I don't care about the statistics... I'm just tired of seeing games where all the 2000+ players get shuffled against a team of 1000 and below players simply because they have one 0-100 player (and sh_teamstats shows average team scores as even)
I'm tired of seeing 4.0 kdr players with hive scores below 2k.. I'm tired of seeing <1000 players who should have 2000+ easily...
I don't care about the statistics... I'm just tired of seeing games where all the 2000+ players get shuffled against a team of 1000 and below players simply because they have one 0-100 player (and sh_teamstats shows average team scores as even)
I'm tired of seeing 4.0 kdr players with hive scores below 2k.. I'm tired of seeing <1000 players who should have 2000+ easily...
But most of all I'm tired of one sided shuffles
The problem is very much that the 0-100 are playing in the same server as the 2000+. Even if they are all people who play together, know each other and pick teams manually and fairly, chances are that the game is going to suck. If you have a bad mix of players, the evaluation and balancing is not the problem.
... then you could just rely on your score as a metric of skill.
It's already a system that's weighted and mostly completed, counting things like assists and building.. just add more things like defending an RT under attack etc and you could rely on that.
The only reason I've thought that this might be a good measurement is that most of the time the players contributing the most to the team's win are at the top of the board due to the Score points - and very often they have poor KDR... so it's already counting things that aren't too abuse-able at least.
...It is inherently easier to get higher scores per minute and kills, if you are on the winning team. Not just because you will be leading in TECH, but you will also have the momentum and the map control.
So how do you account for that?
You account that by averaging the team SPM (this is the strict answer to your question), checking the deviation from the individual SPM on that match to the team SPM on that match, and weighting the awarded points based on that difference.
Give more skill points to those who contributed the most on the winning team.
Take less skill points from those who contributed the most in the losing team.
This + adding skill deviation minimization to FET will give you - my opinion - more exciting matches than we have today. A separate alien/marine SP rank will further improve on this.
After this is done the discussion will be about tuning the score points for constructing a gorge tunnel, giving some points to the comm when a building is built, etc. The optimal values can be interpreted and extracted from historical analysis of matches.
I don't care about the statistics... I'm just tired of seeing games where all the 2000+ players get shuffled against a team of 1000 and below players simply because they have one 0-100 player (and sh_teamstats shows average team scores as even)
I'm tired of seeing 4.0 kdr players with hive scores below 2k.. I'm tired of seeing <1000 players who should have 2000+ easily...
But most of all I'm tired of one sided shuffles
The problem is very much that the 0-100 are playing in the same server as the 2000+. Even if they are all people who play together, know each other and pick teams manually and fairly, chances are that the game is going to suck. If you have a bad mix of players, the evaluation and balancing is not the problem.
The game isn't going to suck nearly as bad if the 2000+ players get balanced onto separate teams...
It should never shuffle 2 or more 2000+ players against a team where the highest score is under 1500. Unfortunately that's something that happens a lot..
The rookies need too be split as much as the high level players do.. I'm sure most vets hate trying to carry a team of <1000 rookies.
You are correct. Teams should have an even distribution of skill numbers to make for the most meaningful game. This is widely agreed upon and discussed in multiple threads.
But how does this have anything, absolutely anything, to do with the metric by which the skill is measured? You would have the exact same problems if the skill system was based on something arbitrary like score-per-minute or kills-to-deaths ratio. You would still have players with different skill ratings, and the system would still have to place those players in teams based on those skill ratings. I do not understand how you associate the problem you are experiencing with skill being measured purely by win-loss. The only difference in changing this would be that the skill number would no longer truthfully represent a player's ability to win a game, but rather that player's tendency to accumulate the arbitrary measures you'd introduce.
Also, you mention that the win-loss measurement encourages stacking and discourages balancing. It does not, as has been discussed in a myriad of threads before. It's the very basis of an Elo ranking system that you cannot inflate your rating by playing against weaker opponents nor, conversely, deflate it by playing against stronger opponents.
