Ladder Site Online...

Discussion of all aspects of multiplayer development: unit balancing, map development, server development, and so forth.

Moderators: Forum Moderators, Developers

Post Reply
User avatar
eyerouge
Posts: 380
Joined: June 29th, 2007, 4:37 am
Location: wtactics.org
Contact:

Re: Ladder Site Online...

Post by eyerouge »

Thanks for all input people. It's been an interesting and intense reading. I'll summarize my thoughts this far, based on what I've read. I'll do it using key-words as topics and keep it short.

match-spammers
Who is and who isn't a so called match-spammer isn't important. The definition I use of a match spammer is however a version of the one Doc Paterson suggests: A person that gets his good rating mainly because he plays very many games compared to a majority of all other active players. I also include so called bottom-feeders in the definition since they tend to play many games, but, don't have to to be bottom-feeding. Let's not waste any more time and effort at pointing fingers on people and/or defending our playstyle. Nobody has question anyones reasons for a) playing many games or b) playing lower ranked opponents.

Match-spamming could very well be a problem that is just theoretical. We could in the real life, on the ladder, at current time have zero match spammers. Even then the ladder system would still allow the problem with match spammers to become real one day, if and when such persons appear. Experience with people, internet and also game development tell me that the problem will arise sooner or later unless it's mended.

A match-spammer isn't equal to being a bad person or an idiot. It could be in some cases, but there is no real relation. It simply means that the person plays very many games. We could perhaps have used another word filled with less negative connotations, like hard-core competitor or whatever.

Lastly, it is also hard to measure who is one and who is not unless the person is very extreme. What I'm saying is that even if we don't currently have the problem, we are still inviting it over by not being prepared and/or by not working on a solution on it.

the cap: amount of allowed games within a period
I'm aware and I agree with the following: The cap is crude. The cap isn't a solution to all problems. The cap is bad in several ways. The cap must be complemented by other, better, functions and mechanics. All that is true. Nevertheless, there are some various reasons for why it might be a good idea to have some kind of cap. I will now present my main one below this paragraph. It won't show that a cap is necessary, nor do I try to defend the cap. I personally would like it removed. However, it is worth some really serious consideration:

the real problem solved by a cap
Nobody in this thread seems to mention any solution whatsoever to what I have described as a problem in previous posts. This is the problem:

When running a public ladder where you want many competitors, and using a system that somehow rewards those who play the most, you will make the ladder unattractive for the non-hardcore players, the players that don't "match spam" and the ones that don't have serious amounts of time to spend on playing.

Some of you may not see that as a problem. You might ask "So what? Of course people that play the most should be reawarded, of course they shouldn't be held back etc etc."

The problem is that the match spammers are always a minority per definition. (If all players play 1000 games / day then a player that playes 1000 games / day isn't a match spammer. One that would play 10 000 would be. And so on.) Despite the fact that they are a minority, they will still get to dictate the pace of the ladder. Why? Because in order to keep up with the match spammers you must play approximately the same amount of games as them (or play less but win more often and/or win more points). If you don't keep up, you could never become a top player on the ladder.

So, my hypothesis is that by having a few match-spammers. we create an environment that will, in the long run, scare of people from the ladder or scare them from joining in the first place. In both cases it could very well depend on the fact that, since they belong to a majority of players that don't play super-many-games due to real life or whatever, there are match-spammers around. This all translates into one thing: A steeper "entry level" for players that want to battle for the top positions found in i.e. the highest 30% strata of the ladder. A higher requirement on any player that wants to be able to compete in a serious fashion.

The questions I want to ask many you that take the time to read this is: Is what I write plausible? Doesn't it really make some sense? Can I be correct that the pace on the ladder will be set by the most hardcore-players? And if I am, wouldn't I also be correct in fearing that it could, in the long run over time, scare away more people than it would attract?

I know much of my reasoning is highly speculative, but it has to be thought of nevertheless. I have to think, and have to question everything, else I can't guarantee you guys or myself that the ladder is taken into the right direction. I also don't think I'm far fetched, really. I seriously believe that people don't register and get active on a ladder where you have to spend countless of hours and be a die hard fan to ever get even a chance to compete in a serious matter. To me it doesn't sound illogical at all.

This is also where the cap comes in, for that reason. A cap slows down the pace of individuals. In contrast to what some may believe, it must not mean that the ladder, overall, is slowed down. Ladder won't be slowed down if more players join and play with the cap in place compared to less players without the cap. Again, we can all imagine both scenarios. Which is the most reasonable and probable?

I know that the cap will make slow down players from getting their correct rating, but, then again - how many of them will really be slowed down? Per definition of the match spammer, a small minority would. A majority wouldn't since a cap isn't supposed to affect the average player in any way at all.

So, if what I write is true, we are at a crossroad where we need to choose between giving the match-spammers total freedom that can potentially harm the ladder from developing user wise, and restricting them somewhat in order to make it easier for more people, the average player, to join and start competing.

Again: I know a cap doesn't solve all problems on earth and that we need other solutions to other problems. But, when it comes to setting the pace of the ladder, a crude cap like the one in place seems to do the trick. Question is, do you follow what I present, and do you share my notion that it might be an idea to have some limitations to how many games a player is allowed to play within x days?

skill measurement
The system I'm probably leaning towards is one which would add an additional Elo. Let's call the current Elo for Elo and the new additional Elo for RE, which is short for "Real Elo" in lack of a more creative name ;)

This is, in short, how I imagine it could be done. Much if this was probably presented by me or chains some time ago, so some of you might recognize the thought:
  • Elo would still work like it does today. Nothing would be touched at all. There is also no relation whatsoever between RE and Elo. They are two totally separate statistics.
  • RE works like Elo with only one exception: It only changes when you won/lost against a player that is within a certain range of RE-points in relation to your own RE-rating.
  • RE ranges are made up of a highest and lowest number. The highest number tells us how high rating, at it's most, your opponent may have. The lowest tells us how "crappy" opponents you are allowed to play against.
  • Example 1: High and low numbers are both set to 100, for the sake of simplicty and this example. If my RE is 1500 it would mean that I can play any player that is between 1400 to 1600.
  • Whenever you play a player that is within the RE range your RE would be altered, using same formula as Elo does, but using the RE ratings.
  • Whenever you play against a player that is below or above the RE-range your RE-rating stays unchanged. The only thing that changes is your Elo. