Author Topic: Community - Player ratings  (Read 44517 times)

BabyAlien

  • Posts: 58
  • Turrets: +0/-0
Community - Player ratings
« on: May 23, 2006, 03:54:36 am »
One thing that might be very valuable and add to the gameplay of Tremulous would be to somehow track and indicate the skill level of players.  One major factor in the outcome of a match is the balance of skill across teams.  If you have a L33T player on the alien team, for example, they can rack up 6-7 kills in the first couple minutes of the game and rocket the aliens on to S2.

I actually experienced a match like this where a really good alien player, against a group of normal humans racked up 15 or 16 kills in the first two minutes killing off humans in their base.

As another example of this, I'm sure everyone experiences joining a game and seeing something like this...

DoodJ0bs     34   12   95
GumboJam   24   12   104
JimDiddly     8   13   86
TomTom      6   13   120
Bob             0   13   110
Billy            0    13   95
Bobby         0    13   96

Then the other team has kills like this: 9, 8, 6, 7, 5, 0, 0

So to cut to the chase, if there was some way of indicating the skill level of players so that when they join and when you size up the balance of the teams you can tell immediately that there are 2 killing machines, 3 moderate players, and 2 noobs on "Humans" and 0 killing machines, 4 moderate players and 3 noobs on "Aliens".

One reality of this game is that you have to get kills to go up stages, and the more players you have the more kills you need to get to the next stage.  So a team with 10 alien moderate players vs a team of 2 L33t and 2 N00b, my money is on the latter team
ah, that wasn't a tyrant you killed, it was just a baby tyrant.

Teiman

  • Posts: 286
  • Turrets: +0/-0
Re: Community - Player ratings
« Reply #1 on: May 23, 2006, 08:25:25 pm »
This is a strategic game, so a builder with zero frags is still very valuable to the team. You already have the frag count + weapon/suit/alientype that is mostly the leet level. Morons dont play as Tyrant forever!.

Maybe frags can be altered, so indicate this on a more accurate way...

you kill a tyrant = 5 frags
your turret kill a drench = 1 frag
you are killed = -2 frags

werepants

  • Posts: 69
  • Turrets: +0/-0
    • http://www.nine18design.com
Community - Player ratings
« Reply #2 on: May 23, 2006, 10:11:17 pm »
it would be nice to see some accountability for deaths, whether it is a give and take thing like Teiman said or just a kill list.  it is much more impressive in my eyes if someone gets 20 kills with 0 deaths than if someone gets 20 kills and 20 deaths.  Also shows feeders for what they are.

[db@]Megabite

  • Posts: 613
  • Turrets: +3/-0
    • http://www.tremulous.info
Community - Player ratings
« Reply #3 on: May 23, 2006, 11:29:59 pm »
Quote from: "werepants"
Also shows feeders for what they are.
Mh, you could just get it wrong, either... ;)
An effective tactic I know (for good dretch players at least) is rushing the human base, doing a kill or damaging a turret and finally get killed by a turret.
You are able to stack up about 10 kills with 20-25 deaths in minutes on niveus if human defense is weak. Just strafejump in backway and aim for medipad at headlevel... and you are feeding less kills then you get. Just make sure you know the point when they finally got it. ;)

It does not tell you much about a player in such a case (and many other cases). It may also keep people from playing aliens because they see how often they actually died.

The whole highscore issue is actually not worth the discussion as stacking kills is not the goal of the game, just a way that may lead to victory. As Cata said in that highscore threat he nearly got 200 frags in one match but his team still lost it. By then dozens of frags are worth nothing, just testament for camping one spot and spamming a chokepoint way too long instead of actually hunting and trying to win.

Just my 2 cents,

Danny
url=http://www.tremulous.info][/url]


Neo

  • Posts: 760
  • Turrets: +2/-0
Community - Player ratings
« Reply #4 on: May 24, 2006, 12:39:08 am »
But High frags are a display of your ability to keep the enemy down, if your team can't keep up you'll still lose.

chompers

  • Posts: 224
  • Turrets: +4/-0
Community - Player ratings
« Reply #5 on: May 24, 2006, 03:57:00 am »
I'd like to see deaths as well as kills on the score screen.

Only deaths to other players should count though - dying to turrets or gravity or whatever else isn't important.

[HUN]N.M.I.