The bottom line is this: kills, score, build time, you name it, can always be farmed without actual corresponding utility to the team. By definition, wins cannot be farmed without being useful towards the goal of the game, which is to win.
You are correct. Teams should have an even distribution of skill numbers to make for the most meaningful game. This is widely agreed upon and discussed in multiple threads.
But how does this have anything, absolutely anything, to do with the metric by which the skill is measured? You would have the exact same problems if the skill system was based on something arbitrary like score-per-minute or kills-to-deaths ratio. You would still have players with different skill ratings, and the system would still have to place those players in teams based on those skill ratings. I do not understand how you associate the problem you are experiencing with skill being measured purely by win-loss.
Because it's compounded by the players who have a hive score that is nowhere near what it should be. It seems to me like its two sides of the same problem.
The bottom line is this: kills, score, build time, you name it, can always be farmed without actual corresponding utility to the team. By definition, wins cannot be farmed without being useful towards the goal of the game, which is to win.
I know of a two players who prove that statement is false... One is really good, coordinates his team well, has great strategies when he commands, and is overall a great player... Last time i checked his hive score was under 1000... The second is a jerk who a lot of people in this community dislike due to his constant stacking, switching to the winning team, constant complaining about how other players suck, and he's not even that good.. His hive score was over 4000
And those are just examples at each end of the spectrum... My hive score is around 1100 yet players who are better than me in every way, aim, positioning, communication, coordinating with the team, etc.. Their score is lower..
I don't know maybe hive 2.0 will work better than the atrocity we have now.. I just don't think going purely by w/l will ever be successful at determining a players individual skill level... The whole notion that a player can get 80+ kills, kill 8+ resource towers, and have double (even triple) the score of anyone on the server, yet have his hive score drop if his team screws up and loses.. That just seems broken.
You're basing your arguments on personal anecdotes. They hold no weight. I can argue that milk is poisonous because once I drank some and the next day I felt ill. Does not make it true.
You are unable to distinguish individual, unlikely scenarios from the big picture.
I know of a two players who prove that statement is false... One is really good, coordinates his team well, has great strategies when he commands, and is overall a great player... Last time i checked his hive score was under 1000... The second is a jerk who a lot of people in this community dislike due to his constant stacking, switching to the winning team, constant complaining about how other players suck, and he's not even that good.. His hive score was over 4000
Did you really just use a player that purposefully exploits and breaks the system to prove that W/L does not work?
I don't know maybe hive 2.0 will work better than the atrocity we have now.. I just don't think going purely by w/l will ever be successful at determining a players individual skill level... The whole notion that a player can get 80+ kills, kill 8+ resource towers, and have double (even triple) the score of anyone on the server, yet have his hive score drop if his team screws up and loses.. That just seems broken.
Kill death rate is already including in the win/loss metric. Kill death rate correlates strongly with winning, which means they are related.
Yeah stop all this crappy nit picking and bitching...
You know you are terrible as NS2 when... You use "cold hands" as an excuse for your poor aim!
cold hands is a legit excuse. warm shower increases acc by increases reaction time by 5%
Fixed that for you. But for real... having cold hands does a lot to your reaction speed which directly affects your aim.
Agreed. Anecdotal evidence suggests that accuracy plummets to 18~% when window is left slightly ajar during the winter. Solutions: turn up heater for warm hands and hot everything else, wear gloves and risk accidental key presses, sit on hands during respawn for temporary relief, or simply close window.
In response to the argument that ppl are seeing players better than their hive skill may indicate, the more they play, the more accurate their elo will become.
I can remember several players, but one in particular, who was carrying his team hard. I didn't recognize this player, suspected smurf, checked his steam profile. Was legit accnt hundreds of hours in other games, csgo included. His elo was low, as to be expected with a new player.
I encountered this player several times over a few months and his elo is now well reflective of the carry I witnessed at first. Was a naturally good player who had just started logging stats, and they eventually caught up with him.
You're basing your arguments on personal anecdotes. They hold no weight. I can argue that milk is poisonous because once I drank some and the next day I felt ill. Does not make it true.
