|
On September 02 2014 08:34 Ghanburighan wrote:
Thanks for the post, Hider. But could you elaborate on `The adjustments takes into account distribution, which means that if there is an excessive amount of race X, the win/rates of the race are adjusted downwards'? I'd like to know the formula for the adjustment downwards.
So the theoretically correct approach is to make an assumption of the distribution and then calculate expected win/rates based on what percentage of the total playerbase we are looking at. The issues with this approach are twofold:
(1) We do not have any data on the percentage of the total playerbase which is being reported on Aliguac database
(2) Even if we make an assumption, then this assumption will be skewed as there is more games reported by top 0.001% players than of 0.01% players.
So this approach is likely to be flawed. So rather I simply opted for a different approach where I weighted the importance of distribution of players in order to replicate how history is general perceived. Thus, the graph shows the imbalance of the patch-zerg era. The imbalance in terran of the first year of WOL and the imbalance of WOL TvP.
It's not attempting to come up with exact numbers of the balance (because the methodlogy is too unprecise for that), however, it's an unbiased approach to showing the developments over time. Thus, we don't have to go around making subjective discussion with bias in order to see whether race X has benefited from a buff.
Regardless I also think this methodlogy is still precise enough to show that terrans never needed a nerf in TvZ late 2013. And Blizzard should have reacted much quicker when balance kept getting gradually worse. Unlike Blizzards ladder statistics, this statistics can actually show whne a shift in balance occurs, and unlike Aliguacs naked win/rates, this shows whether an increase in win/rates are due to an improvement in balance or whether it's merely increasing the imbalance that already existed.
|
The way Terrans are being more represented seems to convey that Terrans are finally being able to fight against Zerg variety or Protoss unit compositional strength. It would still be nice to see more variety of units and options for terrans to use. The reaper is a nice opening, but it is only for the early game. I'd like to see the reaper used more for the mid game. The banshee, i don't know if this unit should be cloaked. because the tradeoff of cloaking is low HP. so you always see banshees getting crushed mid to late game.
for Zerg, I'd like to the hyrdra slightly buffed, maybe the nydus should be reduced in price about 10%. The immortal seems awfully strong but wait and see. Of course i'd like to see the ghost at least somewhat useful against zergs which would add another layer of depth and strategy for zerg vs Terran matchups.
i think we should wait and see how the game evolves.
It's funny every sport evolves. Basketball with their spacing and bigs shooting long range. Football with their spread offence. Baseball with bigger and bigger pitchers that throw 95+. Hockey, bigger and bigger goalies, scoring is still too stingy compared to the 80's. And Soccer, i just don't understand why they don't evolve the game to make it more fan friendly. SC2 will too. Jangbang was the final top level player to take broodwar to another level for protoss. Someone will take SC2 to another level. hopefully Blizzard fosters that evolution instead of stifling it.
I'd also like to see bigger maps like in BW for more challenging and higher level tournments like Blizzcon, Star-leagues too.
|
On September 02 2014 08:59 Nebuchad wrote:Show nested quote +On September 02 2014 08:26 Hider wrote:
An ignored element in this period was that protoss actually was a bit favored vs terran as well. Given the fact that more terrans were eliminated from tournaments than protossses, we would expect that win/rates of terrans should be over 50% in 2012 vs protoss. Since win/rates were slightly below 50% instead, terran was actually underperforming during this period. Your whole post is interesting but this is particularly interesting to me because I wouldn't think that's true, my memory of TvP at the end of WoL was that it was pretty good for terran in general, with ghost viking performing really well against the lategame toss army (I remember surprised reactions when it turned out at the start of HotS that protoss could handle ghost viking really well, if not even too well). Did you take into account that there weren't a ton of protosses in GSL at the end of WoL? 2 out of the first 20 qualified players were protoss, at some point. So even though terrans lost more badly to zergs, I'm not sure protosses qualified a ton more (I don't remember the rest of the events well enough, but I also remember that some of the good results of protoss came from the wcs event, which held qualifiers at an earlier time when protoss was doing really well in Korea).
