The skill variances for the middle are too close to say which is better. Both sets are basically fair teams because the mix of players is fine.
Also, quantitative equality is not necessarily good. The standard deviation might look better, but I have serious misgivings about not having the best player by a margin and having the dead weight 500 score player. Winning engagements is important and the 3000 would probably dominate the field while the 500 is just there not doing much.
That said, the 3000 has to (and probably will) carry harder with shuffle as the next three best players are on the other team. Either way, I think that the 3000's team is more likely to win.
Aside from this, there is the problem of commanders (and lifeforms). Teams are assigned before commanders and seemingly balanced teams are not anymore when the wrong person jumps into the chair.
IxianDenmarkJoin Date: 2014-03-16Member: 194783Members, Squad Five Blue
@Nordic Adamant? It has never been my intention to have immobile openions - I do find that letting go of them whenever they mean any resistance at all isn't viable. Instead try to find where the difference in the option and the current, to understand what it is based upon. I too often find that the premisis of opinions have huge influence in what opinion one holds.
This is the major point of my disagreement with the current system. Shine is based upon equaling the odds/elo to make a game that would be considered fair, and with equal skill on either side. I would rather see a more equally destributed pressure to perform up to your own elo/skill level - right now, most of the time, one has to make up for other players lack of skill, which seems like a unfair prospect, and counter-intuitive in a team based FPS. The Captains method is the best method I know which is able to do this. On this premisis, I find Captains method to be superior. If the premisis changes, so does the superior. The shine method is superior in other premisis, such as mathematical fairness. (which, as mendasp mentioned, can't really be obtained with the current playground, with no matchmaking, and a unsatisfyingly small group of players for this method). But if the logic changes, so does the superior. Maybe Captains method isn't the superior, maybe it is 1-1-1-1-1-1. But it have so far found it not to be - not on this premisis atleast.
The reason I made this thread is because I keep seeing players saying shuffle does not work. I see statements like shuffle does X, but when in reality shuffle does not ever do X.
Shuffle is pretty simple in that it maximizes mathematical equality. It does a really good job at that, but that is all it does. I think the problems people are blaming shuffle for are not problems with shuffle. It is other factors like the inadequacy of hive skill. It is the fact that shuffle only matches teams at the beginning. After shuffle people switch teams. Sometimes people leave and others join. Then there is the fact that ns2's playerbase is so small that we can not find near skilled teams. I typically see the standard deviation of a team have a low end of 400, which is really high. These are the factors that mess up the so called balance, not shuffle.
Even after reading both your posts I can't think of a better way teams could be balanced given the factors I just mentioned. Captains often does a pretty good job. That is a human choosing teams, not a computer. Captains also relies on people having a good idea of others actual skill, not just what their hive skill is.
This doesn't even consider the possibility of 3k or so players commanding as well. I don't know but last I checked I was above 3k in pubs when I used to play NS2 a lot more but I would also marine comm a lot as well in the past. If the server wasn't particularly good or the split wasn't good, I would feel forced to play only as a grunt rather than having a choice of playing grunt or comm which is ironic when a lot of players can feel forced into commanding because no-one else wants to.
When I came back to the game a few months ago after a long break, all I had was stacked games because the ELO shuffle just didn't work. I only got about 1/20 pub games that weren't a stomp. Had better quality games on woozas server than general pubs.
Shuffle doesn't work, whether that's to the method of shuffling or extraneous variables, its not an adequate system as my experience of it over the past few months has been some of the worst pub experiences I've ever had in any game. It's not worth me playing anymore since I was fairly casual anyway and don't have an interest in competitive because the overall quality of games was so bad that I felt no enjoyment in the game anymore even when we're winning/stomping, it wasn't fun whatsoever.
I don't think that is the issue. Shuffling teams in game does not always/usually/ever* work. It doesn't matter if it is the algorithm or hive skill or bad pool of players or whatever else, what they see is that the resulting game sucks and thus the tool that purports to solve that problem is broken.
