Actually I don't think the score in game reflects if a player is good or not. There are a lot of stuff that is very important that doesn't give you any point. Like standing in some hallway trying to block a retreating fade from getting back to the hive, etc. There are to many things to take in to account in a game like NS2, and getting a +2 (x5?) for welding a resource tower is most of the time not worth it when you could (should) be out killing life forms, RT's and upgrades instead. Also, this isn't CoD, score doesn't make you a "good" player.
ELO would most preferably only be based on wether you win or lose the round, and decide how many points you gain depending on the ELO of the other team.
while wins/losses may be the preferred way to do it, im not sure you can say just win/loss record could really reflect if an individual player is "good" all the time, either. Someone could easily be carried, especially if the system allows individual players to join another group of players. I am not saying score is the answer to what should contribute to some kind of player rating, just giving an idea. There are positive and negative sides to each way it could potentially be implemented. This is assuming some kind of rating/matchmaking is even what the developers want. They could easily leave it out and allow the community to use this tool the way people want, setting up their own even matches(though i would very much like match making for pubs).
In game score's entire point seems to be for reflecting how much a player contributes in a game. In your examples, someone blocking a hallway for a fade would more than likely assist with the kill, gaining some points(nsStats servers at least give points for assists.) If they happen to be blocking a way the fade didnt go, should they get points for potentially being in a good position? I'm not sure about that, maybe? Welding/healing most definitely are intended to give points; they just added that semi recently so that is clearly their intention. Maybe you could argue that it is a waste of time to weld certain structures since it takes so long, but that is different from saying it doesn't contribute to the team. All of the other things you listed as what marines should be doing instead give points as well, so all of those add to a players score.
I kinda find it hard to buy that score doesn't reflect good play, as anything a good player will be doing will give them points. Can you think of examples in this game where someone is doing something useful for the team yet not getting any points for it? If so, maybe the problem is certain activities arent being given enough credit in the score.
Well, first of all the scores that are given aren't really balanced, as in they don't really reflect if you actually did something good or if you spend 15 minutes in base welding every structure to full hp when there's a hive with 1% hp 2 rooms away.
I don't believe that individual rating based on scores is the way to go, as there is just too much going on and if I have to see a "+5" for every single thing I do (scouting, positioning) I would have a mental breakdown.
Thing is with simple win/loss rating the bad player will eventually get evened out to his appropriate level, as a 1300 ELO-player (1500 is usually the starting point, 0 games won, 0 games lost) will be greatly outplayed when he's got 1800+ ELO-rating and is facing similar opponents. Over time it all evens out.
Maybe when this system is introduced we could have some kind of "select skill level" from 1-5 where you would automatically start with 1500 up to 1800 ELO or something.
First, "ELO" should be written "Elo" - It's not an acronym, it's a person's name. Arpad Elo. When you write it in all caps, you're talking about the rock band from the seventies.
Second, Elo doesn't refer to just any old method of keeping score. It refers to the system that Arpad Elo invented, or one of several very similar systems that are derivatives of his system. The key feature of Elo is that it's not based on things like k/d or score or win/loss record. It's based strictly on three factors: wins, losses, and (this is key) the strength of the opponents that you win or lose to. You can read more about it here. If you don't already know that you don't need to read that article, then you need to read that article.
The reason that games like chess and LoL use Elo to rank players is because it's about as good a measurement of what we think of as "skill" as anyone has yet devised. In particular: for single-player games, it measures the probability with which a given player will beat a given other player. And for multiplayer games (which the original Elo system doesn't handle, but which can be handled by some relatively straight-forward extensions of the system), it measures the degree to which having a given player on a team increases the probability that the team will win. In other words, it measures your contribution to your teams' chance of victory without having to measure any specific components that might go into that victory (like k/d or score or how well you lead a team or follow orders).
Elo is the shizzle. I very much hope that UWE implements something like it. It would be superior to any other kind of ranking or metric, including the commonly suggested ones such as time played, score, k/d, and win/loss.
The problem with Elo is that it was designed for 1v1 play. The more players added to each side the more likely that a gain/loss is unjust. In a game of 12v12, even if you're amazing, 2k elo, and everyone else is 1k, there is almost no hope of you carrying that game on marines, you will lose anyway, and a sizable loss of Elo will come included.