So this doens't look at GSL only, but all Aliguac games. Thus, it bases balance ranging from competitive play and upwards. Under the assumption that the average foreign terran player is as good as the average foreign protoss player, terran should have had better win/rates than protos in 2012 - but they didn't.
The reason you probably remember this period differently is that most games probably were quite close. The underdistribution of terran players wasn't a known story, rather the story was "zerg was op" instead of "terran is UP and zerg is OP". Thus, it probably wasn't well-known that fewer foreign terran players were good enough to be considered competitive compared to the protoss counterparts But those who were, could still do decently vs the protoss opponents.
Also note, that this was a slow development. It wasn't just from one month to another that the imbalance occured, which made it even harder to spot.
|
On September 02 2014 09:12 Hider wrote:Show nested quote +On September 02 2014 08:59 Nebuchad wrote:On September 02 2014 08:26 Hider wrote:
An ignored element in this period was that protoss actually was a bit favored vs terran as well. Given the fact that more terrans were eliminated from tournaments than protossses, we would expect that win/rates of terrans should be over 50% in 2012 vs protoss. Since win/rates were slightly below 50% instead, terran was actually underperforming during this period. Your whole post is interesting but this is particularly interesting to me because I wouldn't think that's true, my memory of TvP at the end of WoL was that it was pretty good for terran in general, with ghost viking performing really well against the lategame toss army (I remember surprised reactions when it turned out at the start of HotS that protoss could handle ghost viking really well, if not even too well). Did you take into account that there weren't a ton of protosses in GSL at the end of WoL? 2 out of the first 20 qualified players were protoss, at some point. So even though terrans lost more badly to zergs, I'm not sure protosses qualified a ton more (I don't remember the rest of the events well enough, but I also remember that some of the good results of protoss came from the wcs event, which held qualifiers at an earlier time when protoss was doing really well in Korea). So this doens't look at GSL only, but all Aliguac games. Thus, it bases balance ranging from competitive play and upwards. I think its important to remember that only korean terrans performed in 2012, but under the assumption that the average foreign terran player is as good as the average foreign protoss player, terran should have had better win/rates than protos in 2012 - but they didn't. The reason you probably remember this period differently (besides the GSL-effect) is that most games probably were quite close. The underdistribution of terran players wasn't a known story, rather the story was "zerg was op" instead of "terran is UP and zerg is OP". Thus, it probably wasn't a very dicussed thing (untill HOTS) how many fewer foreign terran players were good enough to be considered competitive.
Yeah I understand that you took all of aligulac, it feels weird that there's such a gap between GSL (where there isn't a single round of code S where there are more protosses than terrans from 2012 season 4 to end of WoL) and the total of stats. I didn't think foreign tournaments could have that much influence, especially since protoss also got rolled at IGN which was the only other big tournament I could remember from the period
Btw, considering protoss had better winrates vs zerg than terran, how come it was the number of protosses which dropped from code S and not the number of terrans? That also seems kind of weird, right?
|
Thoughts on how terran went from overrepresented (2010) to underpresented (post 2012)
- In early 2010, everyone accepted terran as being OP --> thus they were nerfed.
- Terran was the most micro-intensive race, and thus when the race was repeaditely nerfed, foreigners suffered more than koreans.
- Untill Queen buff, game was probably roughly balanced in the sense, that terran TvZ was slightly T favored in Korea, and slightly Z favored in the foreign scene. Blizzard however had more emphasize on balancing Korean scene, and thus terran was further nerfed in the match-up.
- Blizzards ladder win/rates remained relatively close to 50/50 (as they always do) and early on Koreans terrans still did okay'ish, so blizzard didn't notice a huge imbalance shift. But using Aliguac statistics, balance looked pretty Z favored early/mid 2012. .
- As Zergs got better, they started dominating terrans also in korean scene. In the foreign scene, toss performed somewhat better due to their timing attacks, which meant they could maintain a higher amount of competitive players. In Korea TvP was probably reasonably balanced, but postEMP nerf, terrans also struggled alot, but this was hidden by the patch-zerg story. Due to the history of korean terrans always being able to adjust their play after patches have occured, Blizzard didn't act too quickly too the signs of Infestor-imbalance. So when MVP won against random foreign zergs in Summer 2012 using Ravens, they delayed a needed patch. This caused a further reduction in the terran-representation.
