What do you think about a league in open clonk? Some of you will say now we don´t have enough szenarios and objects. But in the league of Clonk Rage the players always play the same 3-4 szens. And OC have some interesting meele and race. It would be a start and promote the network activity. Also some players would start to create more szenarios.
Well, it would probably be possible to allow OC to use the CR league system with only minor adjustments. With the problem being that even minor adjustments would have to be done by yours truly and in PHP (*shudder*). From a practical standpoint, the upside and downside of that would be that it would require CR registrations of players.
For a new OC league, there's the two problems that a lot of stuff needs to be developed first, as well as that we don't have any meaningful way to establish identities, so I fear the problem of double-accounts might become much worse.
For a new OC league, there's the two problems that a lot of stuff needs to be developed first, as well as that we don't have any meaningful way to establish identities, so I fear the problem of double-accounts might become much worse.
I know some free games like supremacy. They allow only one account per ip-adress. I don´t know if you can do this too. But there are also a second possibility. In some free games like travian there are some players who find out double-accounts. I´m sure you find some player how would make that. For example me :D
For a settlement league with a well structured score system multiple accounts are not a problem, and we are working towards a settlement release.
So I think it can be fruitful to start working on that (shaders first though!).
So I think it can be fruitful to start working on that (shaders first though!).
Yes by a settlement league double accounts are not really a problem. But peterw is right. A meele liga for a free game like oc has many problems. In Clonk Rage there was also the problem with multiple accounts. And today? Nobody play with a double account. We see you can stop them.
Well, to be fair OC would not be the only free game with a league. There have been ways to counter that problem in the past.
I'm not sure how a league would work in OC, since I have not played any previous clonks. I have a small problem with leagues, on other games that I have played the leagues cause people to quit if they are losing because they are just wasting time. There is one game whose league does not count losses, the result is that even if one team is losing there is no incentive to quit, but that game also has a very small player base.
Usually the league revolves around some sort of score. That score /should/ reflect the player skill at any moment in the league - so if you get better you would quickly rise in your score. Independent of how many losses you had until that point. (those systems are usually similar ELO or Microsoft-TrueSkill).
And if you do it right there should be nearly no incentive to start a new account just because you lose a lot :)
The ClonkRage league is a good example for that: You start with 0 and can only gain points. Even if you lose a lot you can never be lower than your starting score.
The problem there is that the top players make new accounts on which they can casually fool around without any fear of losing points
And if you do it right there should be nearly no incentive to start a new account just because you lose a lot :)
The ClonkRage league is a good example for that: You start with 0 and can only gain points. Even if you lose a lot you can never be lower than your starting score.
The problem there is that the top players make new accounts on which they can casually fool around without any fear of losing points
What's wrong about that? When your loss is inevitable, prolonging the game is wasting the winner's time as well. Obviously the ranking effects should be the same whether you quit early or not.
If I play a game in league, and enjoy it
is that time really wasted?
is that time really wasted?
You enjoy grinding your already hopeless opponent to death? I'm not sure that's the kind of fun we should advocate.
let's say we play CR waggon fight and it's 13 to 1. And both just respawned. The game at this point is played exactly like when it started only that one player is probably going to lose anyway
Why is it less fun? :x
Why is it less fun? :x
I kinda do enjoy having my turrets set up around the money ball and own the enemy pros when coming from the spawn in SMNC, but that's not my point.
Having 5 HP and no relaunches left doesn't mean you lost. http://www.youtube.com/watch?feature=player_embedded&v=vxGEx_DS4LY#! If the map is well done, it's not over until it's over.
Having 5 HP and no relaunches left doesn't mean you lost. http://www.youtube.com/watch?feature=player_embedded&v=vxGEx_DS4LY#! If the map is well done, it's not over until it's over.
Are you trying to tell me that spawnkilling is fun? Really? ;)
And if it isn't really over, I fully agree that you should continue playing. I'm just saying that if you really have lost, there's no point to continue playing for either side. Better start a new game so there's something at stake again.
And if it isn't really over, I fully agree that you should continue playing. I'm just saying that if you really have lost, there's no point to continue playing for either side. Better start a new game so there's something at stake again.
