|
On January 03 2016 06:54 Nebuchad wrote:Show nested quote +On January 03 2016 04:54 Lexender wrote:On January 03 2016 04:48 TheWinks wrote:On January 03 2016 03:48 BronzeKnee wrote:On January 03 2016 03:48 TheWinks wrote:
If we had blindly followed aligulac win rates the widow mine revert would have never happened. Aligulac's "balance" page has always been a poor source of information. Why would we blindly follow the win rates? I don't see anyone arguing for that. They are a source of information to be considered. But not ignored. Therefore we should neither follow them blindly, nor totally discard them. The fact that PvZ is at it's worst point in terms of win rates ever should be very concerning and is cause for some kind of change. Aligulac's win rates are worth ignoring. There's too much garbage included that obscures the end result. I remember when ByuN was being number 1 terran, in the middle of HotS We know how it works. When people play a lot of online tournaments, they end up advancing their rating way more (and way easier) than the people who play only GSL and the occasional week-end tournament, even though they are playing weaker opposition. You have demonstrated that the ranking is flawed, but that's completely different from collecting results and calculating a win percentage based on them. What is described as "garbage" is simply things that contradict the narrative you're pulling for.
The two things you're describing are linked. If a top or near top level Korean goes into an online tournament and wipes the floor with a bunch of foreigners he gets a bigger boost in aligulac rating that he really should AND all those games get added into the win rate graph despite their lower quality.
|
On January 03 2016 07:12 TheWinks wrote:Show nested quote +On January 03 2016 06:54 Nebuchad wrote:On January 03 2016 04:54 Lexender wrote:On January 03 2016 04:48 TheWinks wrote:On January 03 2016 03:48 BronzeKnee wrote:On January 03 2016 03:48 TheWinks wrote:
If we had blindly followed aligulac win rates the widow mine revert would have never happened. Aligulac's "balance" page has always been a poor source of information. Why would we blindly follow the win rates? I don't see anyone arguing for that. They are a source of information to be considered. But not ignored. Therefore we should neither follow them blindly, nor totally discard them. The fact that PvZ is at it's worst point in terms of win rates ever should be very concerning and is cause for some kind of change. Aligulac's win rates are worth ignoring. There's too much garbage included that obscures the end result. I remember when ByuN was being number 1 terran, in the middle of HotS We know how it works. When people play a lot of online tournaments, they end up advancing their rating way more (and way easier) than the people who play only GSL and the occasional week-end tournament, even though they are playing weaker opposition. You have demonstrated that the ranking is flawed, but that's completely different from collecting results and calculating a win percentage based on them. What is described as "garbage" is simply things that contradict the narrative you're pulling for. The two things you're describing are linked. If a top or near top level Korean goes into an online tournament and wipes the floor with a bunch of foreigners he gets a bigger boost in aligulac rating that he really should AND all those games get added into the win rate graph despite their lower quality.
and the underlying assumption unless proven differently (statistically or deductively, not by giving a piece of evidence) is:
such a player can be of any race, so the winrates get influenced by it in the same ways
also you gotta prove that the samplesize is small enough for it mattering to begin with. time for some math if you want to support your hypothesis
|
On January 03 2016 08:22 Big J wrote:Show nested quote +On January 03 2016 07:12 TheWinks wrote:On January 03 2016 06:54 Nebuchad wrote:On January 03 2016 04:54 Lexender wrote:On January 03 2016 04:48 TheWinks wrote:On January 03 2016 03:48 BronzeKnee wrote:On January 03 2016 03:48 TheWinks wrote:
If we had blindly followed aligulac win rates the widow mine revert would have never happened. Aligulac's "balance" page has always been a poor source of information. Why would we blindly follow the win rates? I don't see anyone arguing for that. They are a source of information to be considered. But not ignored. Therefore we should neither follow them blindly, nor totally discard them. The fact that PvZ is at it's worst point in terms of win rates ever should be very concerning and is cause for some kind of change. Aligulac's win rates are worth ignoring. There's too much garbage included that obscures the end result. I remember when ByuN was being number 1 terran, in the middle of HotS We know how it works. When people play a lot of online tournaments, they end up advancing their rating way more (and way easier) than the people who play only GSL and the occasional week-end tournament, even though they are playing weaker opposition. You have demonstrated that the ranking is flawed, but that's completely different from collecting results and calculating a win percentage based on them. What is described as "garbage" is simply things that contradict the narrative you're pulling for. The two things you're describing are linked. If a top or near top level Korean goes into an online tournament and wipes the floor with a bunch of foreigners he gets a bigger boost in aligulac rating that he really should AND all those games get added into the win rate graph despite their lower quality. and the underlying assumption unless proven differently (statistically or deductively, not by giving a piece of evidence) is: such a player can be of any race, so the winrates get influenced by it in the same ways also you gotta prove that the samplesize is small enough for it mattering to begin with. time for some math if you want to support your hypothesis
There is clearly not an equal race distribution just by looking at the balance report itself.