- Going into HOTS; terran foreign scene was roughly as UP as zerg was in 2010. What was required was an adjustment period of 5-6 months where terran win/rates would be above 50% in order to get the distribution right. In order to get win/rates above 50/50, terrans would need buffs in both matchups.
They got one vs zerg, and early on it looked like they got one vs toss, and thus for 2-3 months, improvemenets were underway. But as protoss figured out TvP, it became obvious toss actually got the better end of HOTS and as a consequence, we saw a further reduction of terran players. Eventually it was as bad as under the patch-zerg era.
For a long time, the story was "toss imba" in early 2014, but just like in WOL, the story of terrans struggling vs Zerg was also undertold. However, eventually something changed here - Koreans also started to struggle in TvZ (unlike what they did in TvP WOL). Overall, with HOTS we have started to see balance being more symmetrical with foreign and korean scene, which I believe can be attributed to three factors:
(1) WCS putting Koreans into EU ladder
(2) NA ladder dying --> All good NAs play on EU ladder which means EU ladder is much stronger than ever
3) An equal increase in the absolute skill level of all players which had reduced the relative skilldifference in percentage between Koreans and Foreigners.
Thus, after Terrans figured out blink all-ins, it became obvious that there were problems in TvZ as well, and Blizzard has patched the game - although a little late since Widow Mines never should have been nerfed in the first place. The buff to the Thor was probably also needed since it's likely that terran would have suffered even if the Widow Mine had not been nerfed. That said, terran may have too many early game options in the Banshee and Hellbat now, which can create volatility in the early game.
Sum up
- Foreign terrans struggled due to skill-cap assymetry and Blizzard balanced the game around the Korean scene.
- Blizzard was too late with fixing Infestor-imbalance --> Consistent reduction in terran representation. Meanwhile toss didn't fall as much due to Immortal-allins + being favored vs terran (foreign scene)
- Early HOTS, Blizzard didn't properly understand the adjustment-proces where terran win/rates had to go above 50% in both matchups in order to readjust the representation. Thus, when toss started getting above 55%+ win/rates vs terran, they should have reacted a ton quicker.
- Over the next couple of months, terrans will need 50%+ win/rates in both matchups untill distribution assymetry has been reduced. Given that distribution % increased by 2% in August, it will take roughly 3-4 more months at the same pace to reestablish it.
- Adjusted win/rates are likely to be closer to 50/50 the future due to a reduction in the relative skill-difference between Koreans and foreigners
|
On September 02 2014 09:29 Hider wrote:
- As Zergs got better they started domaiting terrans also in korean scene. Toss performed somewhat better as they had the soul-train build, which meant they could maintain a higher amount of competitive players.
But that's the thing, they didn't. That's why the results are so weird to me.
|
On September 02 2014 09:34 Nebuchad wrote:Show nested quote +On September 02 2014 09:29 Hider wrote:
- As Zergs got better they started domaiting terrans also in korean scene. Toss performed somewhat better as they had the soul-train build, which meant they could maintain a higher amount of competitive players.
But that's the thing, they didn't. That's why the results are so weird to me.
So they didn't perform that well becasue they still lost to zergs (not as badly as terrasn though). However, they were still doing better vs terrans. Why else would there suddenly be around 68% protosses, 60% terrans and 71% zergs in late 2012? On year prior to that it was like 68% terrans, 63% protosses and 66% zergs (non-mirror matchups). Why would terran reduce so much, while protoss would increase a bit? When you take into account that EMP was nerfed, doesn't it make sense that foreign terrans suffered a lot after that?
Thus this indicatoes - that at least when you include competitive foreign scene - that only the very best terrans were left against slightly lesser skilled toss's and even lesser skilled zergs. So when a terran was going 50-50 vs a toss, it was actually a sign of a small imbalance, but given how the game played out, you might actually be tempted to think that the matchup is balanced since you wouldn't neccesarily be aware that this terran player was of a higher caliber than his protoss counterpart.