In SMNC you really have to be an idiot to let yourself get spawn killed. Unlike in Dota, you can't damage someone when he is in the spawn, also, it heals you up in a few seconds, so it kind of is an interesting challenge to spawnkill someone there.
What I actually should be telling you: There should not be a situation where the game is practically over and there won't be a Game Over popping up the next second.
What I actually should be telling you: There should not be a situation where the game is practically over and there won't be a Game Over popping up the next second.
Well, I'm in full agreement there. That's essentially what I said yesterday in #openclonk-dev: There shouldn't be hopeless situations, just situations where you have to take higher risks. Just think of Poker - the smaller your stack, the more you are forced into aggressive play.
> There should not be a situation where the game is practically over and there won't be a Game Over popping up the next second.
Well, Starcraft manages to be popular despite virtually every game ending in such a situation. So when in doubt, design the rules to be fun instead of satisfying the "the computer has to be as good as the players at predicting the outcome" constraint.
As does Chess. And it's not about predicting the outcome, it's about the outcome being as unpredictable as possible in every state of the game. It's unfortunate if that's (close to) impossible given the concrete game mechanics, but that doesn't mean it's not a good idea.
Chess is actually a good example of a game that could probably be modified to end by the rules at about the same time a human player would concede defeat in the standard game. You only have to write a program that can accurately predict the outcome by the standard rules when observing a game between two players. As soon as the program can predict the game with enough confidence, the new rules would declare a winner. But that's not a set of rules anyone would use for a tournament. :-)
Fundamentally, a competitive game tries to measure the skill of the players using rules that are vastly simpler than the players. It's only natural that a human can predict the outcome. We can try to fool them, but that requires to make the game less dependent on the skills of the players. There's only so far one can go down that route before one gets a game that's mostly dependent on luck.
So I don't think this is a solvable problem until artificial intelligence is widespread enough to make the above chess solution acceptable. In the meantime, the best we can do is to make the transition from an unpredictable to a predictable outcome as sharp as possible. For example, it is probably better to make decisions later in the game influence the outcome a bit more than earlier ones. But not so much that the early game gets boring or that long-term planning becomes irrelevant.
Fundamentally, a competitive game tries to measure the skill of the players using rules that are vastly simpler than the players. It's only natural that a human can predict the outcome. We can try to fool them, but that requires to make the game less dependent on the skills of the players. There's only so far one can go down that route before one gets a game that's mostly dependent on luck.
So I don't think this is a solvable problem until artificial intelligence is widespread enough to make the above chess solution acceptable. In the meantime, the best we can do is to make the transition from an unpredictable to a predictable outcome as sharp as possible. For example, it is probably better to make decisions later in the game influence the outcome a bit more than earlier ones. But not so much that the early game gets boring or that long-term planning becomes irrelevant.
Eventually AI will become so intelligent that it can measure all parameters from the contestants and is able to predict the outcome just after the first move thus ending the game! Maybe it will even guess the first move right.
That's nigh impossible though, if the players themselves do not play a perfect strategy :)
As I said, the AI has to consider the player parameters! Scanning the brains to get to know the knowledge of the persons, their behaviour in certain situations etc. and of course their physical states e.g. if they're stressed out or chilled and then prospect the individual moves.
And then, quantum mechanics decides to destroy a radio isotope which the AI was not able to measure because of Heisenberg's uncertainty principle, the resulting particle radiation destroys a brain cell and your result changes.
I know you know that such an AI is technically impossible. I just wanted to say, that it's physically not possible to make an AI that is 100% sure.
I know you know that such an AI is technically impossible. I just wanted to say, that it's physically not possible to make an AI that is 100% sure.
According to physics as we understand nowadays! Future Einstein will solve that issue for sure.
I see, well then that's fairly easy. I will see if I can spare half an hour tomorrow to implement that AI
Surely, humanity will thank you for saving all this pointless competition time of finding out who's better.
Not sure about that. Many players would, say, concede a game where there's a significant material advantage for one side without the pressure to make up for it. Reasoning being that it's impossible to generate pressure if one side will just exchange value for value whenever possible from now on - every exchanged piece serves double-duty as making the advantage more pronounced as well as taking steam out of any initiative. That's how a lot of chess games (would) end - relatively mechanical exchanges where the material advantage just snowballs. Waste of time for both sides.