Well, first of all the scores that are given aren't really balanced, as in they don't really reflect if you actually did something good or if you spend 15 minutes in base welding every structure to full hp when there's a hive with 1% hp 2 rooms away.
seems like a contrived example: if aliens are doing that much structure dmg to your base, how is welding not contributing? it isnt like you get points for welding things that are fully hp. I also have a hard time believing a player wouldnt generate any points while they are scouting/positioning themselves. It isnt like people just sprint through maps scouting, doing nothing. They eventually kill/build/etc... If score was somehow imbalanced, why couldnt they balance it?
Anyways, my thoughts on possibly including more than just win/loss would be that it accounts for individual contribution vs team based. Obviously win/loss would still be part of it.
Like someone who is a killing machine murdering all the aliens/marines would get slightly more rating than some scrub who is 0-27 sitting in base, yet is on the winning team. Then hopefully they would rise in rating faster than the others, placing them with people of their skill level sooner.
It could very easy be based on pure win/loss if say teams of 6/8/whatever were set and you had to queue as that team. Win/loss on a team level makes a lot of sense. Not quite as much for personal ratings, IMO. Obviously, as crazy eddie pointed out, it also depends on the skill level of the other team.
as much as id like some kind of matchmaking/player rating, i am doubtful it would even be implemented for exactly these reasons: ns2 is complex and there is a lot of things people can do to contribute to wins/losses.
Yes it is kind of a contrived example but the point remains valid. Gamesense is a BIG part of this game and a lot of people have a really bad game sense even if they have 300+ score. There are so much stuff in this game that is situational and score doesn't really meassure how well some one has been doing at all. There are just so many examples that prove that scores mean nothing. If you land one bullet on a fade, and it dies, do you _really_ earn the +20? Fana had some interesting ideas for scores with getting points for scouting and what not, but I'd rather shoot myself than seeing "+2" and "+5" pop up on my screen every second.
And I believe Elo (sorry!) could be used for NS2 because you win or lose as a team. You do not get different rating depending on how well you as an individual performed, you gain it if you win and in order to win you need to work together as a team. If you have one guy that is horrible in your team but still manage to win, that horrible player will, with time, be given a rating that is accurate to his skill if he's playing enough games, and will be queued with players of similar skill.
Gamesense is a BIG part of this game and a lot of people have a really bad game sense even if they have 300+ score. There are so much stuff in this game that is situational and score doesn't really meassure how well some one has been doing at all. There are just so many examples that prove that scores mean nothing. Fana had some interesting ideas for scores with getting points for scouting and what not, but I'd rather shoot myself than seeing "+2" and "+5" pop up on my screen every second.
Someone isnt going to get 300+ score in a normal game unless the game lasts forever. Even still, they were doing stuff, be it killing, building, or welding/healing for those points. That is contributing to the team. I also dont think you could truly give all these examples of score not meaning anything, simply because doing anything that is even remotely useful gives points. List one or two examples, or even list something that doesnt contribute to score, aside from map awareness. The only point i think you really have is that scouting or map awareness contributes to the team when it isnt currently counted towards score. Whether or not gaining points show on the screen is fairly irrelevant.
If you land one bullet on a fade, and it dies, do you _really_ earn the +20?
I would definitely say yes. Considering that assists can give points (at least on servers with nsStats), anyone who contributed to the fade kill gets points. How could you say that killing a fade, even if you only get that last shot, doesnt contribute? You even used the example of blocking off a retreating fades path earlier, which is basically attempting to get that last hit.
This all seems off-topic, anyways. My point was that we already have a system in game to measure individual contribution, which is score. The reason it is there is to emphasize that k/d ratio isnt everything, rewarding people for things aside from kills. I think win/loss would still be weighted far more than score anyways, as it doesnt make sense for someone who losses to gain some sort of rating. And since they broke down entities to the player level, there needs to be some kind of weighting at the player level, even if just a little. IF they even want to add matchmaking/rating:P
The problem with Elo is that it was designed for 1v1 play.
TrueSkill is (essentially) an enhanced version of Elo. It is designed to handle both 1v1 and team play. It works very well.