Nobody is in a position to preach to others as to how they should play the game, but I think people should concede more readily when they see that the other team is clearly better. I think that helps the issue much more than an improved shuffle or captains or whatever else because bad games happen regardless. Just go on to the next game and get a fresh start rather than getting frustrated. That is not to say that you shouldn't try for a comeback, but when the skill difference is obvious, hanging on just makes it painful to play unless you enjoy turtling.
IxianDenmarkJoin Date: 2014-03-16Member: 194783Members, Squad Five Blue
@Nordic Standart deviation is only a concern with the current system. In a captains method, where its pressure to perform to your own skill level, rather than make up for other players lack thereof, it isn't a noteworthy factor.
An actual person to pick the players wouldn't be required! Ranking by elo and then destributing 1-2-2-2-2 would be enough. If the system would be REALLY cool, it would learn from Evolve, and make sure the aliens have atleast minimum 2-3 lerk players, and minimum 2 fade players at the very least. It would require additional features, and additional information that the system has to handle, but it would, without any doubt in my mind, make for better games.
It is however widely disagreed upon what a good game is. For many it is an epic long game that lasts 50 minutes. Personally, I often find those games disgusting - these games last this long due to people tunnelvisioning on a single 10 minute push, no resbiting, lollygagging, inaction, flashing onoses, slow commanders who doesnt med or support, failue to perform an actual assault on a hive after JPs are up and so on and so forth.
Experiencing a stomp often depends on the first 1-2 minutes, where one team gains a good advantage, to build upon, which they then continue to build upon, the failiue of the enemy team to find an ansver, and then the huge advantage in technology ending the game within 15 minutes. Often disguised as "sheer luck" or "imbalanced teams", while it really just comes down to oppotunities exploited. Two good friends going together as marines to an enemy natural and then mudering everything on their vay while denying the natural and further expansions isn't as much stacking, as it is an oppotunity exploited. The aliens simply didn't cover for this risk, and created their own disadvantage, sometimes without even creating a difference advantage. If this "minmaxing" is stacking, then i am a stacker - blame me for playing to the best of my ability, and exploiting your teams lack!
Sometimes you fuck up, and play poorly, but people have a habit of blaming the teams. On servers like woozas, these advantages are smaller, due to each player having less influence, meaning fewer "stomps". Long games doesnt mean equal skill - it means a lack of oppotunities spotted and exploited. with 20 players on each team, fewer oppotunities are available, which means in the end it really becomes about moving the mass of players, and hoping you win the engagement - something I find very boring. It may create the epic feeling, but there are no room to develop your wonderful brain, to think in options, effects, and then the new options. With 6v6 to 9v9 it is a fine piece of chess being played, where every player is valuable and their actions carry consequence, and their death has obvious damage. In 20v20 these qualities dissapear and the chess game has everything but the queen, king and towers removed, a very boring game as you might imagine compared to the full setup, with all its beautiful opening, gambits, trades, exploitations, agressions, pre-emptive movements, and defensive setups. The more players, the more options simply dissapear, due to the nature of the map, as opposite to the chessboard. If there were maps, and scaled HP for structures made for 20v20 then this wouldn't be true. But since many maps are made for 6v6 to 10v10, they are half the size and complexity that a 20v20 game would require to function as elequently as 6v6 to 10v10. IMO 12v12 is pushing it.
The aliens simply didn't cover for this risk, and created their own disadvantage, sometimes without even creating a difference advantage. If this "minmaxing" is stacking, then i am a stacker - blame me for playing to the best of my ability, and exploiting your teams lack!
Not really much the alien team can do in this situation. If more than two Skulks need to react and withdraw from the forward position, they've already lost valuable grounds at other places. Same goes for Marines, when Skulks bypass and group to get that new arms lab or an IP. But then a lost arms lab is never as damaging as loosing the natural RT.
For me, it doesn't need to be a 50 minute game. Just something longer than 10 minutes that doesn't feel like a complete stomp. I don't really care if my team isn't doing everything 100% the way competitive players do, I just want to play the game and not by constantly facing stompfests in 90% of my games because whether or not you win isn't the objective anymore but trying to get a balanced team to at least have a chance at an entertaining game.