TvZ: 987
PvT: 635
PvZ: 1061
Why would we assume what you're saying? Even if low quality games did mostly cancel each other out, why include them? It's unnecessary noise. Also, your second sentence doesn't make any sense.
|
Now I don't feel so bad for having a 30% win rate in PvZ.
|
Lots of noise due to skill mismatch will have a tendency to wash out signal, pushing things towards 50% winrate. In other words, by symmetry there's no reason to expect that a much-lower skilled Z plays a higher skilled P more often than the other way around.
So if you have a large number of samples (like aligulac) and still see a big deviation like we do with PvZ, that's a pretty strong indication that there's a problem.
|
On January 03 2016 09:17 Athenau wrote:
In other words, by symmetry there's no reason to expect that a much-lower skilled Z plays a higher skilled P more often than the other way around.
There's no reason to make this assumption though. We clearly have a race population mismatch.
|
On January 03 2016 08:43 TheWinks wrote:Show nested quote +On January 03 2016 08:22 Big J wrote:On January 03 2016 07:12 TheWinks wrote:On January 03 2016 06:54 Nebuchad wrote:On January 03 2016 04:54 Lexender wrote:On January 03 2016 04:48 TheWinks wrote:On January 03 2016 03:48 BronzeKnee wrote:On January 03 2016 03:48 TheWinks wrote:
If we had blindly followed aligulac win rates the widow mine revert would have never happened. Aligulac's "balance" page has always been a poor source of information. Why would we blindly follow the win rates? I don't see anyone arguing for that. They are a source of information to be considered. But not ignored. Therefore we should neither follow them blindly, nor totally discard them. The fact that PvZ is at it's worst point in terms of win rates ever should be very concerning and is cause for some kind of change. Aligulac's win rates are worth ignoring. There's too much garbage included that obscures the end result. I remember when ByuN was being number 1 terran, in the middle of HotS We know how it works. When people play a lot of online tournaments, they end up advancing their rating way more (and way easier) than the people who play only GSL and the occasional week-end tournament, even though they are playing weaker opposition. You have demonstrated that the ranking is flawed, but that's completely different from collecting results and calculating a win percentage based on them. What is described as "garbage" is simply things that contradict the narrative you're pulling for. The two things you're describing are linked. If a top or near top level Korean goes into an online tournament and wipes the floor with a bunch of foreigners he gets a bigger boost in aligulac rating that he really should AND all those games get added into the win rate graph despite their lower quality. and the underlying assumption unless proven differently (statistically or deductively, not by giving a piece of evidence) is: such a player can be of any race, so the winrates get influenced by it in the same ways also you gotta prove that the samplesize is small enough for it mattering to begin with. time for some math if you want to support your hypothesis There is clearly not an equal race distribution just by looking at the balance report itself. TvZ: 987 PvT: 635 PvZ: 1061 Why would we assume what you're saying? Even if low quality games did mostly cancel each other out, why include them? It's unnecessary noise. Also, your second sentence doesn't make any sense.
