lot of people , always complain about unbalance games

all i can say is , if you press autojoin , its 50% you win / lose "CTF"

whe need a , globel stat server , that can make balance teams

it would be ground breaking , if it would work ..

even bether woulld be , artificial intelligence weapons , that are based on stats
if you high in only nex kills , you next nex weapon would be denied , so you need to pick a other gun , whe need need global stats for that , so whe can make the most balance , unique pubic ctf game there is

"Always balance CTF games based on stats"

I'd support tracking stats from the last game to balance teams the next game, but I'm completely against global, persistent stat-tracking. It's also impractical without some registration system which I would also be against. If you just use the stats from the last game instead, you can split up the best players for the start of the next one and see how that works out. If it doesn't, you'll have a new set of "best" players to split for the next match. People don't always play to the same level at all times, so even in that regard global stats wouldn't be representative (you could try keeping separate DM, TDM, CTF, KH, LMS, etc stats, but it's not worth the effort).

Blocking weapons because you're using it too much is just a bad idea, sorry. What next? Deny the laser to people who use it too much to move? Slow people down who are too fast? The whole idea of penalizing people because they're good is such an abysmally bad idea that it's not even worth considering. For the record, I HATE that some people can almost always snipe me yet I can't aim for shit. I have no chance against certain players if they have a nex because they will kill me 99/100 times... as much as I'll swear when it happens, I wouldn't vote for a change that prevents them from getting the nex just to give me an advantage.




But yeah, team balance sucks. I suggested a system to dynamically balance teams when the score difference reaches a threshold based on the match type and the number of players... it's buried somewhere in the forum next to Atlantis.

It's also impractical without some registration system which I would also be against.

IIRC, this is because you were paranoid about someone being able to track your activity or something?

Anyway, what if it were opt-in?

If it were opt-in, most people wouldn't register and it'd be mostly useless then...

however, in theory one can save stats on the player's machine as cvars, that get stored to config.cfg. Sure, then everyone could modify his own stats, but the only effect it would have is that he would end up in the wrong team for his skill.

But I already had tried this, and it simply did not work out. The problem was not that the idea is bad, but that it is hard to judge "how good" a player is based on the mere actions that happened during a game.

Dokujisan wrote:IIRC, this is because you were paranoid about someone being able to track your activity or something?

Anyway, what if it were opt-in?

Belittling replies aren't really necessary, are they? Simply because you don't see the utility of something doesn't make it meaningless or useless. The truth is you have no idea why I care about such things nor will you ever. Perhaps I have good reasons, perhaps I'm just batshit paranoid (which in turn might have a reasonable basis).

Btw, apart from the anonymity aspect, global stats just leads to more prick-waving as people get more focused on trying to be better than everyone else. I prefer the friendlier atmosphere, especially of the EU servers and although that won't last forever, I'd prefer to not precipitate its downfall, which some movements in the community are clearly doing already. (yeah, I'm being vague there to avoid the shitstorm and ignorant replies that that discussion will bring)

EDIT: fixed a typo

Instead of balancing the number of members in each team, why not balance the teams current ability.
(I'm assuming two teams)

ability_A == current number of players in team A * current score of team A
ability_B == current number of players in team B * current score of team B

if (ability_A > ability_B) then the new player joins team B

Obviously, you would have to include some checks to make sure that the teams didn't become very unbalanced.

When the larger team is being spammed with messages "One of you join the smaller team" display also spam the smaller team "You're team is under size, the other team has been informed"

Add the number players in each team to all the score boards.

There are servers where you can't choose which team you will join. The teams are always even there. Do this in each CTF/TDM server and problems are solved.

What happens if one of the teams is full of Nex gods and the other has the same number of new players?

Heh, then you have a problem

victim [v] wrote:What happens if one of the teams is full of Nex gods and the other has the same number of new players?

I believe we can solve this problem using math.


God team has 10 players. at the end of the game, it has 500 points

New team has 10 players, at the end of the game, it has 100 points.

the average of 500 and 100 is 300.

from the god team, remove the player with the highest score that when added to the score of the new team, would not exceed 300.

if the score of the new team would now make less than 300, repeat the step again with a lower scoring member of the god team, who when added to the new total, would still not make the new team's score exceed 300. do this until there are no players left with scores that would not in total end up causing the new teams score to exceed 300.

in the next round, switch these selected god team members into the new team.


this round should be a much more evenly matched.

