White has an advantage, that doesn't mean that if you make two equally skilled players play against each other with one always playing white and the other always black that the former is not going to win more of the time(i.e. that he has an advantage) such that it would be fair to play any other way than to alternate colors in each set.

For anyone who knows anything of learning algorithms in machine learning, there are two phases of any learning algorithm, exploration and exploitation, which are pretty much what they sound like, exploration being the activity of searching out new strategies, usually a task with uncertainty involved. Then there's exploitation which is the act of executing an established strategy and while there is zero potential for discovering a new strategy, the probabilities of success and it's overall feasibility are much more well known.

If we have 3 races and large enough sample size of players from each race we can assume the qualities of each skill attribute for each race to be reasonably similar. Therefore if we can clearly see that one race is spending more time exploring their potential options trying to expand their options to find a successful set of strategies to exploit to gain wins, and another is spending much less time exploring new strategies and exploiting a long since established set of strategies that are yielding greater win rates, that given enough time you can reasonably infer an imbalance in the match up. Keep in mind that subjective does not mean worthless, wrong or inaccurate. The validity of such an inference is based on the amount of time spent exploring, the relative results between the races and the sample size of each race.

Day9 wants to claim that the point at which we can consider a sufficient amount of time to have passed to still be way out in the future and Idra wants to suggest that that point has since come and gone. We can take Day9's view if you want and even extend it out towards infinity if you like, but it's completely fucking worthless as a yardstick, even if Idra's is subjective. But the key point here is that Day9 is not taking into account that there are in fact subjective models that give us a reasonable picture of balance and that we can in fact use numbers to build common models that we can discuss. If we were to take Day9's argument at its word, alpha roaches would still be in the game and protoss and terran would have no grounds to complain or cry imbalance because they haven't explored enough options yet for us to truly know if there is really an unfair advantage to be inferred.

You can either take Day9's view and completely shut off all discussion, but you're effectively arguing for the alpha roach if you do and the alternative is actually not to take Idra's side and agree that the game is imbalanced, but simply to engage in real discussion. To do that in a useful way means basically two things.

1.) Determining how much of a skew is considered imbalance. Not just by a percentage amount though, but also by how long the skew is present and it's magnitude as well as taking into account subjective measures such as accounting for the introduction of a new strategy or significant shift in the meta game and what is a reasonable amount of time to allow for exploration of new strategies.

2.) Coming up with reasonable models in which we can actually discuss mechanics and suggest why specific unit interactions may or may not be imbalanced in their own contexts. Yes they are subjective, no they are not perfect but the models that we as players use to determine balance are the same ones we used to formulate new strategies and do in fact have some basis in logic and practical application that we can go on.

----------

A lot of TLers seem to think not touching anything and waiting half a decade for some new strategy to evolve that balances out a seemingly perpetually broken match up is the holy grail of game balance. It's really not, and it's not like that really happened in Brood War, Blizzard released fucktons of balance updates as well as an expansion and the game was always going to be rebalanced after the patches stopped to some degree by virtue of new maps of which the designs are guided by their impact on win percentages, so not even the metagame developed in complete isolation. It is in no way an argument to cease balance updates and again would have been an argument for keeping the alpha roach, of course that's not what those TLers would want because the point at which updating should cease is incedibly fucking arbitrary. Believe it or not there is in fact likely a tipping point at which you can infer the likely hood of a particular strategy being overcome and determine an acceptable set of thresholds after which you take over and apply changes to restore equilibrium and maintain a competitive game. If you want to argue against constant patching a la Brood War you have to accept that there was in fact justified patching to begin at which point for your argument to hold up you need some reasonable metric as to determine when to actually stop patching, or as I said before, you have to accept the alpha roach.

In all honesty a lot of players and fans who played and watched Brood War have a lot of bad habits and really bad inferences from it that make absolutely no sense because they were borne of a situation where most of the options available now either didn't exist or were in their infancy when brood war came out and that somehow because of that it elevates itself above them. It's really fucking stupid when you think about it, especially when you consider that they are working from a dataset of one and the fact that no one has any idea how Brood War would have turned out if it had the same kind of balance philosophy behind as sc2 has now or as much attention from Blizzard due to their awareness of the significance of its E Sports potential. There's nothing sacred or special about it.



Overlords are armored, so it's 14 damage every 1.44 seconds, making roughly 10 dps not 7.