Eventually, zergs benefited even more, and thus the increase in results of foreign toss's were hidden by the even larger increase in results of foreign zergs.
Going into HOTS, noone was really aware of this, and thus terran took undeservely big nerfs in the TvP matchups.
|
On September 02 2014 09:43 Hider wrote:Show nested quote +On September 02 2014 09:34 Nebuchad wrote:On September 02 2014 09:29 Hider wrote:
- As Zergs got better they started domaiting terrans also in korean scene. Toss performed somewhat better as they had the soul-train build, which meant they could maintain a higher amount of competitive players.
But that's the thing, they didn't. That's why the results are so weird to me. So they didn't perform that well becasue they still lost to zergs (not as badly as terrasn though). However, they were still doing better vs terrans. Why else would there suddenly be around 68% protosses, 60% terrans and 71% zergs in late 2012? On year prior to that it was like 68% terrans, 63% protosses and 66% zergs (non-mirror matchups). Why would terran reduce so much, while protoss would increase a bit? When you take into account that EMP was nerfed, doesn't it make sense that foreign terrans suffered a lot after that?
I meant that they didn't in the korean scene, as was stated in your quote, I should have repeated it for clarity I apologize. The global situation didn't correlate at all with the korean scene, where terran got 136-93 in TvP on the last three seasons (the zerg era), got to keep around 13 players in code S while protoss went down from 10 to 5, and wasn't eliminated faster than protoss at any point.
|
On September 02 2014 09:55 Nebuchad wrote:Show nested quote +On September 02 2014 09:43 Hider wrote:On September 02 2014 09:34 Nebuchad wrote:On September 02 2014 09:29 Hider wrote:
- As Zergs got better they started domaiting terrans also in korean scene. Toss performed somewhat better as they had the soul-train build, which meant they could maintain a higher amount of competitive players.
But that's the thing, they didn't. That's why the results are so weird to me. So they didn't perform that well becasue they still lost to zergs (not as badly as terrasn though). However, they were still doing better vs terrans. Why else would there suddenly be around 68% protosses, 60% terrans and 71% zergs in late 2012? On year prior to that it was like 68% terrans, 63% protosses and 66% zergs (non-mirror matchups). Why would terran reduce so much, while protoss would increase a bit? When you take into account that EMP was nerfed, doesn't it make sense that foreign terrans suffered a lot after that? I meant that they didn't in the korean scene, as was stated in your quote, I should have repeated it for clarity I apologize. The global situation didn't correlate at all with the korean scene, where terran got 136-93 in TvP on the last three seasons (the zerg era), got to keep around 13 players in code S while protoss went down from 10 to 5, and wasn't eliminated faster than protoss at any point.
Well I wasn't being precise in my original quote. What I meant was that zergs did better in the korean scene, and when I talked about protosses, I spoke generally.
Anyway, I realized that given these statistics, you can actually quantify the skill-cap difference (which means that one race requires more skills than another race). If korean win/rates are high, but aliguac win/rates aren't, that points to assymetry in the skill-cap. Might be interesting to see how that has developed over time. My theory - as stated - is that we have seen a huge reduction in the skillcap over the last year.
|
On September 02 2014 09:59 Hider wrote:Show nested quote +On September 02 2014 09:55 Nebuchad wrote:On September 02 2014 09:43 Hider wrote:On September 02 2014 09:34 Nebuchad wrote:On September 02 2014 09:29 Hider wrote:
- As Zergs got better they started domaiting terrans also in korean scene. Toss performed somewhat better as they had the soul-train build, which meant they could maintain a higher amount of competitive players.