You forgot to incorporate time. If pavlvs joins with 30 seconds left and gets 1 kill, he's going to get ranked as a bad player.

The essential ingredients for ranking CTF players would be:
kills, deaths, flag scores, flag drops, flag returns and time. Ideally, flag carrier kills and accuracy would be incorporated but that would need bigger code changes. But ignoring those factors we can do something like this:

(k-d)+2*(1+f*(r-e))/t

Where:
k = kills
d = deaths
f = flag scores
e = flag drops
r = flag returns
t = time

In english, You'll get a combination of your killing power (kills minus deaths) plus your value as a team player (returns minus drops multiplied by the number of flag scores times 2). Notice, if the player has more drops than returns, they will be punished by the same function that would reward them if they weren't such a selfish player. Option, 2 could be subbed in with a value set by the server admin, v. Then divide those two numbers by time.

to summarize:

(k+v(x))/t

where:

k = killing power
v = value as a team player
x = bonus/punishment multiplier
t = time


Examples:

((k-d)+2*(1*f*(r-d)))/t

25 kills, 5 deaths, 3 flag caps, 1flag drop, 3 flag returns, 6 minutes
(team player)

((25-5)+2*(1*3*(3-1)))/6
(20+2*(1*3*(2)))/6
(20+2*(3*(2)))/6
(20+2*6)/6
(20+12)/6
(32)/6 = 5.333

20 kills, 15 deaths, 0 flag caps, 3 flag drop, 1 flag returns, 6 minutes
(selfish / unaware)

((20-15)+2*(1*0*(1-3)))/6
(5+2(-2))/6
= 1/6 = .1667

deaths are not necessarily a bad thing wrt to team value. for example, a player prepared to sacrifice himself to take the brunt of the counter attack while hes team got the other ones flag are a valuable team member. flag return scores are not that important, anyone can (and should!) return the flag. carrier kills however needs to be factored in.

kills are not always good. take a map like hydronex. killing off to much will mean the enemy base is crawling with 150 hp players armed with, at least, a rather nasty shotgun. making the attackers job near impossible. and then of course if you spend time killing instead of trying to cap or defend, or fulfill whatever job the middleman needs to be doing. your not helping your team.

killing power has little to do with k/d ratio, this part is bogus imo. k/t are more like it.

weapon / item hogging are a BIG - in team play. someone grabbing a gun they allready have _while teammates are close_ or gobbeling up more then 200/100 hp/ar need to have their team-worth-score cut drastically regardless of other factors. im sure theres other things like this that needs to be thought of, this was just of the top of my head when i read this.

what really would bring some valuable info to this equation is if the player a) announces hes role and b) tried to fulfill it. hard to make happen tho.

in all i think its good to try to develop a generic score formula. but in the end its hard to evaluate a players worth by statistics. perhaps a karma system could help. eg let players say i dis/like this player. to be effective that would need reliable between game tracking tho.

[-z-] wrote:You forgot to incorporate time. If pavlvs joins with 30 seconds left and gets 1 kill, he's going to get ranked as a bad player.

The essential ingredients for ranking CTF players would be:
kills, deaths, flag scores, flag drops, flag returns and time. Ideally, flag carrier kills and accuracy would be incorporated but that would need bigger code changes. But ignoring those factors we can do something like this:

(k-d)+2*(1*f*(r-d))/t

Where:
k = kills
d = deaths
f = flag scores
d = flag drops
r = flag returns
t = time

In english, You'll get a combination of your killing power (kills minus deaths) plus your value as a team player (returns minus drops multiplied by the number of flag scores times 2). Notice, if the player has more drops than returns, they will be punished by the same function that would reward them if they weren't such a selfish player. Option, 2 could be subbed in with a value set by the server admin, v. Then divide those two numbers by time.

to summarize:

(k+v(x))/t

where:

k = killing power
v = value as a team player
x = bonus/punishment multiplier
t = time


Examples:

((k-d)+2*(1*f*(r-d)))/t

25 kills, 5 deaths, 3 flag caps, 1flag drop, 3 flag returns, 6 minutes
(team player)

((25-5)+2*(1*3*(3-1)))/6
(20+2*(1*3*(2)))/6
(20+2*(3*(2)))/6
(20+2*6)/6
(20+12)/6
(32)/6 = 5.333

20 kills, 15 deaths, 0 flag caps, 3 flag drop, 1 flag returns, 6 minutes
(selfish / unaware)

((20-15)+2*(1*0*(1-3)))/6
(5+2(-2))/6
= 1/6 = .1667

I didn't forget to incorporate time, I chose not to. there is no study to my knowledge on what amount of time is statistically significant in an fps game of capture the flag. so best leave it out, and let the unbalance exist for one more round than anything else. even then, the match should be fairly close, its no fun if its always perfectly balanced.