I meant such players are distributed in the same way as the rest of the sample, not equally distributed. and i do it because it's the reasonable thing to do unless you actually create an objective criterion when a player outclasses the competition and then actually count the instances when it happened that such a player wiped the floor with others (and compare it across the races. my assumption: it's similar to the racial representation). everything else is delusional. that's why comments such as yours trigger me to respond in such a way. you try to diminish numbers without doing actual work. you give an example of something that could or does happen, but you don't do the real work behind it which is grinding the numbers why your example actually deserves to be mentioned. creating hypothesis is always nice. but that's not work, the actual work is giving your hypothesis significance and you don't do that by spitting out single names in the context of thousands of data points. just do the statistics if you are so sure you can show the numbers are meaningless.
and why include them? to cut them you need a criterion by what you cut. good luck creating one.
|
There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric.
|
On January 03 2016 09:30 Big J wrote:
that's why comments such as yours trigger me to respond in such a way. you try to diminish numbers without doing actual work.
Just because someone assembles numbers doesn’t mean someone else has to generate their own database to criticize them. A quick count the first GSL pre-season qualifier, ignoring walkovers, had these no-names: 9 protoss, 11 terrans, and 7 zergs. Every single one was beaten by the first notable player they reached. Ignoring mirrors we had these matchups (notable player is the first race):
TvP: 5
PvT: 4
ZvP: 7
PvZ: 1
TvZ: 2
ZvT: 4
A quick count of ZvP/PvZs in the second preseason qual had 3 and 4. Even if we argue about who's "notable", there's no reason to assume numbers are going to magically balance out and they have an impact on win rates.
On January 03 2016 09:51 Athenau wrote:Show nested quote +There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric.
The problem is it's not an inherent assumption you can make. For example:
David Kim wrote:
It’s the only major tournament going on right now, and it represents a sample size that is too small to draw any broad conclusions from, but Protoss players have lost at a noticeable clip in that tournament. As of this writing, they’ve recorded only 11 wins in 35 non-mirror matchups.
When this happened far more protoss were being fielded and hardly any terrans. The terrans seeing play were disproportionally higher skilled terrans. There was a ~50% win rate, but the average skill of the players being fielded was higher for Terran. This is the danger of just assuming everything cancels out.
|
During Zerg dominance near the end of WoL, aside from win-rate, mirror-matchup count was significantly higher for Zergs. Protoss too were relatively well represented, since they could soul-train zergs before reaching BL/infestor.
/edit
I did some digging around. Here is what I have found so far: Aligulac was used in this Bnet thread to justify balance during Brood/Infestor Era. Aligulac List 100 was posted as evidence.
Now in hindsight, we all acknowledge that Brood/Infestor Era was imbalanced. Here is the balance report for the period of List 100 in Aligulac. Note the mirror matchup count.
![[image loading]](http://i.imgur.com/WaQQFDq.jpg?1)
Here's a look at the most current period of Aligulac:
![[image loading]](http://i.imgur.com/aRNOTeV.jpg?1)
Zerg strength in both periods is strongly indicated by the mirror matchup count.
Links to the Aligulac pages:
http://aligulac.com/periods/153
http://aligulac.com/periods/100
|
On January 03 2016 09:51 Athenau wrote:Show nested quote +There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric.
The total pool of players are not meeting each other randomly. They advance through tournaments, requiring them to win in the first place.
|
On January 03 2016 10:49 plogamer wrote:+ Show Spoiler +During Zerg dominance near the end of WoL, aside from win-rate, mirror-matchup count was significantly higher for Zergs. Protoss too were relatively well represented, since they could soul-train zergs before reaching BL/infestor. /edit I did some digging around. Here is what I have found so far: Aligulac was used in this Bnet thread to justify balance during Brood/Infestor Era. Aligulac List 100 was posted as evidence. Now in hindsight, we all acknowledge that Brood/Infestor Era was imbalanced. Here is the balance report for the period of List 100 in Aligulac. Note the mirror matchup count.Here's a look at the most current period of Aligulac: Zerg strength in both periods is strongly indicated by the mirror matchup count. Links to the Aligulac pages: http://aligulac.com/periods/153
http://aligulac.com/periods/100
Yes, it's actually quite simple come to think of it. In a tournament where you have to qualify to compete it is natural that the competing players actually have a fighting chance against each other. If one race has an advantage in the meta, the result would be an overrepresentation of players of that race, but this means that the average skill level of the current pool of players playing that race is decreased and vice versa. Hence the top players of other races will do decently against them despite the disadvantage. All in all, in the format of tournaments requiring qualification it might just be a mathematical fact that the average winrate of each racial matchup will be equal in the long run.