But that's the thing, they didn't. That's why the results are so weird to me. So they didn't perform that well becasue they still lost to zergs (not as badly as terrasn though). However, they were still doing better vs terrans. Why else would there suddenly be around 68% protosses, 60% terrans and 71% zergs in late 2012? On year prior to that it was like 68% terrans, 63% protosses and 66% zergs (non-mirror matchups). Why would terran reduce so much, while protoss would increase a bit? When you take into account that EMP was nerfed, doesn't it make sense that foreign terrans suffered a lot after that? I meant that they didn't in the korean scene, as was stated in your quote, I should have repeated it for clarity I apologize. The global situation didn't correlate at all with the korean scene, where terran got 136-93 in TvP on the last three seasons (the zerg era), got to keep around 13 players in code S while protoss went down from 10 to 5, and wasn't eliminated faster than protoss at any point. Well I wasn't being precise in my original quote. What I meant was that zergs did better in the korean scene, and when I talked about protosses, I spoke generally. Anyway, I realized that given these statistics, you can actually quantify the skill-cap difference (which means that one race requires more skills than another race). If korean win/rates are high, but aliguac win/rates aren't, that points to assymetry in the skill-cap. Might be interesting to see how that has developed over time. My theory - as stated - is that we have seen a huge reduction in the skillcap over the last year.
Can we still talk about imbalance then? Your argument is that because the population of terran is smaller, terrans would play against weaker protosses on average, and should therefore have a higher winrate than 50%. But now we've established that the best terrans did have a good winrate in the hardest tournaments, against the best of the protosses. It seems to point to TvP being harder rather than PvT being overpowered.
|
On August 29 2014 03:52 Thieving Magpie wrote:Show nested quote +On August 29 2014 03:33 Grumbels wrote:On August 29 2014 00:49 Thieving Magpie wrote:On August 29 2014 00:43 sibs wrote: When zerg and protoss was the main race winning GSL in 2010 => 2011 was that a sign of underpowered terrans? Was it a sign that terran sucked when only MKP, MVP, and MMA would consistently do well amongst terran players only to have MC/Nestea/Fruitdealer get all the trophies? No, because making arbitrary population cross cuts proves nothing. This is some insane SC2 history revisionism, Terran was insanely strong, there was Terran everywhere, Zerg was carried by Nestea/FD, Protoss was carried solely by the boss toss MC. http://aligulac.com/misc/balance/You can also check individual periods. Which is why I kept saying making arbitrary cut offs in population size is ludicrous. GSL winners in 2010 was Zerg Zerg Protoss. GSL finalists in early 2011 were majority Protoss/Zerg. Making arbitrary population cut offs will always make the game look imbalanced depending on which population you wish to focus on. For example, most American prisoners are non-white, does that mean that whites are less prone to crime? Assuming a fair justice system: Yes. (in actuality the system is biased/racist i.e. "imbalanced") Most women are paid less than men in America, does that mean all Americans are misogynist? Assuming men are better than women: Not necessarily. (yet here again the system is actually sexist/imbalanced) Really, the numbers can tell you that there is something wrong. They do in your examples, each forces you to have to accept something you might not want to, that either there is something wrong with the system or with the people in the system. Either terran players are worse or terran is weak. Pick your poison. Which is my point. The population shows a symptom of something--but it's not proof of any one thing. It takes specific tests to prove imbalance. Low Terran representation is merely a random data set that, with other evidence, can lead to something actually truthful. But to point at representation and automatically conclude it's something is foolish. we're not automatically concluding that. It's just one part of the puzzle or equation we are trying to figure out. you second sentence is true. Correlation may or may not lead to causation. Low terran representation can mean many things. We are trying to narrow down what the low terran representation means. looking at unit attributes is probably the best method of analyzing cost/benefit of each unit and measure effectiveness race balance.
|
On September 02 2014 10:12 Nebuchad wrote:Show nested quote +On September 02 2014 09:59 Hider wrote:On September 02 2014 09:55 Nebuchad wrote:On September 02 2014 09:43 Hider wrote:On September 02 2014 09:34 Nebuchad wrote:On September 02 2014 09:29 Hider wrote:
- As Zergs got better they started domaiting terrans also in korean scene. Toss performed somewhat better as they had the soul-train build, which meant they could maintain a higher amount of competitive players.