Xeno The Blind wrote:Belittling replies aren't really necessary, are they? Simply because you don't see the utility of something doesn't make it meaningless or useless. The truth is you have no idea why I care about such things nor will you ever. Perhaps I have good reasons, perhaps I'm just batshit paranoid (which in turn might have a reasonable basis).

I don't think it was belittling (especially not compared to some of the comments you make from time to time).

You seem concerned about a threat stemming from this.
I don't see any reason for concern, even in theory, from someone who participates in IRC and on forums. I don't see the difference. Thus, it appears like paranoia to me. You don't seem interested in explaining any further, so we're at an impasse.

My concern is that you could be holding back development of a much-wanted feature while giving very little explanation. I don't want to know your personal life, but it would be useful to know something more to support the concern. I don't see how it could be any more dangerous than using other internet services, like a web forum, IRC or email.

Btw, apart from the anonymity aspect, global stats just leads to more prick-waving as people get more focused on trying to be better than everyone else.

Competitive gameplay doesn't spawn "prick-waving" any more than forums spawn heated arguments. It is really dependent upon the community. Many people value competition for adding excitement to a game. Some might use stats as a method for tracking improvement.

tZork wrote:deaths are not necessarily a bad thing wrt to team value. for example, a player prepared to sacrifice himself to take the brunt of the counter attack while hes team got the other ones flag are a valuable team member.

lol wut? You're still dying, it doesn't matter if it's for the good of the team, that's a whole 'nother level of statistics that my equation is not concerned with. Regardless of the fact that Nexuiz can't predict whether you died to help your team or not. Dying is not a good thing, I don't believe you should get points even if you did die ~for your team~. You still died, and lost all your weapons and your location.

tZork wrote:flag return scores are not that important, anyone can (and should!) return the flag. carrier kills however needs to be factored in.

Well now, your ideology and public server realities are two distinct pictures. You'd like to assume everyone knows to retrieve the flag but spec on galts for a while and you'll see just how oblivious some players can be.

tZork wrote:kills are not always good. take a map like hydronex. killing off to much will mean the enemy base is crawling with 150 hp players armed with, at least, a rather nasty shotgun. making the attackers job near impossible. and then of course if you spend time killing instead of trying to cap or defend, or fulfill whatever job the middleman needs to be doing. your not helping your team.

In this example, the deaths are also high... so if we go back to my point about deaths being a bad thing, you'll see how this balances out.

tZork wrote:killing power has little to do with k/d ratio, this part is bogus imo. k/t are more like it.

it's all over time... kills minus deaths times a multipler OVER TIME.

tZork wrote:weapon / item hogging are a BIG - in team play. someone grabbing a gun they allready have _while teammates are close_ or gobbeling up more then 200/100 hp/ar need to have their team-worth-score cut drastically regardless of other factors. im sure theres other things like this that needs to be thought of, this was just of the top of my head when i read this.

Again some sort of ideology here. I'm speaking about public servers. Public servers.

tZork wrote:what really would bring some valuable info to this equation is if the player a) announces hes role and b) tried to fulfill it. hard to make happen tho.

While I'd love a team fortress mode, that's hardly a solution to balancing.

tZork wrote:in all i think its good to try to develop a generic score formula. but in the end its hard to evaluate a players worth by statistics.

Really because there are quite a few out there that do a pretty good job.

tZork wrote:perhaps a karma system could help. eg let players say i dis/like this player. to be effective that would need reliable between game tracking tho.

What could possibly go wrong?!?


I feel like you're trying to hold my formula up against unrealistic hopes and dreams. I'm tired of this "it won't work" attitude. Experiments are how we build knowledge about our science.

