|
On March 15 2013 07:36 Excalibur_Z wrote:
It's a little difficult to tell but let's delve into the theory. Mendelfist is right (and Josh the former ladder designer pointed this out as well) that if you were to lop off the bottom 20%, the relative positions between players won't change, and this is what the Elo format measures. Blizzard would necessarily step in and redefine the boundaries if this happened, of course. However, what makes a Silver player Silver? It's because most of the time, he beats Bronzes and loses to Golds.
Just looking at MMR (or Elo) numbers, if you lop off the bottom 20%, the person who's at the 20.0000001th %-ile will begin to lose a much higher percentage of his games, and his score will drift downward as a result. This will pull everyone else with them until the system achieves a new equilibrium in which the score accurately predicts win/loss likelihood once more.
Thus, everyone's actual rating number will decrease, and depending on how the math works they may or may not spread out, but like you say, they'll retain the same relative positions.
|
On March 15 2013 06:27 Mendelfist wrote:Show nested quote +I'm also pretty sure you'd get MMR drift if you removed everyone in Bronze from the playing population. I think not. This question has come up before, and Excalibur_Z actually asked some dev about it. In that example I think it was "every gold player quits" or something, and the answer is that gold will stay empty, because MMR difference translates directly to winning chance, and that prevents any stretching or contracting of the MMR scale. There is of course the question of where the complete MMR scale is "anchored". My theory is that it's anchored at the top, and in that case removing all bronze players will have no effect.
I tend to think that the scale is not "anchored" at all. Yes, relative positions remain the same, but if you took off either end of the scale, the new lowest player would always lose, causing the scale to drift downward and the new highest player would always win. Only if you remove the league in the middle does the system fail to drift up or down.
|
This is a good move from blizzard. That's a way to keep a good amount of active players.
I think it would be even better if it was: B 8% S 20% G 25% P 25% D 20% M 2% GM[200]
|
On March 15 2013 07:45 Lysenko wrote:Show nested quote +On March 15 2013 07:36 Excalibur_Z wrote:
It's a little difficult to tell but let's delve into the theory. Mendelfist is right (and Josh the former ladder designer pointed this out as well) that if you were to lop off the bottom 20%, the relative positions between players won't change, and this is what the Elo format measures. Blizzard would necessarily step in and redefine the boundaries if this happened, of course. However, what makes a Silver player Silver? It's because most of the time, he beats Bronzes and loses to Golds. Just looking at MMR (or Elo) numbers, if you lop off the bottom 20%, the person who's at the 20.0000001th %-ile will begin to lose a much higher percentage of his games, and his score will drift downward as a result. This will pull everyone else with them until the system achieves a new equilibrium in which the score accurately predicts win/loss likelihood once more. Thus, everyone's actual rating number will decrease, and depending on how the math works they may or may not spread out, but like you say, they'll retain the same relative positions.
No, I understand Excalibur_Z's explanation and I'm pretty sure he's right and I'm wrong. With all my capybara-cannon victims dead, the lowest-ranked survivor suddenly starts losing (say) 60% of his games because of the lack of weaker opponents. But his ranking already reflects the fact that he's expected to lose about 60% against the range of players he's being matched up against, so his MMR should stay put.
|
Good change! I never understood why games such as LoL and SC2 insists on having tons of leagues for the few highest ranked players and then have huge percentile of players in the few lower leagues.
|
I did my placement matches yesterday and received gold. I was hoping for plat, but oh well. I wonder if I'm real gold or fool's gold?
|
On March 15 2013 07:39 Lysenko wrote:Show nested quote +On March 15 2013 04:54 EleanorRIgby wrote:On March 15 2013 02:54 monkybone wrote:On March 15 2013 01:19 Evangelist wrote:
Someone might have a near perfect theoretical understanding of the game, but not necessarily be very good at playing it. This is one of the biggest myths out there. Funny how it's still prevalent. indeed, if you had near perfect understanding of everything you can get to masters with less then 100apm and minimal hotkey usage. Are you saying that there can't be a person with excellent understanding of the game and awful reaction time? Edit: I suppose someone could argue that it's impossible to gain such an understanding of the game without actually playing at that level. I think this is a hard argument to make, because watching a game presents a person with all the same (in fact often more!) information than the players see. Of course, such a person would have to be a great talent who's held back in some fundamental mechanical way.
There are examples of people who have a ton of knowledge about the game as a whole, but maybe not professionals or GM players. The prime example is: Artosis. Every pro says he has an amazing level of knowledge about the game, builds and players.
So when someone say that they are a low level player with great game knowledge, ask yourself. Are they the Artosis of Gold/Plat players? 98% of the time the answer is going to be no and the person just believes they are.