But that's the thing, they didn't. That's why the results are so weird to me. So they didn't perform that well becasue they still lost to zergs (not as badly as terrasn though). However, they were still doing better vs terrans. Why else would there suddenly be around 68% protosses, 60% terrans and 71% zergs in late 2012? On year prior to that it was like 68% terrans, 63% protosses and 66% zergs (non-mirror matchups). Why would terran reduce so much, while protoss would increase a bit? When you take into account that EMP was nerfed, doesn't it make sense that foreign terrans suffered a lot after that? I meant that they didn't in the korean scene, as was stated in your quote, I should have repeated it for clarity I apologize. The global situation didn't correlate at all with the korean scene, where terran got 136-93 in TvP on the last three seasons (the zerg era), got to keep around 13 players in code S while protoss went down from 10 to 5, and wasn't eliminated faster than protoss at any point. Well I wasn't being precise in my original quote. What I meant was that zergs did better in the korean scene, and when I talked about protosses, I spoke generally. Anyway, I realized that given these statistics, you can actually quantify the skill-cap difference (which means that one race requires more skills than another race). If korean win/rates are high, but aliguac win/rates aren't, that points to assymetry in the skill-cap. Might be interesting to see how that has developed over time. My theory - as stated - is that we have seen a huge reduction in the skillcap over the last year. Can we still talk about imbalance then? Your argument is that because the population of terran is smaller, terrans would play against weaker protosses on average, and should therefore have a higher winrate than 50%. But now we've established that the best terrans did have a good winrate in the hardest tournaments, against the best of the protosses. It seems to point to TvP being harder rather than PvT being overpowered.
Depends on your definition. Some people say only measure balance for the very best, but what does that even mean?. You could also just compare the best 2 terran against the best 2 protosses at any given point in time. What if the best two protosses wins against the best two terrans, does that say anything about balance?
You gotta measure balance using some larger sort of data, and the issue with only looking at GSL is that sample size is way too low on a month-per-month basis. Thus you can only asses balance over multiple months, which makes it apretty bad appraoch to measure differences in balance. This has been further worsened over the last years due to the reduction in games played in GSL/PL. Further, if you wanna adjust for differences in the distribution, that adjustments suffers extremely much from volatility as we can see a 100% increase from one season to another, and to be fair, whether there are 4 terrans in GSL or 7, doesn't change expected win/rates that much. Instead, it seems to be a much more reliable indicator of a balance shift if we can see that the percentage of games played by one race has dropped by 6-8 % over 3-4 months without an offsetting increase in win/rates.
So yes when the top 1000th terran players losses to the top1000th protoss players, it's an indicaiton of an imbalance. That said, there can be different types of imbalances at different levels and I believe multiple sources of data must be combined in order to get a full understanding of the issue.
Going into HOTS, there should obviously have been a focus on increasing the skill-cap of protoss and Zerg in order to improve overall balance.
|
On September 02 2014 10:19 Hider wrote:Show nested quote +On September 02 2014 10:12 Nebuchad wrote:On September 02 2014 09:59 Hider wrote:On September 02 2014 09:55 Nebuchad wrote:On September 02 2014 09:43 Hider wrote:On September 02 2014 09:34 Nebuchad wrote:On September 02 2014 09:29 Hider wrote:
- As Zergs got better they started domaiting terrans also in korean scene. Toss performed somewhat better as they had the soul-train build, which meant they could maintain a higher amount of competitive players.