[-z-] wrote:

tZork wrote:deaths are not necessarily a bad thing wrt to team value. for example, a player prepared to sacrifice himself to take the brunt of the counter attack while hes team got the other ones flag are a valuable team member.

lol wut? You're still dying, it doesn't matter if it's for the good of the team, that's a whole 'nother level of statistics that my equation is not concerned with. Regardless of the fact that Nexuiz can't predict whether you died to help your team or not. Dying is not a good thing, I don't believe you should get points even if you did die ~for your team~. You still died, and lost all your weapons and your location.

But deaths already have their consequence, the other team gets a point. Why have double-punishment?

[-z-] wrote:

tZork wrote:flag return scores are not that important, anyone can (and should!) return the flag. carrier kills however needs to be factored in.

Well now, your ideology and public server realities are two distinct pictures. You'd like to assume everyone knows to retrieve the flag but spec on galts for a while and you'll see just how oblivious some players can be.

Yes, returning the flag is good, but then people will just race toward the flag. I feel the person who killed the FC should be rewarded, not the guy who returned it.

[-z-] wrote:

tZork wrote:kills are not always good. take a map like hydronex. killing off to much will mean the enemy base is crawling with 150 hp players armed with, at least, a rather nasty shotgun. making the attackers job near impossible. and then of course if you spend time killing instead of trying to cap or defend, or fulfill whatever job the middleman needs to be doing. your not helping your team.

In this example, the deaths are also high... so if we go back to my point about deaths being a bad thing, you'll see how this balances out.

I agree with tZork on this one.

[-z-] wrote:

tZork wrote:killing power has little to do with k/d ratio, this part is bogus imo. k/t are more like it.

it's all over time... kills minus deaths times a multipler OVER TIME.

I disagree with both of you here. Maybe on the scores box you could put how many deaths the player had, so they could realize "oh, I'm dying way too much, and everyone can see it. Better pull back on the carelessness"

[-z-] wrote:

tZork wrote:weapon / item hogging are a BIG - in team play. someone grabbing a gun they allready have _while teammates are close_ or gobbeling up more then 200/100 hp/ar need to have their team-worth-score cut drastically regardless of other factors. im sure theres other things like this that needs to be thought of, this was just of the top of my head when i read this.

Again some sort of ideology here. I'm speaking about public servers. Public servers.

Even on public servers. Maybe we could add a sprite to people who just have starting weapons so it's easier to tell? (Of course, this would only be shown to team-mates)

[-z-] wrote:

tZork wrote:what really would bring some valuable info to this equation is if the player a) announces hes role and b) tried to fulfill it. hard to make happen tho.

While I'd love a team fortress mode, that's hardly a solution to balancing.

meh, that sounds sort of like a completely different gamemode. Maybe some sort of fireteam system would help?

[-z-] wrote:

tZork wrote:in all i think its good to try to develop a generic score formula. but in the end its hard to evaluate a players worth by statistics.

Really because there are quite a few out there that do a pretty good job.[/url]

It's possible, but we have to stop the bickering and allow for constructive criticism.

[-z-] wrote:

tZork wrote:perhaps a karma system could help. eg let players say i dis/like this player. to be effective that would need reliable between game tracking tho.

What could possibly go wrong?!?


I feel like you're trying to hold my formula up against unrealistic hopes and dreams. I'm tired of this "it won't work" attitude. Experiments are how we build knowledge about our science.

User moderation works, but only if the user is willing to do it, and only if the user is reasonable. I don't think people are willing to fill out a questionnaire at the end of every map. On top of that, I'm afraid people may get the mentality of "if he's on my team I'm going to say he was good so we win, but if he's on the other team, he sucked."

I think you guys are missing the point that my calculations would be for weighting on public servers. Done behind the scenes. Most users wouldn't even know about. I believe however, it would create fairer matches than what currently exist.

Each user gets a score used internally at the end of the match and then the new teams created would used the data from the previous match to try and weight the teams fairly.


I like psychcf' idea about the 'new player' sprite but unless we match a user's ip against a database, how would we know? Let the user control it? That'll lead to abuse. Global login? Don't hold your breath.

If you think my formula is flawed, I challenge you to argue it with math, not just an opinion.

[-z-] wrote:I like psychcf' idea about the 'new player' sprite but unless we match a user's ip against a database, how would we know? Let the user control it? That'll lead to abuse. Global login? Don't hold your breath.

No no no... if the player has /just/ the starting weapons, then he'd have a sprite above him. That way his teammates know he needs a weapon.

Or, if the player could only use the starting weapons because ammo from another weapon has been depleted, then it'd show the sprite.