@Ixian Doesn't matter why a long game is well...longer than usual, its at least something out of the ordinary for me that is much better than the ordinary < 10 minute games. While I'm not looking for something as brain dead as woozas server, I don't care if people perform poorly as much in 16-24 if the game is in a state where either team can win. It's got to the point where I don't even play this game anymore because I just couldn't be bothered dealing with the poor balance system but I like to keep in touch with the updates to see what changes in the future. As a pubby, any bad hit detection or any glitches and bugs or design flaws felt like annoyances but I could look past them because I enjoyed the core gameplay so much but when it comes to this balance system, it got me to quit the game outright. I don't know what happened in the past few months but its turned into shit, might've just been my luck but it sucked.
@Nordic Standart deviation is only a concern with the current system. In a captains method, where its pressure to perform to your own skill level, rather than make up for other players lack thereof, it isn't a noteworthy factor.
An actual person to pick the players wouldn't be required! Ranking by elo and then destributing 1-2-2-2-2 would be enough. If the system would be REALLY cool, it would learn from Evolve, and make sure the aliens have atleast minimum 2-3 lerk players, and minimum 2 fade players at the very least. It would require additional features, and additional information that the system has to handle, but it would, without any doubt in my mind, make for better games.
I don't agree that 1-2-2-2 is the way to go. It really depends on what players are available and how large the skill gap between each player is. Using this method, you are not really solving anything but trading the problems of one solution for another. It needs to be more flexible. I think ranking players by tier could be better but it might just produce the same results as score averaging.
IxianDenmarkJoin Date: 2014-03-16Member: 194783Members, Squad Five Blue
Elo is the "ranking by tier" you are talking about. And if they cannot live up to being "first pick" they lose elo. Finally elo is a system where you have to live up to your own skill, and not make up for others lack thereof.
Elo is the "ranking by tier" you are talking about. And if they cannot live up to being "first pick" they lose elo. Finally elo is a system where you have to live up to your own skill, and not make up for others lack thereof.
No. What I meant is by grouping, for example, 0-500 as a tier, 1000-1200 as another tier and so on, where there is less emphasis on raw numbers and more on match ups. It could very well result in the exact same teams as normal shuffling, but it seems fairer to me.
Also, your insistence on elo/hive being a reflection of your own skill does not hold no matter how much you wish. It is a team game. What your team does matters as much as what you do.
IxianDenmarkJoin Date: 2014-03-16Member: 194783Members, Squad Five Blue
edited February 2016
Seems like i constantly have to repeat myself in this discussion: I have never insisted that elo is a direct reflection of your actual mechanical skill, theoretical knowledge, awareness and communicative skills. the player base is simply too small for elo to actually work. And referencing something i have said earlier somewhere else - if you want fairness, join a team and play PCW's and officials.
I didn't say anything about elo not making sense on a small player base.
I did say that it's stupid to argue about hive while it uses a single number for all 4 roles (marine, alien, and both comms).
And if it did that, it can't predict who's going to command for doing the shuffle to use the correct elo to balance.
In any case it should have been a hidden number, and it being based on community (pub) servers is another weakness, it would work for actual matchmaking, where you can't "choose" your opponent.
Seems like i constantly have to repeat myself in this discussion: I have never insisted that elo is a direct reflection of your actual mechanical skill, theoretical knowledge, awareness and communicative skills. I have agreed with mendasp's very true statement that the player base is simply too small for elo to actually work. And referencing something i have said earlier somewhere else - if you want fairness, join a team and play PCW's and officials.
Elo is the "ranking by tier" you are talking about. And if they cannot live up to being "first pick" they lose elo. Finally elo is a system where you have to live up to your own skill, and not make up for others lack thereof.
Maybe try reading your own posts. And no one is talking about game knowledge, just skill evaluation, but sure, go toot your own horn at every available opportunity.
Let me ask a bit different question regarding shuffle given the evaluation of teams is focused on average and deviation.
There is a well know puzzle about choosing a die in a set and still losing no matter which you picked. https://en.wikipedia.org/wiki/Nontransitive_dice
Assume that a player with higher skill always kills a player with lower skills, which team/die would you pick?