Or you can look at the: http://en.wikipedia.org/wiki/Dunning_kruger_effect
Dunning and Kruger proposed that, for a given skill, incompetent people will:
1. tend to overestimate their own level of skill;
2. fail to recognize genuine skill in others;
3. fail to recognize the extremity of their inadequacy;
4. recognize and acknowledge their own previous lack of skill, if they are exposed to training for that skill.
This simple theory explains 90% of the internet and SC2.
|
Top 2% is still the top 2%. This makes me happy. 
|
On March 15 2013 07:39 Lysenko wrote:Show nested quote +On March 15 2013 04:54 EleanorRIgby wrote:On March 15 2013 02:54 monkybone wrote:On March 15 2013 01:19 Evangelist wrote:
Someone might have a near perfect theoretical understanding of the game, but not necessarily be very good at playing it. This is one of the biggest myths out there. Funny how it's still prevalent. indeed, if you had near perfect understanding of everything you can get to masters with less then 100apm and minimal hotkey usage. Are you saying that there can't be a person with excellent understanding of the game and awful reaction time?Edit: I suppose someone could argue that it's impossible to gain such an understanding of the game without actually playing at that level. I think this is a hard argument to make, because watching a game presents a person with all the same (in fact often more!) information than the players see. Of course, such a person would have to be a great talent who's held back in some fundamental mechanical way.
The point is that if you have excellent understanding of the game you won't be in Bronze or Silver, no matter what your reaction time is (not accounting for physical disablilities). People often claim great insight and blame lack of execution for their lower rankings, which seems absurd.
|
On March 16 2013 00:36 Vorenius wrote:
The point is that if you have excellent understanding of the game you won't be in Bronze or Silver, no matter what your reaction time is (not accounting for physical disablilities). People often claim great insight and blame lack of execution for their lower rankings, which seems absurd.
I agree the vast bulk of people who make such claims don't know what they're talking about (and I would never, ever make such a claim myself), but there are a number of reasons someone might not be able to play at top levels and still have a strong understanding of the game. (Having trouble clicking accurately, having trouble with the coordination involved in hitting the right keys, reaction time, issues with focus that cause one to pay attention to the wrong things, etc.)
|
On March 15 2013 20:02 AmericanUmlaut wrote:
No, I understand Excalibur_Z's explanation and I'm pretty sure he's right and I'm wrong. With all my capybara-cannon victims dead, the lowest-ranked survivor suddenly starts losing (say) 60% of his games because of the lack of weaker opponents. But his ranking already reflects the fact that he's expected to lose about 60% against the range of players he's being matched up against, so his MMR should stay put.
Thanks for this, I see the argument now, and I think you're probably right.
|
On March 16 2013 03:45 Lysenko wrote:Show nested quote +On March 16 2013 00:36 Vorenius wrote:
The point is that if you have excellent understanding of the game you won't be in Bronze or Silver, no matter what your reaction time is (not accounting for physical disablilities). People often claim great insight and blame lack of execution for their lower rankings, which seems absurd. I agree the vast bulk of people who make such claims don't know what they're talking about (and I would never, ever make such a claim myself), but there are a number of reasons someone might not be able to play at top levels and still have a strong understanding of the game. (Having trouble clicking accurately, having trouble with the coordination involved in hitting the right keys, reaction time, issues with focus that cause one to pay attention to the wrong things, etc.)
I think in that case its more of a physical disability when it comes to clicking. And as the above poster said that doesn't really count.
And game knowledge at a lower level is like, he built mutas so if i build a stargate i can counter and win. While in theory that should work, it won't due to the fact that he will have way too many mutas for you to deal with by the time you get 4 phoenix.
|
On March 12 2013 03:23 Inori wrote:
Wouldn't that mean that there will be a huge variety of skill level at gold league?
The move actually makes sense. Skill isn't uniformly distributed. I'm not sure it's normally distributed either (bell curve), but this isn't one. They're probably trying to make it look something like a gamma distribution, which makes sense as the best first-order approximation to me. Look those up on wikipedia if you don't have a stats background.
|
United States12235 Posts
On March 18 2013 02:04 Bombadil819 wrote:Show nested quote +On March 12 2013 03:23 Inori wrote:
Wouldn't that mean that there will be a huge variety of skill level at gold league? The move actually makes sense. Skill isn't uniformly distributed. I'm not sure it's normally distributed either (bell curve), but this isn't one. They're probably trying to make it look something like a gamma distribution, which makes sense as the best first-order approximation to me. Look those up on wikipedia if you don't have a stats background.
It's not a gamma distribution, but it's doesn't 100% mirror a normal distribution either. It's a normal distribution with a little extra bump in the direction of the Diamondish region where the more hardcore players tend to gather, but that's not reflected in my graphical representation because it's not clear where exactly the bump is, so I just omitted it.
|
Hopefully will make the game more pleasant for newer players and, keep them playing. This is very smart of Blizzard. Hopefully, we will see less players quit HOTS after a couple months as a result of this. We will see I guess.
|
|
|
|
|
|
|
|
|
|
EPT