That kind of thing is hard to pin-point for computers, though. How exactly do you define "pressure"? How exactly do you determine that there's not some sort of hole in the reasoning, however unlikely? Computers could only really calculate this for, say, five-move combinations - but that's the least interesting case, as - one - it's not clear whether the players actually saw it (as combinations are harder for humans to see) and - two - if they did, it wouldn't make much difference in terms of time whether they play it out or not.
So I have doubts of this being practical for Chess. On theoretical grounds, I feel the important thing to realize here is - the call the computer referee has to make is whether he can not see one player losing under any circumstance. Every competitive game is a game of mistakes, therefore he has to impose realistic bounds on the mistakes one side is going to make, then judge whether or not the situation can be derived as unlosable from that assumption. That's a job that's by definition harder than playing the game itself, therefore giving a computationally interesting game it can be assumed to be pretty damn hard. Allowing for conceding the game seems like a far more elegant solution.
... and I'm obviously with you on the "later decisions should have larger impact".
That kind of thing is hard to pin-point for computers, though. How exactly do you define "pressure"? How exactly do you determine that there's not some sort of hole in the reasoning, however unlikely? Computers could only really calculate this for, say, five-move combinations - but that's the least interesting case, as - one - it's not clear whether the players actually saw it (as combinations are harder for humans to see) and - two - if they did, it wouldn't make much difference in terms of time whether they play it out or not.
So I have doubts of this being practical for Chess. On theoretical grounds, I feel the important thing to realize here is - the call the computer referee has to make is whether he can not see one player losing under any circumstance. Every competitive game is a game of mistakes, therefore he has to impose realistic bounds on the mistakes one side is going to make, then judge whether or not the situation can be derived as unlosable from that assumption. That's a job that's by definition harder than playing the game itself, therefore giving a computationally interesting game it can be assumed to be pretty damn hard. Allowing for conceding the game seems like a far more elegant solution.
... and I'm obviously with you on the "later decisions should have larger impact".
The referee only has to predict the outcome accurately enough that a human can't take advantage of the new rules by playing to them, so that the same strategies work as before. Sure, once in a while the referee will declare a winner surprisingly, but then again once in a while a human will give up despite having a better chance to win. Still, I'll believe you that automated chess analysis isn't yet up to the task.
So do you agree that the "humans know the result of the game before the game rules do" problem is unsolvable without making the decisions in the early game too irrelevant or making the influence of luck too great for a competitive game? Or did I miss something in my argument? Before I wrote that post, I didn't think that the problem is as fundamental as it now seems to me.
Hm, thinking about it, distracting the player by requiring constant concentration to other details or making the games really short would be options. Fighting games, basically. But I'd guess that an observer still would be able to usually predict the result before the game ends.
So do you agree that the "humans know the result of the game before the game rules do" problem is unsolvable without making the decisions in the early game too irrelevant or making the influence of luck too great for a competitive game? Or did I miss something in my argument? Before I wrote that post, I didn't think that the problem is as fundamental as it now seems to me.
Hm, thinking about it, distracting the player by requiring constant concentration to other details or making the games really short would be options. Fighting games, basically. But I'd guess that an observer still would be able to usually predict the result before the game ends.
Well, what would an advantage be? The referee is taking away the possibility for both sides to make mistakes. There are always ways to play to that - should I try to think through a 10-move combination to get a mate, or should I go for the secure material advantage? Both involve risks doing certain types of mistakes... And depending on exactly how good your referee is at spotting certain options, you are certainly changing the equation.
And that's where I'm getting doubts about employing intelligence of any form... It might lead to "surprising" results. And generally speaking, "surprising" in this context is rarely good. I'd rather vote to tighten up the rules to predictably (!) cover up as many known-winning situations as possible.
Well, yes, it is unsolvable - but it's such a rabbit-hole of problems that it's hard to pick just one reason. The other important factor here is the meta-game... If I'm playing against a grand-master, I'm probably going to yield even after a minor material loss. On the other hand, a newbie player is likely to make enough mistakes for the game to go on further. That's another thing a good referee would have to decide.