The answer is that B in average beats A, C in average beats B, but A in average beats C.
The winning move is pick the die after the other player.
So the question is; what would an optimal shuffle heuristic actually optimize?
Average and deviation are easy to calculate, but that really is not an argument for using them.
It is a team game. What your team does matters as much as what you do.
You dragged me into, talking down to me, I say ok, and ethos beccomes a thing. Im gonna toot my horn, when you try to break it. Leave personas out of it, and we wont have to talk about each others achievements, rather only the subject at hand which is much more interesting.
I disagree with elo and the way it is used and calculated - hence the " ". The statement was also to respond to your wish for a ranking by tier, to which I added that I believed such a thing was already in place. I apolegise if I have been too unclear about these things.
It is a team game. What your team does matters as much as what you do.
You dragged me into, talking down to me, I say ok, and ethos beccomes a thing. Im gonna toot my horn, when you try to break it. Leave personas out of it, and we wont have to talk about each others achievements, rather only the subject at hand which is much more interesting.
I disagree with elo and the way it is used and calculated - hence the " ". The statement was also to respond to your wish for a ranking by tier, to which I added that I believed such a thing was already in place. I apolegise if I have been too unclear about these things.
Now I am confused. Are you being sarcastic or sincere on this?
And if they cannot live up to being "first pick" they lose elo. Finally elo is a system where you have to live up to your own skill, and not make up for others lack thereof.
You have been repeating it so I took it as being sincere and it really is not true. It should work out in the aggregate (because large sample size should in theory cover everything), but really, you need your whole team to live up to their skill and most importantly, your commander to do at least the minimum. Even the best players will be limited by a bad commander. It is much easier to carry a 200 elo deadweight on the field than to carry him in the chair, for example.
But if you were being sarcastic, my entire point is moot. In that case, whatever.
I did say that it's stupid to argue about hive while it uses a single number for all 4 roles (marine, alien, and both comms).
And if it did that, it can't predict who's going to command for doing the shuffle to use the correct elo to balance.
This is the biggest flaw imho of any of the shuffles out there .. none take the commander into account .. the shuffle should never shuffle until commanders are in place .. and not shuffle the commanders. Regardless of all the data you collect to try get a balanced game going, unless you can weight the commanders skill based on some metrics .. (eg. commander akill being based on the skill of their fieldteam and the oppositions comm+fieldteam skill)
As for those 3k guys joining pub servers .. imho .. it's on THEM to pull back when they can obviously stomp a pub game and carry the entire team .. especially with such broken match making systems.. when you see one marine with 40-2 k/d on a public server where the next highest kill count is ~6 .. then that player needs to either pull their punches or go to another server .. it happens so often, bored pro lads ..I'd be on for handicapping these guys play on a pub game and instead of stomping, accept a handicap for the server?
offtopic That might be a decent idea for a server mod :P stompers can get landed with a handicap to keep them honest .. anyone over hive skill X difference with server average.. if they hit a 4/1 k/d ratio get hit with .. I dunno, armour/weap/speed penalty.. surley stompers couldnt complain too much .. they'd get a decent challenge on pub games and no doubt a mod like this wouldn't affect anyones ratings .. which in itself is a bit unfortunate since if it became popular would break itself :P
/offtopic
Let me ask a bit different question regarding shuffle given the evaluation of teams is focused on average and deviation.
There is a well know puzzle about choosing a die in a set and still losing no matter which you picked. https://en.wikipedia.org/wiki/Nontransitive_dice
Assume that a player with higher skill always kills a player with lower skills, which team/die would you pick?
The answer is that B in average beats A, C in average beats B, but A in average beats C.
The winning move is pick the die after the other player.
So the question is; what would an optimal shuffle heuristic actually optimize?
Average and deviation are easy to calculate, but that really is not an argument for using them.
If the player with the higher skill ALWAYS kills the lower player, than team B has two players who NEVER die. They could literally walk into the hive/marine base and win the game outright. Which is more or less exactly what happens when you have 4k players.