Ooooh even better idea. Sounds doable too!

Dokujisan wrote:

Xeno The Blind wrote:Belittling replies aren't really necessary, are they? Simply because you don't see the utility of something doesn't make it meaningless or useless. The truth is you have no idea why I care about such things nor will you ever. Perhaps I have good reasons, perhaps I'm just batshit paranoid (which in turn might have a reasonable basis).

I don't think it was belittling (especially not compared to some of the comments you make from time to time).

You seem concerned about a threat stemming from this.
I don't see any reason for concern, even in theory, from someone who participates in IRC and on forums. I don't see the difference. Thus, it appears like paranoia to me. You don't seem interested in explaining any further, so we're at an impasse.

My concern is that you could be holding back development of a much-wanted feature while giving very little explanation. I don't want to know your personal life, but it would be useful to know something more to support the concern. I don't see how it could be any more dangerous than using other internet services, like a web forum, IRC or email.

Btw, apart from the anonymity aspect, global stats just leads to more prick-waving as people get more focused on trying to be better than everyone else.

Competitive gameplay doesn't spawn "prick-waving" any more than forums spawn heated arguments. It is really dependent upon the community. Many people value competition for adding excitement to a game. Some might use stats as a method for tracking improvement.

wouldn't a system like this require that users need passwords?

I kinda hate to spoil all the hard work you put into this topic, but what good will it be if people can still choose what team to play on? What use is it to ASK people to switch if all they do is continue to bash the other team? People are just selfish idiots at times and trying to autobalance things will just make them choose the teams themselfs and continue screwing around..
Or maybe to put it gentler, how do you expect all your work to really help improve public matches? A lot of players just play one or two matches, thats hardly time enough to gather data about them and for having a sort of database, a lot of players play with fakenames for what ever reason. And i think those players tend to be good players. As soon as they choose a new name the system will have to collect data about them again for two games or so. And even if one would get a working system for this, what about the players that leave in mid-game, rebalance the players each times? Well maybe it would not be more annoying to be rebalanced then it is to point out teams are unbalanced and being ignored.
Basicly i'm still thinking that there are too many unknows to get such a system working really well and that people themself should do the work, but during the last weeks i slowing was convinced that a lot of people are just mean and will abuse any system. You should be sure that its worth to put work into this before you continue..

On some servers the only team balancing is simply a message informing the bigger team. The method the server uses seems to be very basic - a simple head count.

On other servers, some form of enforced auto balancing takes place. The server actually makes somebody change teams, again the method seems to be a basic head count. The server tries to balance the teams by moving a player from the oversized team.

Do we want our team based games to end with all the teams having similar scores or do we want the players to decide for themselves (free market vs state control)?

I personally want the server to tell both teams when they appear to be seriously out of balance. When I'm in the weaker team I want to see a message telling me that we are currently below par.

As esteel has said (and have others) to balance the teams at the start of the game the server needs to be able to positively identify the players. The server could identify the players from the previous game that stay connected with their IP. However, the server cannot identify a player who connects after the game has started.

If a players disconnects and subsequently reconnects, even with both the same IP and player name there is no guarantee that it is the same person.

Although if some kind of secret token based authentication was used (a line in the players config or even a cookie), then it would be possible for each server to identify a player. Each server would have its own stats for all the participating players. The server would then know the players token and the server admins could easily use it.

Maybe we need a third way. Reduce the bigger teams potential points, award them less points per flag capture, keys collected. Or give them less health, reduce their speed. Simply penalise the stronger team.

esteel wrote:Basicly i'm still thinking that there are too many unknows to get such a system working really well and that people themself should do the work, but during the last weeks i slowing was convinced that a lot of people are just mean and will abuse any system. You should be sure that its worth to put work into this before you continue..

This is why I chose to divide by time im my equation. It gives us an average. I'm not saying it's a perfect solution but it's better than what we currently have. In theory my code is a much simpler solution than a global user system. A little math, get a number to assert a compound 'value' on a player, group based on individual values.

Jikes seems i steeped on some sore toes providing some input here.

[-z-] wrote:

tZork wrote:deaths are not necessarily a bad thing wrt to team value. for example, a player prepared to sacrifice himself to take the brunt of the counter attack while hes team got the other ones flag are a valuable team member.

lol wut? You're still dying, it doesn't matter if it's for the good of the team, that's a whole 'nother level of statistics that my equation is not concerned with. Regardless of the fact that Nexuiz can't predict whether you died to help your team or not. Dying is not a good thing, I don't believe you should get points even if you did die ~for your team~. You still died, and lost all your weapons and your location.