The more players added to each side the more likely that a gain/loss is unjust. In a game of 12v12, even if you're amazing, 2k elo, and everyone else is 1k, there is almost no hope of you carrying that game on marines, you will lose anyway, and a sizable loss of Elo will come included.
As long as the following are all true:
A team with eleven 1000 players and a 2000 player has at least a slightly better than 50/50 chance versus a team with twelve 1000 players (no matter how slight)
A team with eleven 1000 players and a 1500 player has at least a slightly better than 50/50 chance versus a team with twelve players (no matter how slight)
The team with the 2000 player has a larger advantage over the 12x1000 team than the team with the 1500 player does
... then over time, everyone's rating will adjust such that their ratings reflect the degree to which their presence on a team improves that team's chances. Assuming you're using a system that extends the basic concepts of Elo to include team play. Like TrueSkill, or something else like it.
The problem with Elo is that it was designed for 1v1 play.
TrueSkill is (essentially) an enhanced version of Elo. It is designed to handle both 1v1 and team play. It works very well.
The more players added to each side the more likely that a gain/loss is unjust. In a game of 12v12, even if you're amazing, 2k elo, and everyone else is 1k, there is almost no hope of you carrying that game on marines, you will lose anyway, and a sizable loss of Elo will come included.
As long as the following are all true:
A team with eleven 1000 players and a 2000 player has at least a slightly better than 50/50 chance versus a team with twelve 1000 players (no matter how slight)
A team with eleven 1000 players and a 1500 player has at least a slightly better than 50/50 chance versus a team with twelve players (no matter how slight)
The team with the 2000 player has a larger advantage over the 12x1000 team than the team with the 1500 player does
... then over time, everyone's rating will adjust such that their ratings reflect the degree to which their presence on a team improves that team's chances. Assuming you're using a system that extends the basic concepts of Elo to include team play. Like TrueSkill, or something else like it.
Microsoft owns Truskill.
Use in other projects
TrueSkill is patented,[3] and the name is trademarked, so therefore it is limited to Microsoft projects and commercial projects that obtain a license to use the algorithm.
I never heard of TrueSkill before but it sounds essentially the same as how Starcraft 2 uses MMR (match-making rank/rating) - each played has a value for "skill" (like Elo) but also has a value for "uncertainty", which increases when they win games the system thought they would lose and vice versa. I don't imagine it would be that hard for clever people like the UWE team to extrapolate enough information about these systems to make something similar for NS2, or at least find someone new who could do that for them.
Basing match-making ranks on anything but pure win/loss (factoring in opponent rating) sounds crazy, winning or losing is the only constant between games. If you used player scores, no matter how much you tweaked the scoring system, there would be all sorts of problems with short games compared to long games, teams might camp and rack up kills without finishing the game, strategies would be developed to maximise score differential rather than just your chance of winning, etc.
I never heard of TrueSkill before but it sounds essentially the same as how Starcraft 2 uses MMR (match-making rank/rating) - each played has a value for "skill" (like Elo) but also has a value for "uncertainty", which increases when they win games the system thought they would lose and vice versa. I don't imagine it would be that hard for clever people like the UWE team to extrapolate enough information about these systems to make something similar for NS2, or at least find someone new who could do that for them.
Basing match-making ranks on anything but pure win/loss (factoring in opponent rating) sounds crazy, winning or losing is the only constant between games. If you used player scores, no matter how much you tweaked the scoring system, there would be all sorts of problems with short games compared to long games, teams might camp and rack up kills without finishing the game, strategies would be developed to maximise score differential rather than just your chance of winning, etc.
Many of them are variations of the Elo-rating system, usually to address its well-known flaws. Since winning is the primary objective of the game, it makes the most sense to use that as either the sole or primary method to determine a players rating. The only non-win tweak I would make in NS2 is to distribute the rating points gained/lost within a team via the relative player scores (i.e. the player on the top of the scoreboard would gain more or lose less of their rating then one on the bottom).