Well, especially making winning more prone to mistakes than coming back. High risk, high reward, as I keep saying.
And that's where I'm getting doubts about employing intelligence of any form... It might lead to "surprising" results. And generally speaking, "surprising" in this context is rarely good. I'd rather vote to tighten up the rules to predictably (!) cover up as many known-winning situations as possible.
> humans know the result of the game before the game rules do
Well, yes, it is unsolvable - but it's such a rabbit-hole of problems that it's hard to pick just one reason. The other important factor here is the meta-game... If I'm playing against a grand-master, I'm probably going to yield even after a minor material loss. On the other hand, a newbie player is likely to make enough mistakes for the game to go on further. That's another thing a good referee would have to decide.
> Hm, thinking about it, distracting the player by requiring constant concentration
Well, especially making winning more prone to mistakes than coming back. High risk, high reward, as I keep saying.
> Well, what would an advantage be?
No "I don't like it when players leave before the game ends" thingy. Of course there's always a chance that the game would have turned around due to a mistake - but that's independent of who's making the call to end the game early. The referee only has to be better than the players at predicting the outcome, and probably have a setting changing the required certainty of the prediction to end the game. I can believe that automated chess analysis is still far away from that, but I don't think its too different from game analysis for automated chess players to be totally inconceivable. It's probably a matter of culture whether players would accept the referee - I could image that at least some would prefer to loose due to mistakes they make in the game instead of mispredicting the outcome. And it could speed up practice games.
But I'm playing devil's advocate here. I doubt anybody is building such a referee any time soon.
<HJK> jetzt verliere ich mindestens 40 pkt, wetten? -.-
<HJK> und das nur wegen makrousern und rumgeteame
<HJK> logisch -43 bei einem 3 vs 2 vs 1 -.-
<HJK> wie ich diese punktesystem hasse
<HJK> welcher alkoholiker hat sich das ausgedacht?
<_Andy_> ich konnte null spieln, bei dem laag...
<_Andy_> ka...
<Maddin_> was würdest du anders machen?
<_Andy_> wenn selbst ich 10pverliere
<Maddin_> HJK?
<HJK> .. 3 vs 2 vs 1 .. ich war alleine und verliere 543 punkte.. burner war im 3 er team udn verliert 10.. das würde ich z.b. anders machen
<HJK> wenn man mal was riskiert das man nciht gleich mehr punklte verliert als wenn man ncihts verliert
<Maddin_> Wie würdest du das berechnen?
<Maddin_> Mh..
<HJK> wenn sich jmd mal ein 3 vs 1 traut oder so dann verliert er mehr punkte als wenn er ein 1 vs 1 machen würde
<HJK> finde ich unlogisch
<_Andy_> man ich bin echt nur noch am punkte verlieren, durch dumme runden,
<HJK> me too
<_Andy_> du hast immerhin fast 750 ich bin bei grad mal 200
<HJK> joa, aber letztens hatte ich noch 800
<IronMan> wow habt ihr mal in der magieliga den unt
Sorry it is in German. But what do you think about this chat?
Do you really lose more points when you play 3 vs 1 as 1 vs 1?
Improvment on the team calculation is really, really complicated. There are just so many factors: Since teams affect most of the scenarios differently, and it can be a nightmare in some scenarios to actually beat 2 medium players whereas in others its simple to kill the complete enemy team of five or more, if there is no good player involved.
I think its not possible.
Lag affects and disturbs gameplay. This makes it hard for stronger players to play a number of weaker enemies, among others because they loose their speed advantage. I once lost to 2 zero pointers with a lot of lag.
Network improvement will kill some of the frustration, since the player loosing imbalanced rounds (like 1on3) will have no other word than to acknowledge that he lost only due to his mistakes or the great play of his opponents.
I think its not possible.
> <_Andy_> ich konnte null spieln, bei dem laag...
Lag affects and disturbs gameplay. This makes it hard for stronger players to play a number of weaker enemies, among others because they loose their speed advantage. I once lost to 2 zero pointers with a lot of lag.
Network improvement will kill some of the frustration, since the player loosing imbalanced rounds (like 1on3) will have no other word than to acknowledge that he lost only due to his mistakes or the great play of his opponents.