You are unable to distinguish individual, unlikely scenarios from the big picture.
Anecdotes based on his play time and observed events and supported by other players.
Anecdotes sourced from the dissatisfaction of how FET/shuffle causes more imbalanced matches than balanced ones.
Consider true pub servers such as IBIS or DMD.
Do you think it's joyful having about 1 in every 3 matches where alien team cannot even get to Fade res?
Do you think it's normal a game not lasting for more than 8 minutes? (painful 8 minutes btw).
I and many others can tell you these things happen not due to brilliant strategies from the winning team.
Anecdotes from how player's skills can be misrepresented. (Player's who have stable skill points)
Just the fact that people look at the skill points as bins/bands (1000-1500-2000-3000) prove the system is not entirely reliable.
Anecdotes which sustain logical testing if you consider how the tri-component rank system (score point per action, rank algorithm, balancing algorithm) performs in the game environment (team skill points being attributed to individual player performance, no match-making enforced, uncoupled server populations, asymmetrical lifeforms).
These anecdotes are not random/cherry picked. They are common place.
Here's an anecdote:
Calego Captain Server before the hive skill showing next to the name was great. Captains actually picked players by their competency rather than the number thing. Matches were fun. Then, it all changed. Now you had the expectation imposed by the number plus the frustration it caused when the expectation was broken.
Same thing happen in pub matches now. And it causes distrust to some players.
Regarding the +4000 player mofo mentioned, the player told me had Hive rank been revised earlier, his calculations estimated an additional 200 hours of game play to achieve his current skill point number rather than only 250 hours when going from 3.3k to 5k. W/L can be farmed as much as SPM, KDR, etc.
The best solution for all this would be to make the hive skill database public (assuming it records every match stat) and let people fiddle with it.
The best solution for all this would be to make the hive skill database public (assuming it records every match stat) and let people fiddle with it.
Fiddle with it how? I have an incomplete snapshot of the hive data from March 2015. In what ways might you suggest I fiddle with it? I can't share the data but if you have suggestions on how to fiddle with the data I am all ears.
Yes, NovoRei, anecdotes. Anecdotes by people who feel unfairly represented by their rating (even if it would be accurate) and have already biased themselves against the system, causing them to overlook the system when it works and concentrate on individual faulty scenarios, blaming the skill system when they don't even understand how it works.
Yeah stop all this crappy nit picking and bitching...
You know you are terrible as NS2 when... You use "cold hands" as an excuse for your poor aim!
cold hands is a legit excuse. warm shower increases acc by increases reaction time by 5%
Fixed that for you. But for real... having cold hands does a lot to your reaction speed which directly affects your aim.
Haha I thought I was the only one! In order for me to aim at my best I take a hot shower before loading up NS2
Heatbags are almost universal among korean starcraft progamers, mainly due to the fact that the studio temperatures are never reliable and adrenaline can cause cold hands.
Yeah stop all this crappy nit picking and bitching...
You know you are terrible as NS2 when... You use "cold hands" as an excuse for your poor aim!
cold hands is a legit excuse. warm shower increases acc by increases reaction time by 5%
Fixed that for you. But for real... having cold hands does a lot to your reaction speed which directly affects your aim.
Haha I thought I was the only one! In order for me to aim at my best I take a hot shower before loading up NS2
Heatbags are almost universal among korean starcraft progamers, mainly due to the fact that the studio temperatures are never reliable and adrenaline can cause cold hands.
I just crank up the radiators tho
Kind of makes me wish I took the "adrenaline" upgrade prior to playing ns2
Having headach and litte sleep makes you suck at ns2....
And that moment when thats the only thing you've had for 2 weeks and then suddenly you don't makes you a hardcore pro! Im telling you m8's that was me this morning.
Soul_RiderMod BeanJoin Date: 2004-06-19Member: 29388Members, Constellation, Squad Five Blue
edited February 2016
So as this thread is now about the skill system, I thought I'd post one of my bugbears with the current system in this thread:
Whitelisting.