3k players on pubs usually have 130+ping and use the advantage of lag. Such players will rule the ground they walk. People speak about limiting high skill, then just limit high ping...
A 2k player with a 130 > a 2k player with 80. How can you balance lag compensation.
This is the biggest flaw imho of any of the shuffles out there .. none take the commander into account .. the shuffle should never shuffle until commanders are in place .. and not shuffle the commanders.
Shine's shuffle has a setting to ignore commanders. I wish everyone enabled it.
Comments
The skill variances for the middle are too close to say which is better. Both sets are basically fair teams because the mix of players is fine.
Also, quantitative equality is not necessarily good. The standard deviation might look better, but I have serious misgivings about not having the best player by a margin and having the dead weight 500 score player. Winning engagements is important and the 3000 would probably dominate the field while the 500 is just there not doing much.
That said, the 3000 has to (and probably will) carry harder with shuffle as the next three best players are on the other team. Either way, I think that the 3000's team is more likely to win.
Aside from this, there is the problem of commanders (and lifeforms). Teams are assigned before commanders and seemingly balanced teams are not anymore when the wrong person jumps into the chair.
This is the major point of my disagreement with the current system. Shine is based upon equaling the odds/elo to make a game that would be considered fair, and with equal skill on either side. I would rather see a more equally destributed pressure to perform up to your own elo/skill level - right now, most of the time, one has to make up for other players lack of skill, which seems like a unfair prospect, and counter-intuitive in a team based FPS. The Captains method is the best method I know which is able to do this. On this premisis, I find Captains method to be superior. If the premisis changes, so does the superior. The shine method is superior in other premisis, such as mathematical fairness. (which, as mendasp mentioned, can't really be obtained with the current playground, with no matchmaking, and a unsatisfyingly small group of players for this method). But if the logic changes, so does the superior. Maybe Captains method isn't the superior, maybe it is 1-1-1-1-1-1. But it have so far found it not to be - not on this premisis atleast.
The reason I made this thread is because I keep seeing players saying shuffle does not work. I see statements like shuffle does X, but when in reality shuffle does not ever do X.
Shuffle is pretty simple in that it maximizes mathematical equality. It does a really good job at that, but that is all it does. I think the problems people are blaming shuffle for are not problems with shuffle. It is other factors like the inadequacy of hive skill. It is the fact that shuffle only matches teams at the beginning. After shuffle people switch teams. Sometimes people leave and others join. Then there is the fact that ns2's playerbase is so small that we can not find near skilled teams. I typically see the standard deviation of a team have a low end of 400, which is really high. These are the factors that mess up the so called balance, not shuffle.
Even after reading both your posts I can't think of a better way teams could be balanced given the factors I just mentioned. Captains often does a pretty good job. That is a human choosing teams, not a computer. Captains also relies on people having a good idea of others actual skill, not just what their hive skill is.
When I came back to the game a few months ago after a long break, all I had was stacked games because the ELO shuffle just didn't work. I only got about 1/20 pub games that weren't a stomp. Had better quality games on woozas server than general pubs.
Shuffle doesn't work, whether that's to the method of shuffling or extraneous variables, its not an adequate system as my experience of it over the past few months has been some of the worst pub experiences I've ever had in any game. It's not worth me playing anymore since I was fairly casual anyway and don't have an interest in competitive because the overall quality of games was so bad that I felt no enjoyment in the game anymore even when we're winning/stomping, it wasn't fun whatsoever.
I don't think that is the issue. Shuffling teams in game does not always/usually/ever* work. It doesn't matter if it is the algorithm or hive skill or bad pool of players or whatever else, what they see is that the resulting game sucks and thus the tool that purports to solve that problem is broken.
Nobody is in a position to preach to others as to how they should play the game, but I think people should concede more readily when they see that the other team is clearly better. I think that helps the issue much more than an improved shuffle or captains or whatever else because bad games happen regardless. Just go on to the next game and get a fresh start rather than getting frustrated. That is not to say that you shouldn't try for a comeback, but when the skill difference is obvious, hanging on just makes it painful to play unless you enjoy turtling.