  • Posts: 87
  • Turrets: +1/-1
Community - Player ratings
« Reply #6 on: May 24, 2006, 10:51:49 am »
Funny that these kind of games are of a team-based concept, but everyone talks about personal frags :D.
It's ok for me, but a little weird due to the fact that the team AS A WHOLE can win or lose. Also, it happens several times that you're full of money because you shoot and fight well, but don't have any frags because teammates 'steal' (mean word, sry for that) your kills  :D .
url=http://userbars.org][/url]

stahlsau

  • Posts: 160
  • Turrets: +1/-1
Community - Player ratings
« Reply #7 on: May 24, 2006, 11:13:36 am »
I've seen it very often that the human team got more than twice as much kills than aliens, but in the end, they lost. Personal frags are really unimportant.

Stof

  • Posts: 1343
  • Turrets: +1/-1
Community - Player ratings
« Reply #8 on: May 24, 2006, 11:55:16 am »
Personal frags are useful as a rough guess for when the oposing team will start changing into dragoons and to see if we are getting closer to the next stage than them.
urphy's rules of combat
8 ) Teamwork is essential; it gives the enemy someone else to shoot at.
18 ) Make it too tough for the enemy to get in and you can't get out.

[HUN]N.M.I.

  • Posts: 87
  • Turrets: +1/-1
Community - Player ratings
« Reply #9 on: May 24, 2006, 01:19:49 pm »
Quote from: "Stof"
Personal frags are useful as a rough guess for when the oposing team will start changing into dragoons and to see if we are getting closer to the next stage than them.


Teamfragcount has just the same benefits :).
url=http://userbars.org][/url]

BabyAlien

  • Posts: 58
  • Turrets: +0/-0
Community - Player ratings
« Reply #10 on: May 24, 2006, 10:20:46 pm »
I was actually thinking of some kind of way of rating players so that before a match begins it is easier to balance.  For example, it is common to start a new match and if you know the names and skill of players you will know which team is going to win because two 'excellent' players are on one team and the other has a few 'good' players and lots of 'new' players.

So it would have to be something that would carry across games and go with you to different servers.

That's also why I used the word community in the topic name, because it would really have to be something adopted by the whole community that could be recognizable.  Perhaps it could be self-adopted honor system type of thing.  For example:

Skill level  - Prefix / Postfix  - example
-----------    --------------        ------------
Master      - Msta / 99 / X1     BabyAlien (msta), X1 - BabyAlien
Elite          - Lt / Elt / X2        BabyAlien.Lt, EliteBabyAlien
Good        - G / Playa / X3     BabyAlien-G
Lightweight - lite / X4            BabyAlien-lite
Noob        - Nu / 00 / X5        NuBabyAlien, BabyAlien00  

This is a poor example, but the community could put out some 'levels' and some agreed upon tags in a name that would indicate that level somehow.  

One potential negative would be contention of levels.  I could imagine a lot of flak on games where someone claimed to be 'msta' but had a lousy game and the whole team demands he change his tag to 'noob' or something.  Or another team complaining about a master in noob clothing...

The potential benefit could outweigh any negatives, helping to balance out games so that masters get spread out and strategy can be adapted based on the skill of the team.  It would be stupid for a team of 'good' players to rush out and try to mount up the kills when they don't have the skill for that....
ah, that wasn't a tyrant you killed, it was just a baby tyrant.

next_ghost

  • Posts: 892
  • Turrets: +3/-6
Community - Player ratings
« Reply #11 on: May 25, 2006, 11:37:56 am »
Well, there could be some sort of skill counter. A few Czech Counter-strike servers had something similar back in 2002/2003. Balancing hit-based skill counter won't be any problem (just a bunch of equations on 2 skill numbers with a few equipment/evo modifiers). The only problem I can see is rating builders... By structure lifespan and usage/turrets kills perhaps?

The feeding counter is also a good thing. Not just deaths but actual stage-counted deaths.
If my answer to your problem doesn't seem helpful, it means I won't help you until you show some effort to fix your problem yourself!
1.2.0 release's been delayed for 5:48:00 already because of stupid questions.

Desp.

  • Posts: 16
  • Turrets: +0/-0
Community - Player ratings
« Reply #12 on: May 26, 2006, 05:54:55 am »
Quote from: "BabyAlien"
For example, it is common to start a new match and if you know the names and skill of players you will know which team is going to win ....