Im not suggesting a reward. im suggesting its not much of a negative. And, like Psychcf pointed out death has its own bad built in.

[-z-] wrote:

tZork wrote:flag return scores are not that important, anyone can (and should!) return the flag. carrier kills however needs to be factored in.

Well now, your ideology and public server realities are two distinct pictures. You'd like to assume everyone knows to retrieve the flag but spec on galts for a while and you'll see just how oblivious some players can be.

Right, evaluating players worth has nothing do with what you call ideology *giggles*. what ever the current public scene is like would not a balancing system be for the purpose of making that scene better? i dont (pardon the language) give a flying fuck what ppl do on galts, {x} or in your bedroom. if someone plays like a turd he should be evaluated as one.

[-z-] wrote:

tZork wrote:kills are not always good. take a map like hydronex. killing off to much will mean the enemy base is crawling with 150 hp players armed with, at least, a rather nasty shotgun. making the attackers job near impossible. and then of course if you spend time killing instead of trying to cap or defend, or fulfill whatever job the middleman needs to be doing. your not helping your team.

In this example, the deaths are also high... so if we go back to my point about deaths being a bad thing, you'll see how this balances out.

tZork wrote:killing power has little to do with k/d ratio, this part is bogus imo. k/t are more like it.

it's all over time... kills minus deaths times a multipler OVER TIME.

You may agree or disagree with my point of view, but as stated before i see no distinct team negative in getting fragged. with this in mind the kill power factor is simply kills over time, after all the player still killed this many players over this much time. it don't matter if he dies 2 or 200 times doing it.

[-z-] wrote:

tZork wrote:weapon / item hogging are a BIG - in team play. someone grabbing a gun they allready have _while teammates are close_ or gobbeling up more then 200/100 hp/ar need to have their team-worth-score cut drastically regardless of other factors. im sure theres other things like this that needs to be thought of, this was just of the top of my head when i read this.

Again some sort of ideology here. I'm speaking about public servers. Public servers.

Okay! lets design a system that keeps everything the way it is, since we all love the current state so much.

[-z-] wrote:

tZork wrote:what really would bring some valuable info to this equation is if the player a) announces hes role and b) tried to fulfill it. hard to make happen tho.

While I'd love a team fortress mode, that's hardly a solution to balancing.

Right! (?)

[-z-] wrote:

tZork wrote:in all i think its good to try to develop a generic score formula. but in the end its hard to evaluate a players worth by statistics.

Really because there are quite a few out there that do a pretty good job.

Good for them.

[-z-] wrote:

tZork wrote:perhaps a karma system could help. eg let players say i dis/like this player. to be effective that would need reliable between game tracking tho.

What could possibly go wrong?!?

If the majority of ppl would abuse or ignore such a system then theres no point at all in trying to balance anything. im talking about a factor in the balancer, not a replacement.

[-z-] wrote:I feel like you're trying to hold my formula up against unrealistic hopes and dreams. I'm tired of this "it won't work" attitude. Experiments are how we build knowledge about our science.

Get some coffee and read my post again. i never once said anything remotely in the lines your suggesting. I added my ideas on how to make it better, before it gets implemented and cemented.

so, lets first have a look the the problem we need to solve. We have two teams, these need to be better balanced. To do this we need to define what a good teams is. A good teams in ctf needs players doing different thing, some defending, some attacking and perhaps a few middle men. knowing this its easy to see that a single "value" for each player provides to little information to form two balanced teams. we may end up with one team full of attackers and one full of defenders and the game would never know.

Heres a simplified version of what im currently working on for a team balancer:
defensive score = (fc kills + flag returns + kills close to flag) / (time played / time limit)
middleman score = (kills + (kills close to own fc * 5)) / (time played / time limit)
offensive score = ((captures / capture limit) - ((failed captures / cap limit) * 0.5)) / (time played / time limit)

Now, at the start of next game (or at a forced re balance) we could normalize these scores to the best one in each category. we then select the highest one for each player and zero the rest.
we can then select the two best attackers, flip a coin for who goes to what team. we do the same for defenders and middle. then do it over again with the remaining "free" players until the list is empty. Keep in mind that above score evaluation is simplified and likely flawed. what im thinking right now is a system that responds so events like kills, caps, deaths and so on and build a progressive score instead.