But that's the thing, they didn't. That's why the results are so weird to me. So they didn't perform that well becasue they still lost to zergs (not as badly as terrasn though). However, they were still doing better vs terrans. Why else would there suddenly be around 68% protosses, 60% terrans and 71% zergs in late 2012? On year prior to that it was like 68% terrans, 63% protosses and 66% zergs (non-mirror matchups). Why would terran reduce so much, while protoss would increase a bit? When you take into account that EMP was nerfed, doesn't it make sense that foreign terrans suffered a lot after that? I meant that they didn't in the korean scene, as was stated in your quote, I should have repeated it for clarity I apologize. The global situation didn't correlate at all with the korean scene, where terran got 136-93 in TvP on the last three seasons (the zerg era), got to keep around 13 players in code S while protoss went down from 10 to 5, and wasn't eliminated faster than protoss at any point. Well I wasn't being precise in my original quote. What I meant was that zergs did better in the korean scene, and when I talked about protosses, I spoke generally. Anyway, I realized that given these statistics, you can actually quantify the skill-cap difference (which means that one race requires more skills than another race). If korean win/rates are high, but aliguac win/rates aren't, that points to assymetry in the skill-cap. Might be interesting to see how that has developed over time. My theory - as stated - is that we have seen a huge reduction in the skillcap over the last year. Can we still talk about imbalance then? Your argument is that because the population of terran is smaller, terrans would play against weaker protosses on average, and should therefore have a higher winrate than 50%. But now we've established that the best terrans did have a good winrate in the hardest tournaments, against the best of the protosses. It seems to point to TvP being harder rather than PvT being overpowered. Depends on what metrics you are using. You could also just compare the best 2 terran against the best 2 protosses at any given point in time. What if the best two protosses wins against the best two terrans, does that say anything about balance? You gotta measure balance using some larger sort of data, and the issue with only looking at GSL is that sample size is way too low on a month-per-month basis. Thus you can only asses balance over multiple months, which makes it apretty bad appraoch to measure differences in balance. This has been further worsened over the last years due to the reduction in games played in GSL/PL. So yes when the top 1000th terran players losses to the top1000th protoss players, it's an indicaiton of an imbalance. That said, there can be different types of imbalances at different levels and I believe multiple sources of data must be combined to get a full understanding of the issue.
You seem to have changed your mind on this issue, last time we discussed this we ended up agreeing on that:
On July 03 2014 06:56 Hider wrote:
Well my personal opinion is that we with almost certainty can conclude that there is a balance issue if these two conditions are met;
1) Over an extended period (3+months) a race is signifciantly lagging behind the other races in Aliguac statistics (where we look at a combo of representation and win/rates)
2) If the results among the best players in the world are poor for the race as well
If the former requirement is met, but the latter isn't, then I would argue that we are seeing some type of skill ceiling issue where race X is balanced at the highest level but a bit too weak/harder to play at lower levels.
I liked your position better back then.
|
I liked your position better back then.
Matter of definition. They are not mutually exclusive as skill-cap issue is an imbalance as well. When you have one race which is easier and another race which is much harder to play, I believe imbalance fits that category as well... Avilo
That said, check out what I wrote in the post:
That said, there can be different types of imbalances at different levels and I believe multiple sources of data must be combined to get a full understanding of the issue.
Thus, clearly I still believe that it's important to gather as much data as possible to get the best assesment of the balancee-situation. But you have to look at the context of the quote. Here the context is that I suggest an unbiased approach of measuring a shift in balance. If you wanna use GSL-data, then it IMO needs to be combined with subjective analysis due to the low sample size. Given that skillcap assymetry has been reduced in recent years (it's my theory at least), only relying on Aliguac data makes more sense going forward since it's unlikely that there will be a huge difference in results between Aliguac data and GSL data, but the former has much higher sample sizes.
The context I had back then was probably something where I wanted to go into an more indepth depht with subjective analysis since I played the game actively back then. I actually don't think I ever suggested to use GSL today for quantiative purposes, but rather I probably suggested to analyze the games on qualitative basis. That requires you play the game at a pretty high level as well, and today, I no longer have time, and thus I don't feel I am in a postion to talk about balance from a subjective analytical POV. It's probably why you haven't seen in this forum the last month as I didn't really feel like I could contriubte with anything, untill today where August numbers were published.
Why did I talk about the PvT imbalnace in 2012? I did that in order to explain why terrans have a lower repesentation than the other races as some claim that it's just a structural thing. Like terran for some artificial reason is supposed to have fewer competitive players than their counterparts. But given that the race balance assymetry has been reduced significantly, that is not the case anymore.
It's important to understand why the data looks like it does today, in order to understand how to measure balance going forward.
|
On September 02 2014 10:30 Hider wrote:
Matter of definition. They are not mutually exclusive as skill-cap issue is an imbalance as well.
It is different in the way you treat them though. If terran is doing very well vs protoss at high level and getting rekt at slightly lower level, then the right change to make is to buff protoss slightly, in a way that makes the race a lot harder to play. If terran is underpowered, then you still have to make protoss harder to play, but you need to nerf it instead of buff it.
On September 02 2014 10:30 Hider wrote:
Why did I talk about the PvT imbalnace in 2012? In order to explain why terrans have a lower repesentation than its other races as some claim that it's just a strucutral thing.