That's just not true. 2 excellent players can apply a lot of pressure on the opposing team, but if their own team doesn't fill the vacuum they leave behind their 'excellency' will not have much of an effect on the outcome of the game.
Personally I've come to fear the dreaded "Hey, let's move to elevator room" (tactical suicide, imho) or "moving reactor" right when you have the opposing team almost down on their knees - completely destroying momentum and usually turning the tide of battle :roll:

Edit: typos

BabyAlien

  • Posts: 58
  • Turrets: +0/-0
Community - Player ratings
« Reply #13 on: May 26, 2006, 08:59:10 am »
Quote from: "Desp."
Quote from: "BabyAlien"
For example, it is common to start a new match and if you know the names and skill of players you will know which team is going to win ....

That's just not true. 2 excellent players can apply a lot of pressure on the opposing team, but if their own team doesn't fill the vacuum they leave behind they 'excellency' will not have much of an effect on the outcome of the game.
Personally I've come to fear the dreaded "Hey, let's move to elevator room" (tactical suicide, imho) or "moving reactor" right when you have the opposing team almost down on their knees - completely destroying momentum and usually turning the tide of battle :roll:


Yes, this is true.  But if you don't have at least the semblance of balance to start with then you will not really have a game.  I have played enough matches where one team advances quickly to s3 and tactics never really come into play at all.

So the idea would be to get a semblance of equality to start with.

The other factor would be that overall the ratings shouldn't be based on kills but rather overall goodness, i.e. teamwork and strategy as well.  On the other hand, community ratings could be more specific..

For example, slayer and builder ratings might lead to something like:
BabyAlien - S2B4
 .. Slayer level 2, builder level 4 (out of 5)
ah, that wasn't a tyrant you killed, it was just a baby tyrant.

next_ghost

  • Posts: 892
  • Turrets: +3/-6
Community - Player ratings
« Reply #14 on: May 26, 2006, 02:32:58 pm »
I think of a system like this:
- One floting-point number representing overall skill (maybe with building skill counted as a part of overall skill number)
- Every hit inflicted to building or player should be evaluated for skill value and put into queue of skill claims with inflicted damage, when the building or player dies, the queue will be passed through and claims will be assigned to attackers up to the maximum health of dead player/building (any skill claims beyond maximum health will be cancelled and partial skill claims - like 10hp claim with 5hp remaining to max health - will be adjusted to fit max health). Simply, if you let the enemy heal/repair, you don't get your skill.
- When an alien egg dies, it's skill claim list will be compacted (skill claim list will be truncated to maximum health of egg and all skill claims of each player will be added together) and copied to a special creep claim list of all alien buildings within creep range. When some building loses creep and starts losing health, lost health will be added as skill claims with NULL attacker. These claims will be divided among all players in special list of creep claims. The same will be for human buildings that get powered down (with repeaters and reactor instead of eggs), but aliens get skill for the period of time those buildings stay powered down.
- Builder gets skill points for repairing buildings (humans only, the amount should be equivalent to skill claims cancelled but attackers don't lose skill), every use of building (every player that spawns in a telenode or egg, every item bought at armory, every evo point spent while OM exists, every assigned skill claim by turret etc. and some extra points for every building staying up and running. Builder gets points only for buildings he personally built even after he gets a weapon and goes fighting.
- Builders lose skill for time without reactor/overmind (there should be enough time to rebuild reactor/OM without any penalty - no penalty for well planned move). If the reactor/OM is destroyed by enemy, equivalent of skill lost by builder is divided among attackers.
- When a building or player dies, skill claims will be added to attackers' skill number and substracted from skill of the dead player/builder of destroyed building.
- Winners should get some extra skill, losers should lose a little, no skill gain or loss for stalemate.
- General modifiers of skill claims: team balance (team player count ratio), skill balance (victim/attacker skill ratio, only for PvP hit, not against buildings), attacker's equipment, victim's equipment (inverted), number of turrets in active range (even blocked ones).

- Alien evo modifiers:
Granger - 2
Advanced Granger - 1.5
Dretch - 1
Basilisk - 0.9
Advanced Basilisk - 0.8
Marauder - 0.7
Dragoon - 0.6
Advanced Marauder - 0.5
Advanced Dragoon - 0.4
Tyrant - 0.3

Human modifiers:
Builder/unarmed (only blaster) - 2
Soldier with blaster - 1.5
Rifle, no armor - 1
???
Lucifer cannon, battlesuit - 0.3

This means if you slaughter grangers with lucifer in battlesuit, you get almost nothing but if you kill tyrant with blaster on your own as unarmored builder, you get lots of skill points.
If my answer to your problem doesn't seem helpful, it means I won't help you until you show some effort to fix your problem yourself!
1.2.0 release's been delayed for 5:48:00 already because of stupid questions.

BabyAlien

  • Posts: 58
  • Turrets: +0/-0
Community - Player ratings
« Reply #15 on: May 26, 2006, 11:54:28 pm »
Calculated numbers are an interesting option, but in the meantime there could be an opt-in system where people are encouraged to display their own ratings based on their skill.  The problem would be from new players who are starting to think they are good but haven't yet realised that they really aren't.  :)
ah, that wasn't a tyrant you killed, it was just a baby tyrant.

Teiman

  • Posts: 286
  • Turrets: +0/-0
Community - Player ratings
« Reply #16 on: May 27, 2006, 11:16:55 am »
Quote from: "BabyAlien"
Calculated numbers are an interesting option, but in the meantime there could be an opt-in system where people are encouraged to display their own ratings based on their skill.  The problem would be from new players who are starting to think they are good but haven't yet realised that they really aren't.  :)


A fix for that is to create "patrol groups", where you recieve orders from your leader in a special way. And your teammates are marked on the radar, too. Maybe even add 3 tiny "cameras" to see what your teammates see. This way you leader will be still be your leader, even If you score better and more than him. And you still follow orders. :D

SLAVE|Mietz

  • Posts: 672
  • Turrets: +2/-0
    • http://blasted.tremulous.info
Community - Player ratings
« Reply #17 on: May 27, 2006, 11:26:37 am »
For me, there is no well working way to measure "skill" in a game like Tremulous. Because you can be a "specialist", so you build, support your team, move base and this can't be measured in any AI-way.

The best way to measure skill atm is to play lots of games and get to know the community.

next_ghost

  • Posts: 892
  • Turrets: +3/-6
Community - Player ratings
« Reply #18 on: May 27, 2006, 03:16:08 pm »
Quote from: "SLAVE|Mietz"
For me, there is no well working way to measure "skill" in a game like Tremulous. Because you can be a "specialist", so you build, support your team, move base and this can't be measured in any AI-way.


It can't be measured directly but there're enough indirect ways to measure it accurately. Badly designed bases don't survive long, well designed bases do. Badly designed defences don't kill many enemies, well designed defences do. The other team would have to suck really hard to change this. And if it does, bad builder won't get much skill points because the game will be over soon.
If my answer to your problem doesn't seem helpful, it means I won't help you until you show some effort to fix your problem yourself!
1.2.0 release's been delayed for 5:48:00 already because of stupid questions.

Paradox

  • Posts: 2612
  • Turrets: +253/-250
    • Paradox Designs
Community - Player ratings
« Reply #19 on: May 28, 2006, 02:35:04 am »
i know this is very dificult, but can the master server keep track of users, and give them a "rank" like halo 2 has.

∧OMG ENTROPY∧

BabyAlien

  • Posts: 58
  • Turrets: +0/-0
Community - Player ratings
« Reply #20 on: May 28, 2006, 03:21:56 am »
Quote from: "Paradox"
i know this is very dificult, but can the master server keep track of users, and give them a "rank" like halo 2 has.


It might also be great to have some kind of approval system, where people can give positive or negative points to someone.  Basically, if you are new but try hard, listen and learn, you will pretty much always be an asset to a team.  However, if you are inclined to making life difficult for others, don't listen and are inclined to abusive language and behaviour, it doesn't matter how good you are, you aren't a desirable team member.

One thing about a central server is that you would never have duplicate names, which could be annoying on one hand.. but good on another.
ah, that wasn't a tyrant you killed, it was just a baby tyrant.

benplaut

  • Posts: 195
  • Turrets: +0/-0
Community - Player ratings
« Reply #21 on: May 28, 2006, 05:16:47 am »
i like BabyAlien's second idea... SxBx...

i don't think an official AI ranking is needed... if a S1B1 says they're S5B5, it'll become obvious pretty soon, and someone will probably ask them to change it.