Don't confuse this with blizzards statistics. They are doing a sophisticated statistical analysis of the ladder as a whole, taking all factors (such as MMR) in consideration to give a meaningful balance statistic as far as this is possible.
I've actually always thought that race representation in high level tournaments is the best indicator of balance at the top level. Even racial representation in GM could be a good indicator (although possibly not, as the players of a disadvantaged race could be more likely to ladder, since they are less likely to occupied with practicing for a tournament match).
|
On January 03 2016 09:51 Athenau wrote:Show nested quote +There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric.
No because its not a randomized experiment where you check all variants, if there is 100 zerg players and 10 protoss players, sure at some point a top zerg and a top protoss will meet, but until that point the protoss players may have defeated 10 bottom zerg players while the zerg player may have played ten mirrors wich will skew balance, wich is why not only non-mirrors have to be checked but mirrors too.
|
On January 03 2016 11:50 Lexender wrote:Show nested quote +On January 03 2016 09:51 Athenau wrote:There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric. No because its not a randomized experiment where you check all variants, if there is 100 zerg players and 10 protoss players, sure at some point a top zerg and a top protoss will meet, but until that point the protoss players may have defeated 10 bottom zerg players while the zerg player may have played ten mirrors wich will skew balance, wich is why not only non-mirrors have to be checked but mirrors too.
This is exactly why it makes sense to weight the win/loss statistics according to the players respective ELO rank parameters. So that if a zerg player has a 60% chance of winning against a lower level protoss, a win will be weighted as some number < 1, and conversely the protoss winning will be weighted by some number > 1.
However, given an imbalance meta, I suspect that in time as the ELO rankings will adjust and stabilize, it will stop representing the relative skill levels between players, and rather reflect a combination of skill level and current racial advantages/disadvantages. So in time, this weighted calculation might stabilize around even racial winrates. I would appreciate someone well-versed in statistics to chime in on this.
|
On January 03 2016 11:59 cheekymonkey wrote:Show nested quote +On January 03 2016 11:50 Lexender wrote:On January 03 2016 09:51 Athenau wrote:There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric. No because its not a randomized experiment where you check all variants, if there is 100 zerg players and 10 protoss players, sure at some point a top zerg and a top protoss will meet, but until that point the protoss players may have defeated 10 bottom zerg players while the zerg player may have played ten mirrors wich will skew balance, wich is why not only non-mirrors have to be checked but mirrors too. This is exactly why it makes sense to weight the win/loss statistics according to the players respective ELO rank parameters. So that if a zerg player has a 60% chance of winning against a lower level protoss, a win will be weighted as some number < 1, and conversely the protoss winning will be weighted by some number > 1. However, given an imbalance meta, I suspect that in time as the ELO rankings will adjust and stabilize, it will stop representing the relative skill levels between players, and rather reflect a combination of skill level and current racial advantages/disadvantages. So in time, this weighted calculation might stabilize around even racial winrates. I would appreciate someone well-versed in statistics to chime in on this.
What you're asking for is the bottom graph in the Aligulac report. But you run into the same problem here. Racial balance influences players' Elos, much like player skill influences overall winrates, so it'll always trend towards 0. The only thing it might help you with is determining how a change affects a matchup. E.g. if the map pool changes and you see zerg shoot up, you know the new pool is more zerg favourable than the last one - but only as a comparison to the previous pool, not that it's necessarily 'good'. And you'll see it fall back to even as Elos adjust.
EDIT: Unrelated, but whenever these threads pop up, you'll always have one party saying that the statistics are definitive proof, and probably using p-values from binomial tests with invalid assumptions. Then you'll have another saying that they should just be completely ignored because of selection bias and they're not independent and players have differing skills. Funny how the people treating them as gospel play the race doing poorly, and those claiming it's invalid are the ones doing well.