But you don't need to show a PvT imbalance to do that. Terran was weak enough in TvZ and the state of TvZ was stagnant enough that you don't need necessarily need more justification for a lot of terrans quitting the game, regardless of how they performed vs protoss (especially since the majority of people prefer TvZ to TvP, and a lot of the people who quit were probably from a slightly lower level, where protoss was doing well vs terran). And then you add that terran has a higher skillcap, which I would assume influences your motivation to keep on, given the amount of effort you have to put in.
|
But you don't need PvT to do that. Terran was weak enough in TvZ and the state of TvZ was stagnant enough that you don't need necessarily need more justification for a lot of terrans quitting the game, regardless of how they performed vs protoss (especially since the majority of people prefer TvZ to TvP). And then you add that terran has a higher skillcap, which I would assume influences your motivation to keep on, given the amount of effort you have to put in.
FYI, I don't believe terrans are quitting the game. I believe the major effect is that they no longer can perform at a high enough level to be considered competitive, thus fewer of their matches show up on Aliguags database.
Btw, the data I suggested here shows no obvious skill-cap difference between Z and T (I guess that was more of qualitiatve comment on my behalf). But showing that protoss were beating terrans in the foreign scene helps explain why terran was in a much worse postiion going into HOTS than protoss was. And thus this shows that terran isn't always supposed to have fewer players than their toss/zerg counterparts. But that's only the case if all of the below 3 assumptions are met:
(1) Significant difference in skill-cap between the races
(2) Koreans are sigificantly better than foreigners (in percentages)
(3) Game is balanced around Koreans
Given that I think (2) isn't really that true any more, I believe the quality of Aliguac data has increased quite a lot. That said, it's ofcourse never gonne be bad to combine it with GSL data as long as your careful about not making preemptive consequences on too few games.
Anyway, I feel we are going a bit in a circle here - Good night.
|
Terran's doing incredibly overall on aligulac, 58% WR vs Z, 54.5% vs P, 4 out of the 5 top spots, obviously leading race at +6%.
|
On September 02 2014 10:49 Hider wrote:
FYI, I don't believe terrans are quitting the game. I believe the major effect is that they no longer can perform at a high enough level to be considered competitive, thus fewer of their matches show up on Aliguags database.
That sounds like something we could easily prove or disprove. Given how many games we add from early rounds of tournaments or qualifiers nowadays, I tend to think you're wrong there, but I can't be sure. I always wish we could add more.
On September 02 2014 10:49 Hider wrote:
But showing that protoss were beating terrans in the foreign scene helps explain why terran was in a much worse postiion going into HOTS than protoss was. And thus this shows that terran isn't always supposed to have fewer players than their toss/zerg counterparts.
A drop in players of a race points strongly towards the race doing badly. An increase in players of a race points strongly towards the race doing well. However, the number of players staying the same is not conclusive with regards to balance, and an equal number of players between two races does not show that balance has been reached. That's why you also need the winrates, and you can't just stop at the population alone (but I already know that you agree with that).
|
On September 02 2014 11:07 sibs wrote:
Terran's doing incredibly overall on aligulac, 58% WR vs Z, 54.5% vs P, 4 out of the 5 top spots, obviously leading race at +6%.
I'm actually surprised that P isn't the leading race. The calculation for leading race seems to always have the least intuitive result.
As such I wouldn't ever cite the leading race calculation as a balance metric.
|
The mine's radius of damage is silly. In BW or SC 2, not until now, with the mine, have I seen a unit where you can react instantly to seeing a unit at your mineral line, yet it doesn't matter. Doesn't matter if I have a cannon and I've only been focusing on that spot to begin with... they will get kills... You would think, at the least, that the mine would be an expensive unit, given this unique distinction...
If you don't make something sensible, then the game will denigrate to, say, hell bat drops all game, every game. It just seems like an artificial way to increase winrates, to me. It's basically saying, I guess Toss would win fewer games if they were forced to have a 10 worker reduction! While true... it makes for a pretty awkward game. There has to be a better way...
|
|
|
|
|
|
EPT