S1.5B2... to be generous to myself

 8)
img]http://img240.imageshack.us/img240/5443/5863101266io.gif[/img]
}MG{benplaut

next_ghost

  • Posts: 892
  • Turrets: +3/-6
Community - Player ratings
« Reply #22 on: May 28, 2006, 10:08:01 am »
Quote from: "Paradox"
i know this is very dificult, but can the master server keep track of users, and give them a "rank" like halo 2 has.


It wouldn't work very well without some central ranking point. If you'd start with no skill rank each time you connect to a server, this system would be useless.

Quote from: "BabyAlien"
One thing about a central server is that you would never have duplicate names, which could be annoying on one hand.. but good on another.


You can as well keep skill rank against player UID, not only name. Wolf:ET for example uses player UID-based temporary bans.
If my answer to your problem doesn't seem helpful, it means I won't help you until you show some effort to fix your problem yourself!
1.2.0 release's been delayed for 5:48:00 already because of stupid questions.

Paradox

  • Posts: 2612
  • Turrets: +253/-250
    • Paradox Designs
Community - Player ratings
« Reply #23 on: May 30, 2006, 01:12:01 am »
No, what I mean is the central server tracks rank, from server to server, day to day.

∧OMG ENTROPY∧

Norfenstein

  • Posts: 628
  • Turrets: +81/-78
Community - Player ratings
« Reply #24 on: May 31, 2006, 05:10:02 pm »
I'm not really sure what you want to do with player rankings, but I see only two options:

1) Force balanced teams based on rankings. People will really not like being forced to play a certain team, especially the ones who only play humans. This just wouldn't work because there'd be no way to prevent intentional rank degradation and if the rankings aren't accurate the system will just be forcing people to play a certain way for no reason
2) "Elective" team balancing, wherein people voluntarily balance teams. The assumption is that all people want balanced games all the time which just isn't the case. Most people want to win, because winning is more fun than losing, and trying to shoot for a close victory is a lot harder than just stacking teams. It takes skilled, responsible players just to pick the side that a map doesn't favor.

I'm afraid there's no getting around this for public servers. An automatic rating system would not be able to judge the skill level of anyone that doesn't play their best every single time they join a server and you certainly can't force people to play well to avoid intentionally lowering their rank.
Any such system will only benefit people who don't know who they're playing with anyway, unless you're forcing teams. Nothing beats just playing with people and judging them for yourself; regular players already know who can carry a team. Adding a few more raw statistics to the scoreboard would improve this, but it'd be a mistake to have a computer try to do something we can accomplish incomparably better.

Ksempac

  • Posts: 261
  • Turrets: +1/-1
    • http://www.ksempac.info/blog
Community - Player ratings
« Reply #25 on: May 31, 2006, 07:33:38 pm »
I dont believe in any ranking system in Tremulous :

First it is impossible to make a fair ranking :
If you take into account the number of frags or the number of deaths, campers behind turrets will get a high rank despite the fact that they are poor players. A good player dont feed unnecessarily, but knows when it is time to take risks.
If you try to take into account the base building you re in trouble because it is all about teamwork :
- A good base without defensers goes down
- Good players can protect a loosy base (or win the match before the base starts to be really useful)
- A good player can have trouble building the base he wants if a noob that doesnt want to listen is building at the same time.
- A griefer ruin any base in 1 min. So much for the ranks of the real builders.
Moreover being able to play as a teammate is not something you can rank...I prefer a poor killer that listen to his teammates and do his best to follow their advice than a skilled player who think he is the only one to know how to play and do as he wants.

Second how you make the ranking and show it to other players ?
The SxBx system is based on the good will and honesty of other players.
Anyone could put in his nick a S5B5, if the guy isn t too bad it might get unnoticed that he is not as good as he said. To enforce a ranking system you would have to be sure that each player has his nick (it s easy to take the nick of someone else) and that the rank is given by the server rather than the player.

Quote
2) "Elective" team balancing, wherein people voluntarily balance teams. The assumption is that all people want balanced games all the time which just isn't the case. Most people want to win, because winning is more fun than losing, and trying to shoot for a close victory is a lot harder than just stacking teams. It takes skilled, responsible players just to pick the side that a map doesn't favor.