|
On January 03 2016 13:06 dainbramage wrote:Show nested quote +On January 03 2016 11:59 cheekymonkey wrote:On January 03 2016 11:50 Lexender wrote:On January 03 2016 09:51 Athenau wrote:There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric. No because its not a randomized experiment where you check all variants, if there is 100 zerg players and 10 protoss players, sure at some point a top zerg and a top protoss will meet, but until that point the protoss players may have defeated 10 bottom zerg players while the zerg player may have played ten mirrors wich will skew balance, wich is why not only non-mirrors have to be checked but mirrors too. This is exactly why it makes sense to weight the win/loss statistics according to the players respective ELO rank parameters. So that if a zerg player has a 60% chance of winning against a lower level protoss, a win will be weighted as some number < 1, and conversely the protoss winning will be weighted by some number > 1. However, given an imbalance meta, I suspect that in time as the ELO rankings will adjust and stabilize, it will stop representing the relative skill levels between players, and rather reflect a combination of skill level and current racial advantages/disadvantages. So in time, this weighted calculation might stabilize around even racial winrates. I would appreciate someone well-versed in statistics to chime in on this. What you're asking for is the bottom graph in the Aligulac report. But you run into the same problem here. Racial balance influences players' Elos, much like player skill influences overall winrates, so it'll always trend towards 0. The only thing it might help you with is determining how a change affects a matchup. E.g. if the map pool changes and you see zerg shoot up, you know the new pool is more zerg favourable than the last one - but only as a comparison to the previous pool, not that it's necessarily 'good'. And you'll see it fall back to even as Elos adjust. EDIT: Unrelated, but whenever these threads pop up, you'll always have one party saying that the statistics are definitive proof, and probably using p-values from binomial tests with invalid assumptions. Then you'll have another saying that they should just be completely ignored because of selection bias and they're not independent and players have differing skills. Funny how the people treating them as gospel play the race doing poorly, and those claiming it's invalid are the ones doing well.
Im EU GM and I think balance is all about statistics, excuse me?
|
On January 03 2016 13:06 dainbramage wrote:Show nested quote +On January 03 2016 11:59 cheekymonkey wrote:On January 03 2016 11:50 Lexender wrote:On January 03 2016 09:51 Athenau wrote:There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric. No because its not a randomized experiment where you check all variants, if there is 100 zerg players and 10 protoss players, sure at some point a top zerg and a top protoss will meet, but until that point the protoss players may have defeated 10 bottom zerg players while the zerg player may have played ten mirrors wich will skew balance, wich is why not only non-mirrors have to be checked but mirrors too. This is exactly why it makes sense to weight the win/loss statistics according to the players respective ELO rank parameters. So that if a zerg player has a 60% chance of winning against a lower level protoss, a win will be weighted as some number < 1, and conversely the protoss winning will be weighted by some number > 1. However, given an imbalance meta, I suspect that in time as the ELO rankings will adjust and stabilize, it will stop representing the relative skill levels between players, and rather reflect a combination of skill level and current racial advantages/disadvantages. So in time, this weighted calculation might stabilize around even racial winrates. I would appreciate someone well-versed in statistics to chime in on this. EDIT: Unrelated, but whenever these threads pop up, you'll always have one party saying that the statistics are definitive proof, and probably using p-values from binomial tests with invalid assumptions. Then you'll have another saying that they should just be completely ignored because of selection bias and they're not independent and players have differing skills. Funny how the people treating them as gospel play the race doing poorly, and those claiming it's invalid are the ones doing well.
So... you mean people are biased toward the race they play and/or like? Man, you just blew my world right now, thanks a lot for clearing that up.
As far as I know, I've always made it clear that I'm biased for protoss. I think the people who pretend not to be biased are either delusional or attempting to hide their agenda. And when my side presents an argumentation, for example, saying that when a match-up is under 50% and trending consistently downwards for the last three months, and said match-up is about to get patched with two pretty important nerfs to the downtrending race, it is reasonable to have some concerns, I'd like a little more than "yeah but you're biased" in return.