[-z-] wrote:

esteel wrote:Basicly i'm still thinking that there are too many unknows to get such a system working really well and that people themself should do the work, but during the last weeks i slowing was convinced that a lot of people are just mean and will abuse any system. You should be sure that its worth to put work into this before you continue..

This is why I chose to divide by time im my equation. It gives us an average. I'm not saying it's a perfect solution but it's better than what we currently have. In theory my code is a much simpler solution than a global user system. A little math, get a number to assert a compound 'value' on a player, group based on individual values.

I suppose your solution would be ideal in certain scenarios, such as when players only stick around for two games.

but consider that a good player joining for 50 seconds, might simply not have had the chance to score any points. it would be unbalanced.

perhaps we could have all players who haven't played past a certain threshold be assigned to groups randomly. again, not a perfect solution, but better than assigning a leet player you thought was nub to an already leet team.

victim wrote:On some servers the only team balancing is simply a message informing the bigger team. The method the server uses seems to be very basic - a simple head count.

On other servers, some form of enforced auto balancing takes place. The server actually makes somebody change teams, again the method seems to be a basic head count. The server tries to balance the teams by moving a player from the oversized team.

Do we want our team based games to end with all the teams having similar scores or do we want the players to decide for themselves (free market vs state control)?

I personally want the server to tell both teams when they appear to be seriously out of balance. When I'm in the weaker team I want to see a message telling me that we are currently below par.

As esteel has said (and have others) to balance the teams at the start of the game the server needs to be able to positively identify the players. The server could identify the players from the previous game that stay connected with their IP. However, the server cannot identify a player who connects after the game has started.

If a players disconnects and subsequently reconnects, even with both the same IP and player name there is no guarantee that it is the same person.

Although if some kind of secret token based authentication was used (a line in the players config or even a cookie), then it would be possible for each server to identify a player. Each server would have its own stats for all the participating players. The server would then know the players token and the server admins could easily use it.

Maybe we need a third way. Reduce the bigger teams potential points, award them less points per flag capture, keys collected. Or give them less health, reduce their speed. Simply penalise the stronger team.

no, penalizing strong players is a TERRIBLE way to keep balance. it makes the game not fun. the config file idea would be good, except for the fact that a locally stored file, is one that can be locally edited.

tZork wrote:Heres a simplified version of what im currently working on for a team balancer:
defensive score = (fc kills + flag returns + kills close to flag) / (time played / time limit)
middleman score = (kills + (kills close to own fc * 5)) / (time played / time limit)
offensive score = ((captures / capture limit) - ((failed captures / cap limit) * 0.5)) / (time played / time limit)

This is like an over simplified version of what I did. You just made it require new information about events and dividing it up amongst these unknown (to the game) areas of play. Mine only requires math.

The only thing I agree with is your denominator.

I like where you're going with it but I think it's unrealistic unless you're coding all that extra information. Though if that's the case, I'd argue you use your efforts towards a stronger base.

[-z-] wrote:

tZork wrote:Heres a simplified version of what im currently working on for a team balancer:
defensive score = (fc kills + flag returns + kills close to flag) / (time played / time limit)
middleman score = (kills + (kills close to own fc * 5)) / (time played / time limit)
offensive score = ((captures / capture limit) - ((failed captures / cap limit) * 0.5)) / (time played / time limit)

This is like an over simplified version of what I did.

What? How is these two even similar? Like i said i posted a simplified version here so that it would be easy to understand. The point of my system is not the precise formula use to derive each score, its that its split into evaluating the three primary roles of a ctf team. Your system does not pay any attention to this.

[-z-] wrote:You just made it require new information about events and dividing it up amongst these unknown (to the game) areas of play. Mine only requires math.

The only thing I agree with is your denominator.

I like where you're going with it but I think it's unrealistic unless you're coding all that extra information. Though if that's the case, I'd argue you use your efforts towards a stronger base.

coding this is a pice of cake. im still working out some kinks in my actual formula once thats done ill code it and let ppl try.

Simply put, I'm not going to waste my breath.

I have stated my formula, given examples and reasons. The way you and I see CTF is just different and you are going to code it however you want it. So until I learn to code it myself or someone's willing to back me up, I guess we'll just end up playing it your way.