Earlier today I played 4 great rounds of NS2 on a server where the only mod running is UWE Hotfix. I was on the winning team, 3 out of 4 games, even featuring 2 games with a positive K/D, 16/10 & 12/9. Not much for most, but enough of a rarity for me to be really proud anyway.
The games were proper games of NS2 as well, we had 2 rookies on each team, we had comms who talked and teams who communicated. All the games were proper ding-dong games, the kind that we all really play NS2 for.
And for all the fun we all had, rookies and vets alike, and all our perfomances etc, none of it mattered, because this server is not whitelisted.
The point of this post or TL:DR
What is the point of a ranking system that does not automatically include completely vanilla servers? Only modded servers should need whitelisting.
Another problem with hive, it was implemented poorly. Whitelisting being one reason. Wooza's, a popular server, from time to time looses its whitelisting. Hive 2.0 will have unique server keys to resolve this.
Comments
I'm tired of seeing 4.0 kdr players with hive scores below 2k.. I'm tired of seeing <1000 players who should have 2000+ easily...
But most of all I'm tired of one sided shuffles
The problem is very much that the 0-100 are playing in the same server as the 2000+. Even if they are all people who play together, know each other and pick teams manually and fairly, chances are that the game is going to suck. If you have a bad mix of players, the evaluation and balancing is not the problem.
You account that by averaging the team SPM (this is the strict answer to your question), checking the deviation from the individual SPM on that match to the team SPM on that match, and weighting the awarded points based on that difference.
Give more skill points to those who contributed the most on the winning team.
Take less skill points from those who contributed the most in the losing team.
This + adding skill deviation minimization to FET will give you - my opinion - more exciting matches than we have today. A separate alien/marine SP rank will further improve on this.
After this is done the discussion will be about tuning the score points for constructing a gorge tunnel, giving some points to the comm when a building is built, etc. The optimal values can be interpreted and extracted from historical analysis of matches.
Aaand now there's arguing
Yeah stop all this crappy nit picking and bitching...
You know you are terrible as NS2 when... You use "cold hands" as an excuse for your poor aim!
The game isn't going to suck nearly as bad if the 2000+ players get balanced onto separate teams...
It should never shuffle 2 or more 2000+ players against a team where the highest score is under 1500. Unfortunately that's something that happens a lot..
The rookies need too be split as much as the high level players do.. I'm sure most vets hate trying to carry a team of <1000 rookies.
You hear that sound?
Calling all forum mods @IronHorse @Asraniel @Decoy @Narfwak @GhoulofGSG9
You are correct. Teams should have an even distribution of skill numbers to make for the most meaningful game. This is widely agreed upon and discussed in multiple threads.
But how does this have anything, absolutely anything, to do with the metric by which the skill is measured? You would have the exact same problems if the skill system was based on something arbitrary like score-per-minute or kills-to-deaths ratio. You would still have players with different skill ratings, and the system would still have to place those players in teams based on those skill ratings. I do not understand how you associate the problem you are experiencing with skill being measured purely by win-loss. The only difference in changing this would be that the skill number would no longer truthfully represent a player's ability to win a game, but rather that player's tendency to accumulate the arbitrary measures you'd introduce.
Also, you mention that the win-loss measurement encourages stacking and discourages balancing. It does not, as has been discussed in a myriad of threads before. It's the very basis of an Elo ranking system that you cannot inflate your rating by playing against weaker opponents nor, conversely, deflate it by playing against stronger opponents.
The bottom line is this: kills, score, build time, you name it, can always be farmed without actual corresponding utility to the team. By definition, wins cannot be farmed without being useful towards the goal of the game, which is to win.
Because it's compounded by the players who have a hive score that is nowhere near what it should be. It seems to me like its two sides of the same problem.
I know of a two players who prove that statement is false... One is really good, coordinates his team well, has great strategies when he commands, and is overall a great player... Last time i checked his hive score was under 1000... The second is a jerk who a lot of people in this community dislike due to his constant stacking, switching to the winning team, constant complaining about how other players suck, and he's not even that good.. His hive score was over 4000
And those are just examples at each end of the spectrum... My hive score is around 1100 yet players who are better than me in every way, aim, positioning, communication, coordinating with the team, etc.. Their score is lower..