*delete as applicable
An actual person to pick the players wouldn't be required! Ranking by elo and then destributing 1-2-2-2-2 would be enough. If the system would be REALLY cool, it would learn from Evolve, and make sure the aliens have atleast minimum 2-3 lerk players, and minimum 2 fade players at the very least. It would require additional features, and additional information that the system has to handle, but it would, without any doubt in my mind, make for better games.
It is however widely disagreed upon what a good game is. For many it is an epic long game that lasts 50 minutes. Personally, I often find those games disgusting - these games last this long due to people tunnelvisioning on a single 10 minute push, no resbiting, lollygagging, inaction, flashing onoses, slow commanders who doesnt med or support, failue to perform an actual assault on a hive after JPs are up and so on and so forth.
Experiencing a stomp often depends on the first 1-2 minutes, where one team gains a good advantage, to build upon, which they then continue to build upon, the failiue of the enemy team to find an ansver, and then the huge advantage in technology ending the game within 15 minutes. Often disguised as "sheer luck" or "imbalanced teams", while it really just comes down to oppotunities exploited. Two good friends going together as marines to an enemy natural and then mudering everything on their vay while denying the natural and further expansions isn't as much stacking, as it is an oppotunity exploited. The aliens simply didn't cover for this risk, and created their own disadvantage, sometimes without even creating a difference advantage. If this "minmaxing" is stacking, then i am a stacker - blame me for playing to the best of my ability, and exploiting your teams lack!
Sometimes you fuck up, and play poorly, but people have a habit of blaming the teams. On servers like woozas, these advantages are smaller, due to each player having less influence, meaning fewer "stomps". Long games doesnt mean equal skill - it means a lack of oppotunities spotted and exploited. with 20 players on each team, fewer oppotunities are available, which means in the end it really becomes about moving the mass of players, and hoping you win the engagement - something I find very boring. It may create the epic feeling, but there are no room to develop your wonderful brain, to think in options, effects, and then the new options. With 6v6 to 9v9 it is a fine piece of chess being played, where every player is valuable and their actions carry consequence, and their death has obvious damage. In 20v20 these qualities dissapear and the chess game has everything but the queen, king and towers removed, a very boring game as you might imagine compared to the full setup, with all its beautiful opening, gambits, trades, exploitations, agressions, pre-emptive movements, and defensive setups. The more players, the more options simply dissapear, due to the nature of the map, as opposite to the chessboard. If there were maps, and scaled HP for structures made for 20v20 then this wouldn't be true. But since many maps are made for 6v6 to 10v10, they are half the size and complexity that a 20v20 game would require to function as elequently as 6v6 to 10v10. IMO 12v12 is pushing it.
Not really much the alien team can do in this situation. If more than two Skulks need to react and withdraw from the forward position, they've already lost valuable grounds at other places. Same goes for Marines, when Skulks bypass and group to get that new arms lab or an IP. But then a lost arms lab is never as damaging as loosing the natural RT.
@Ixian Doesn't matter why a long game is well...longer than usual, its at least something out of the ordinary for me that is much better than the ordinary < 10 minute games. While I'm not looking for something as brain dead as woozas server, I don't care if people perform poorly as much in 16-24 if the game is in a state where either team can win. It's got to the point where I don't even play this game anymore because I just couldn't be bothered dealing with the poor balance system but I like to keep in touch with the updates to see what changes in the future. As a pubby, any bad hit detection or any glitches and bugs or design flaws felt like annoyances but I could look past them because I enjoyed the core gameplay so much but when it comes to this balance system, it got me to quit the game outright. I don't know what happened in the past few months but its turned into shit, might've just been my luck but it sucked.
I don't agree that 1-2-2-2 is the way to go. It really depends on what players are available and how large the skill gap between each player is. Using this method, you are not really solving anything but trading the problems of one solution for another. It needs to be more flexible. I think ranking players by tier could be better but it might just produce the same results as score averaging.