I don't care for the ELO or trueskill system, ELO doesn't seem to make even matches, and trueskill locks you at levels which you don't belong
I loved whatever they used for halo 2, it was 2-3 wins rank up, 1-2 losses in a row, rank down. Basically the point was, win as a team and you'll rank up, keep losing as a team and you'll rank down. There wasn't any insane math formulas (most of which can be abused)
on that note, screw consoles. Anyway I'm sure they will figure out a system that works well
Hey guys, just bumping, almost a month, care to give us a taste on any news or insider info of what's going on?? Very excited for this as I think it will be a major improvement to the game and everyones enjoyment
Stumbled across an old forum topic that redirected a link to here if we wanted to help support the idea.
Well, I'm here! YES! This system sounds awesome, and I hope it comes sooner than soon.
If you go by the individual scores affecting your 'rating', then something that i was thinking about is if it's in anyway possible to track k/a/d ratio for an individual lifeform, if we're going to have x-amount of random players, it'll be somewhat unfortunate if there's teams full of good fades, but horrible lerks, or skulks. If you choose to go this route, there's so many different things to take into consideration.
But yeah, did the news post seem like it was missing all the valuable information to anyone else? Without Hugh's 'longer' reply i've been mostly dissapointed by the whole thing..
Comments
while wins/losses may be the preferred way to do it, im not sure you can say just win/loss record could really reflect if an individual player is "good" all the time, either. Someone could easily be carried, especially if the system allows individual players to join another group of players. I am not saying score is the answer to what should contribute to some kind of player rating, just giving an idea. There are positive and negative sides to each way it could potentially be implemented. This is assuming some kind of rating/matchmaking is even what the developers want. They could easily leave it out and allow the community to use this tool the way people want, setting up their own even matches(though i would very much like match making for pubs).
In game score's entire point seems to be for reflecting how much a player contributes in a game. In your examples, someone blocking a hallway for a fade would more than likely assist with the kill, gaining some points(nsStats servers at least give points for assists.) If they happen to be blocking a way the fade didnt go, should they get points for potentially being in a good position? I'm not sure about that, maybe? Welding/healing most definitely are intended to give points; they just added that semi recently so that is clearly their intention. Maybe you could argue that it is a waste of time to weld certain structures since it takes so long, but that is different from saying it doesn't contribute to the team. All of the other things you listed as what marines should be doing instead give points as well, so all of those add to a players score.
I kinda find it hard to buy that score doesn't reflect good play, as anything a good player will be doing will give them points. Can you think of examples in this game where someone is doing something useful for the team yet not getting any points for it? If so, maybe the problem is certain activities arent being given enough credit in the score.
I don't believe that individual rating based on scores is the way to go, as there is just too much going on and if I have to see a "+5" for every single thing I do (scouting, positioning) I would have a mental breakdown.
Thing is with simple win/loss rating the bad player will eventually get evened out to his appropriate level, as a 1300 ELO-player (1500 is usually the starting point, 0 games won, 0 games lost) will be greatly outplayed when he's got 1800+ ELO-rating and is facing similar opponents. Over time it all evens out.
Maybe when this system is introduced we could have some kind of "select skill level" from 1-5 where you would automatically start with 1500 up to 1800 ELO or something.
First, "ELO" should be written "Elo" - It's not an acronym, it's a person's name. Arpad Elo. When you write it in all caps, you're talking about the rock band from the seventies.
Second, Elo doesn't refer to just any old method of keeping score. It refers to the system that Arpad Elo invented, or one of several very similar systems that are derivatives of his system. The key feature of Elo is that it's not based on things like k/d or score or win/loss record. It's based strictly on three factors: wins, losses, and (this is key) the strength of the opponents that you win or lose to. You can read more about it here. If you don't already know that you don't need to read that article, then you need to read that article.
The reason that games like chess and LoL use Elo to rank players is because it's about as good a measurement of what we think of as "skill" as anyone has yet devised. In particular: for single-player games, it measures the probability with which a given player will beat a given other player. And for multiplayer games (which the original Elo system doesn't handle, but which can be handled by some relatively straight-forward extensions of the system), it measures the degree to which having a given player on a team increases the probability that the team will win. In other words, it measures your contribution to your teams' chance of victory without having to measure any specific components that might go into that victory (like k/d or score or how well you lead a team or follow orders).
Elo is the shizzle. I very much hope that UWE implements something like it. It would be superior to any other kind of ranking or metric, including the commonly suggested ones such as time played, score, k/d, and win/loss.
tldr; need a better system than Elo.