I don't know maybe hive 2.0 will work better than the atrocity we have now.. I just don't think going purely by w/l will ever be successful at determining a players individual skill level... The whole notion that a player can get 80+ kills, kill 8+ resource towers, and have double (even triple) the score of anyone on the server, yet have his hive score drop if his team screws up and loses.. That just seems broken.
You are unable to distinguish individual, unlikely scenarios from the big picture.
cold hands is a legit excuse. warm shower increases acc by 5%
Kill death rate is already including in the win/loss metric. Kill death rate correlates strongly with winning, which means they are related.
Fixed that for you. But for real... having cold hands does a lot to your reaction speed which directly affects your aim.
I can remember several players, but one in particular, who was carrying his team hard. I didn't recognize this player, suspected smurf, checked his steam profile. Was legit accnt hundreds of hours in other games, csgo included. His elo was low, as to be expected with a new player.
I encountered this player several times over a few months and his elo is now well reflective of the carry I witnessed at first. Was a naturally good player who had just started logging stats, and they eventually caught up with him.
Anecdotes based on his play time and observed events and supported by other players.
Anecdotes sourced from the dissatisfaction of how FET/shuffle causes more imbalanced matches than balanced ones.
Consider true pub servers such as IBIS or DMD.
Do you think it's joyful having about 1 in every 3 matches where alien team cannot even get to Fade res?
Do you think it's normal a game not lasting for more than 8 minutes? (painful 8 minutes btw).
I and many others can tell you these things happen not due to brilliant strategies from the winning team.
Anecdotes from how player's skills can be misrepresented. (Player's who have stable skill points)
Just the fact that people look at the skill points as bins/bands (1000-1500-2000-3000) prove the system is not entirely reliable.
Anecdotes which sustain logical testing if you consider how the tri-component rank system (score point per action, rank algorithm, balancing algorithm) performs in the game environment (team skill points being attributed to individual player performance, no match-making enforced, uncoupled server populations, asymmetrical lifeforms).
These anecdotes are not random/cherry picked. They are common place.
Here's an anecdote:
Calego Captain Server before the hive skill showing next to the name was great. Captains actually picked players by their competency rather than the number thing. Matches were fun. Then, it all changed. Now you had the expectation imposed by the number plus the frustration it caused when the expectation was broken.
Same thing happen in pub matches now. And it causes distrust to some players.
Regarding the +4000 player mofo mentioned, the player told me had Hive rank been revised earlier, his calculations estimated an additional 200 hours of game play to achieve his current skill point number rather than only 250 hours when going from 3.3k to 5k. W/L can be farmed as much as SPM, KDR, etc.
The best solution for all this would be to make the hive skill database public (assuming it records every match stat) and let people fiddle with it.
Haha I thought I was the only one! In order for me to aim at my best I take a hot shower before loading up NS2
I just crank up the radiators tho
Kind of makes me wish I took the "adrenaline" upgrade prior to playing ns2
And that moment when thats the only thing you've had for 2 weeks and then suddenly you don't makes you a hardcore pro! Im telling you m8's that was me this morning.
Whitelisting.
Earlier today I played 4 great rounds of NS2 on a server where the only mod running is UWE Hotfix. I was on the winning team, 3 out of 4 games, even featuring 2 games with a positive K/D, 16/10 & 12/9. Not much for most, but enough of a rarity for me to be really proud anyway.
The games were proper games of NS2 as well, we had 2 rookies on each team, we had comms who talked and teams who communicated. All the games were proper ding-dong games, the kind that we all really play NS2 for.
And for all the fun we all had, rookies and vets alike, and all our perfomances etc, none of it mattered, because this server is not whitelisted.
The point of this post or TL:DR
What is the point of a ranking system that does not automatically include completely vanilla servers? Only modded servers should need whitelisting.