Just imagine it weren't that way; imagine playing 1v5 against much lower opponents gave you a lot of points. Then you would be able to become rank #1 in the league just by playing against large teams of noobs all the time.
To become rank #1, you are supposed to measure yourself against other strong opponents in balanced teams.
To become rank #1, you are supposed to measure yourself against other strong opponents in balanced teams.
>To become rank #1, you are supposed to measure yourself against other strong opponents in balanced teams.
So actually, we should not only discourage unbalanced rounds but make it clear to players, that there is no shame in avoiding these battles. Just do not transfer points in unfair matchings at all?
> <HJK> welcher alkoholiker hat sich das ausgedacht?
Well, I'm certainly not known for excessive amounts of alcohol consumption, that's for sure ;)
There's three effects here:
1. Unbalanced teams are really hard to judge. How hard it is for 3 bad players to win against 1 good player? Depending on the scenario, that can change *wildly*. There's really nothing the league can do here but make a reasonable guess.
2. Same goes for more than two teams. With three teams, did two teams band together first to cripple the better team? It's impossible to say from the outside.
3. The smaller teams have to get more extreme point values, just because the sum of base points lost must be equal to the sum of base points won - otherwise you could game the system to generate infinite points. Playing smaller teams can lose you a lot of points, but you can also gain a lot of points.
Bottom line: I'm pretty sure there's nothing that can be done better that wouldn't require a lot of complicated per-scenario rules. I don't want to prevent people from playing whatever they want, so the league will take whatever results people chose to generate, but if you are concerned about fairness, you should stay away from asymmetrical free-for-all situations.
Edit: And obviously, if there's a concrete result that you feel needs explaining, I can provide all details of how the league system arrived at the decision. That might involve a fair amount of maths though ;)
I like the league as it is. The only thing what annoys me is that inactiv "Pros" come back with 0 points and win against me.. this is really bad sometimes^^
But i don´t think you can change that. Maybe you can insert a highscore of every player? So that your score with them you win and lose points after a game are calculated from the current score and the highscore. But i think that makes new problems.
But i don´t think you can change that. Maybe you can insert a highscore of every player? So that your score with them you win and lose points after a game are calculated from the current score and the highscore. But i think that makes new problems.
On chess.com inactive players keep their rating, but get an increasing uncertainty number. Playing against those players provides less points for both, winning and loosing against them - whereas it is a different story for the reactivated player: He can loose (if he is not that strong anymore) or gain (if he is coming from zero points) points extremly fast, until his rating becomes more certain for the system (meaning he has played some games again).
I can imagine something similar for clonk for the future, but its of course far too much effort for just 30-50 players.
I can imagine something similar for clonk for the future, but its of course far too much effort for just 30-50 players.
Yeah, that's similar to TrueSkill and Glicko - which is pretty much the go-to rating system style once you want a bit more than Elo. It's pretty much the direct answer to Maddin's problem. I rather like giving each player one score only though - I think it makes it easier to identify with.
I agree with you there. I'd rather have a simple system that people can identify with than some weird formula nobody understands.
Also, I don't think the rating needs to be overly correct all the time. People who are overrated or underrated always make the system interesting after all and it may even allow for some not-so-good-players to get into top ranks temporarily.
Also, I don't think the rating needs to be overly correct all the time. People who are overrated or underrated always make the system interesting after all and it may even allow for some not-so-good-players to get into top ranks temporarily.
Well, to be fair, "weird formula nobody understands" would describe the current system rather well already. My hope is basically that at least the results speak for themselves, giving you a simple metric for "this is good, this is bad". When the player needs to read an article in order to even understand what the number means, that's slowly going overboard for me.
Case in point: Battle.net - they use a very good system behind the scenes, but seem to have completely given up on communicating how it works. Instead, they give out mock points and rank people on a mock ladder. It's a bit puzzling.
Case in point: Battle.net - they use a very good system behind the scenes, but seem to have completely given up on communicating how it works. Instead, they give out mock points and rank people on a mock ladder. It's a bit puzzling.
Powered by mwForum 2.29.7 © 1999-2015 Markus Wichitill