Anyways, my thoughts on possibly including more than just win/loss would be that it accounts for individual contribution vs team based. Obviously win/loss would still be part of it.
Like someone who is a killing machine murdering all the aliens/marines would get slightly more rating than some scrub who is 0-27 sitting in base, yet is on the winning team. Then hopefully they would rise in rating faster than the others, placing them with people of their skill level sooner.
It could very easy be based on pure win/loss if say teams of 6/8/whatever were set and you had to queue as that team. Win/loss on a team level makes a lot of sense. Not quite as much for personal ratings, IMO. Obviously, as crazy eddie pointed out, it also depends on the skill level of the other team.
as much as id like some kind of matchmaking/player rating, i am doubtful it would even be implemented for exactly these reasons: ns2 is complex and there is a lot of things people can do to contribute to wins/losses.
And I believe Elo (sorry!) could be used for NS2 because you win or lose as a team. You do not get different rating depending on how well you as an individual performed, you gain it if you win and in order to win you need to work together as a team. If you have one guy that is horrible in your team but still manage to win, that horrible player will, with time, be given a rating that is accurate to his skill if he's playing enough games, and will be queued with players of similar skill.
Someone isnt going to get 300+ score in a normal game unless the game lasts forever. Even still, they were doing stuff, be it killing, building, or welding/healing for those points. That is contributing to the team. I also dont think you could truly give all these examples of score not meaning anything, simply because doing anything that is even remotely useful gives points. List one or two examples, or even list something that doesnt contribute to score, aside from map awareness. The only point i think you really have is that scouting or map awareness contributes to the team when it isnt currently counted towards score. Whether or not gaining points show on the screen is fairly irrelevant.
I would definitely say yes. Considering that assists can give points (at least on servers with nsStats), anyone who contributed to the fade kill gets points. How could you say that killing a fade, even if you only get that last shot, doesnt contribute? You even used the example of blocking off a retreating fades path earlier, which is basically attempting to get that last hit.
This all seems off-topic, anyways. My point was that we already have a system in game to measure individual contribution, which is score. The reason it is there is to emphasize that k/d ratio isnt everything, rewarding people for things aside from kills. I think win/loss would still be weighted far more than score anyways, as it doesnt make sense for someone who losses to gain some sort of rating. And since they broke down entities to the player level, there needs to be some kind of weighting at the player level, even if just a little. IF they even want to add matchmaking/rating:P
As long as the following are all true:
- A team with eleven 1000 players and a 2000 player has at least a slightly better than 50/50 chance versus a team with twelve 1000 players (no matter how slight)
- A team with eleven 1000 players and a 1500 player has at least a slightly better than 50/50 chance versus a team with twelve players (no matter how slight)
- The team with the 2000 player has a larger advantage over the 12x1000 team than the team with the 1500 player does
... then over time, everyone's rating will adjust such that their ratings reflect the degree to which their presence on a team improves that team's chances. Assuming you're using a system that extends the basic concepts of Elo to include team play. Like TrueSkill, or something else like it.Basing match-making ranks on anything but pure win/loss (factoring in opponent rating) sounds crazy, winning or losing is the only constant between games. If you used player scores, no matter how much you tweaked the scoring system, there would be all sorts of problems with short games compared to long games, teams might camp and rack up kills without finishing the game, strategies would be developed to maximise score differential rather than just your chance of winning, etc.
I was TS38 in 2v2 but would come across TS45 and roll them, it wasn't the best system imo.
No system is perfect, but unless you were stomping TS50 or losing to TS25, that's still much better than NS2 atm.
thinly veiled brag
I loved whatever they used for halo 2, it was 2-3 wins rank up, 1-2 losses in a row, rank down. Basically the point was, win as a team and you'll rank up, keep losing as a team and you'll rank down. There wasn't any insane math formulas (most of which can be abused)
on that note, screw consoles. Anyway I'm sure they will figure out a system that works well
But I was bad so it's okay.
Well, I'm here! YES! This system sounds awesome, and I hope it comes sooner than soon.
But yeah, did the news post seem like it was missing all the valuable information to anyone else? Without Hugh's 'longer' reply i've been mostly dissapointed by the whole thing..
Anyway, as far as I know it's still being worked on.