No. What I meant is by grouping, for example, 0-500 as a tier, 1000-1200 as another tier and so on, where there is less emphasis on raw numbers and more on match ups. It could very well result in the exact same teams as normal shuffling, but it seems fairer to me.
Also, your insistence on elo/hive being a reflection of your own skill does not hold no matter how much you wish. It is a team game. What your team does matters as much as what you do.
Having been on a team (and actually taking the championship in the ENSL S5 Premier Division), I would consider myself knowledgable about the game, and its mechanics, theories, and strategies, but thank you for the concern, misplaced as it is.
Edit: Misinterpreted mendaps statement, and has been removed
I did say that it's stupid to argue about hive while it uses a single number for all 4 roles (marine, alien, and both comms).
And if it did that, it can't predict who's going to command for doing the shuffle to use the correct elo to balance.
In any case it should have been a hidden number, and it being based on community (pub) servers is another weakness, it would work for actual matchmaking, where you can't "choose" your opponent.
Meh.
Maybe try reading your own posts. And no one is talking about game knowledge, just skill evaluation, but sure, go toot your own horn at every available opportunity.
There is a well know puzzle about choosing a die in a set and still losing no matter which you picked.
https://en.wikipedia.org/wiki/Nontransitive_dice
Lets have 3 teams/dice:
A:
200
200
600
600
700
700
B:
100
100
500
500
900
900
C:
300
300
400
400
800
800
Assume that a player with higher skill always kills a player with lower skills, which team/die would you pick?
The answer is that B in average beats A, C in average beats B, but A in average beats C.
The winning move is pick the die after the other player.
So the question is; what would an optimal shuffle heuristic actually optimize?
Average and deviation are easy to calculate, but that really is not an argument for using them.
You dragged me into, talking down to me, I say ok, and ethos beccomes a thing. Im gonna toot my horn, when you try to break it. Leave personas out of it, and we wont have to talk about each others achievements, rather only the subject at hand which is much more interesting.
I disagree with elo and the way it is used and calculated - hence the " ". The statement was also to respond to your wish for a ranking by tier, to which I added that I believed such a thing was already in place. I apolegise if I have been too unclear about these things.
Now I am confused. Are you being sarcastic or sincere on this? You have been repeating it so I took it as being sincere and it really is not true. It should work out in the aggregate (because large sample size should in theory cover everything), but really, you need your whole team to live up to their skill and most importantly, your commander to do at least the minimum. Even the best players will be limited by a bad commander. It is much easier to carry a 200 elo deadweight on the field than to carry him in the chair, for example.
But if you were being sarcastic, my entire point is moot. In that case, whatever.
This is the biggest flaw imho of any of the shuffles out there .. none take the commander into account .. the shuffle should never shuffle until commanders are in place .. and not shuffle the commanders. Regardless of all the data you collect to try get a balanced game going, unless you can weight the commanders skill based on some metrics .. (eg. commander akill being based on the skill of their fieldteam and the oppositions comm+fieldteam skill)
As for those 3k guys joining pub servers .. imho .. it's on THEM to pull back when they can obviously stomp a pub game and carry the entire team .. especially with such broken match making systems.. when you see one marine with 40-2 k/d on a public server where the next highest kill count is ~6 .. then that player needs to either pull their punches or go to another server .. it happens so often, bored pro lads ..I'd be on for handicapping these guys
offtopic
That might be a decent idea for a server mod :P stompers can get landed with a handicap to keep them honest .. anyone over hive skill X difference with server average.. if they hit a 4/1 k/d ratio get hit with .. I dunno, armour/weap/speed penalty.. surley stompers couldnt complain too much .. they'd get a decent challenge on pub games and no doubt a mod like this wouldn't affect anyones ratings .. which in itself is a bit unfortunate since if it became popular would break itself :P
/offtopic
If the player with the higher skill ALWAYS kills the lower player, than team B has two players who NEVER die. They could literally walk into the hive/marine base and win the game outright. Which is more or less exactly what happens when you have 4k players.
A 2k player with a 130 > a 2k player with 80. How can you balance lag compensation.
Shine's shuffle has a setting to ignore commanders. I wish everyone enabled it.