But still, thanks.
|
On January 03 2016 10:49 plogamer wrote:+ Show Spoiler +During Zerg dominance near the end of WoL, aside from win-rate, mirror-matchup count was significantly higher for Zergs. Protoss too were relatively well represented, since they could soul-train zergs before reaching BL/infestor. /edit I did some digging around. Here is what I have found so far: Aligulac was used in this Bnet thread to justify balance during Brood/Infestor Era. Aligulac List 100 was posted as evidence. Now in hindsight, we all acknowledge that Brood/Infestor Era was imbalanced. Here is the balance report for the period of List 100 in Aligulac. Note the mirror matchup count.Here's a look at the most current period of Aligulac: Zerg strength in both periods is strongly indicated by the mirror matchup count. Links to the Aligulac pages: http://aligulac.com/periods/153
http://aligulac.com/periods/100
I didn't read through all of this so I won't participate in the discussion, just stopping by to say BL/Infestor was end of 2012/beginning of 2013, not end of 2013, so unless the date in your first graph is shown incorrectly you've got the wrong graph
|
On January 03 2016 13:06 dainbramage wrote:Show nested quote +On January 03 2016 11:59 cheekymonkey wrote:On January 03 2016 11:50 Lexender wrote:On January 03 2016 09:51 Athenau wrote:There's no reason to make this assumption though. We clearly have a race population mismatch.
Relative size doesn't matter. Even if you have, for example, a population of 100 zerg players and only 10 protoss players, the number of ways to match a top 10% zerg player with a bottom 10% protoss player is the same as the other way around. The problem is symmetric. No because its not a randomized experiment where you check all variants, if there is 100 zerg players and 10 protoss players, sure at some point a top zerg and a top protoss will meet, but until that point the protoss players may have defeated 10 bottom zerg players while the zerg player may have played ten mirrors wich will skew balance, wich is why not only non-mirrors have to be checked but mirrors too. This is exactly why it makes sense to weight the win/loss statistics according to the players respective ELO rank parameters. So that if a zerg player has a 60% chance of winning against a lower level protoss, a win will be weighted as some number < 1, and conversely the protoss winning will be weighted by some number > 1. However, given an imbalance meta, I suspect that in time as the ELO rankings will adjust and stabilize, it will stop representing the relative skill levels between players, and rather reflect a combination of skill level and current racial advantages/disadvantages. So in time, this weighted calculation might stabilize around even racial winrates. I would appreciate someone well-versed in statistics to chime in on this. What you're asking for is the bottom graph in the Aligulac report. But you run into the same problem here. Racial balance influences players' Elos, much like player skill influences overall winrates, so it'll always trend towards 0. The only thing it might help you with is determining how a change affects a matchup. E.g. if the map pool changes and you see zerg shoot up, you know the new pool is more zerg favourable than the last one - but only as a comparison to the previous pool, not that it's necessarily 'good'. And you'll see it fall back to even as Elos adjust. EDIT: Unrelated, but whenever these threads pop up, you'll always have one party saying that the statistics are definitive proof, and probably using p-values from binomial tests with invalid assumptions. Then you'll have another saying that they should just be completely ignored because of selection bias and they're not independent and players have differing skills. Funny how the people treating them as gospel play the race doing poorly, and those claiming it's invalid are the ones doing well.
And then you have the few that actually can do a proper analysis rather than just blindly apply a method that may or may not make sense to the case. But if you are not one of them, it is close to impossible to separate them from the rest of the pack.
I'm zerg and I have been arguing in this thread that the P > Z signal is real and important, but I guess you also have confirmation bias when you look at the race of posters?  People kindof take the easy way out and just go "your biased, thus you are wrong", and somehow takes that as evidence of the opposite ("CIA denied it, so it must be true"). It is perfectly possible to be biased but still get things right. You only have to be more careful that what you are doing actually makes sense, be self-critical and be open to comment from others. You also have to be able to do the analysis to start with...
|
On January 02 2016 20:26 Ghanburighan wrote:This is somewhat disingenuous by Bronzeknee as such stats have been discussed in the designated dalance discussion thread already and been found wanting. Bronzeknee was part of these discussions so he should know better. That's not to say that there isn't an early indication of a problem in PvZ.
Disingenuous? No, I'm actually totally sincere in posting this. I haven't participated in that discussion in literally years. Probably, since around the time of the release of HOTS (likely even further back).
But I do remember you quite clearly arguing for long periods of time that we should adopt a wait and see approach, as you've stated in this thread. I've always found that to be a very ignorant thing to do in light of massive imbalance (PvZ has never been this imbalanced for either side).
The fact that we don't know what we don't know is not a reason to do nothing about what we do know.
And it never will be.
|
|
|
|
|
|
EPT