That's actually happening : When teams were really unbalanced i often saw the best players switch team. It s not a given, but i often noticed it.
url=http://tremulous.net][/url]

BabyAlien

  • Posts: 58
  • Turrets: +0/-0
Community - Player ratings
« Reply #26 on: May 31, 2006, 09:21:12 pm »
The last couple of posts state the problem quite nicely.

Any solution would have to be adopted by the community and bought into by players.  

It is up to the individual players to balance out the teams and often players just don't care.

Rankings would only be an aid, or guide to assist in the balancing of teams. (if desired)

Of course, one thing that has been largely ignored in my SxBx ranking is tacticts & strategy.  I have seen players who are SuperSlayers (S5) but don't have great tactics, to the point of watching matches where one team has 5 times the kills of the other and is S3 vs S1, but still doesn't finish it because they are too focussed on getting kills.

Another thing that has been ignored is alien vs. human.  You can be a human killing machine but a lousy alien.

It is true that as you play on a server you get to know players by name and what they are capable of.  Perhaps the real benefit to player rankings would just be to raise awareness of how skill levels affect the game and to help new people to get their bearings.
ah, that wasn't a tyrant you killed, it was just a baby tyrant.

Neo

  • Posts: 760
  • Turrets: +2/-0
Community - Player ratings
« Reply #27 on: June 01, 2006, 01:45:05 am »
Might be something like Player Ratings, where you have a compiled list of players, preferably by some unique ID, and their most used name. Then have a section for that player so people can leave rep comments and rate that person as a player. Granted it wouldn't have much use apart from as an out of game reference point.

Paradox

  • Posts: 2612
  • Turrets: +253/-250
    • Paradox Designs
Community - Player ratings
« Reply #28 on: June 01, 2006, 04:19:38 am »
How about ranks dont mean anything, but a number by your name.

∧OMG ENTROPY∧

next_ghost

  • Posts: 892
  • Turrets: +3/-6
Community - Player ratings
« Reply #29 on: June 01, 2006, 09:43:26 am »
Trust me guys, I've seen a similar ranking system work and it worked very well.

Quote from: "Ksempac"
First it is impossible to make a fair ranking :
If you take into account the number of frags or the number of deaths, campers behind turrets will get a high rank despite the fact that they are poor players.


As I have already said here, the skill system would have to be based on actual damage inflicted rather than kills and deaths. If some turret hugging hippie gets lots of kills while turrets do most of his work, he gets almost nothing compared to what he loses when aliens come to wipe him out along with his turrets.

Quote
A good player dont feed unnecessarily, but knows when it is time to take risks.


Measured indirectly.

Quote
- A good base without defensers goes down


In that case, skill balance will make sure that the good builder does not lose too much skill because of bad low-skill teammates.

Quote
- Good players can protect a loosy base (or win the match before the base starts to be really useful)


Again, skill balance kicks in and the builder gets very little from turret hits.

Quote
- A good player can have trouble building the base he wants if a noob that doesnt want to listen is building at the same time.


Once the base goes down, it won't be the good builder who loses lots of skill. It will be the noob who reconstructs everything.

Quote
- A griefer ruin any base in 1 min. So much for the ranks of the real builders.


If some loser tries to deconstruct the base, he'll lose LOTS of skill by doing that.

Quote
Moreover being able to play as a teammate is not something you can rank...I prefer a poor killer that listen to his teammates and do his best to follow their advice than a skilled player who think he is the only one to know how to play and do as he wants.


Such player is often on the winning side and has his own share of credit measured indirectly by total damage he inflicts during the game. You do not have to kill someone to get skill. You just have to hit him and make sure someone else finishes him off before he can heal.

Quote
To enforce a ranking system you would have to be sure that each player has his nick (it s easy to take the nick of someone else) and that the rank is given by the server rather than the player.


Once again, YOU DO NOT HAVE TO IDENTIFY PLAYERS ONLY BY NAMES! Server-generated unique client IDs work much better. And if you want to reset your skill counter, you just delete your old ID and get a new one.
If my answer to your problem doesn't seem helpful, it means I won't help you until you show some effort to fix your problem yourself!
1.2.0 release's been delayed for 5:48:00 already because of stupid questions.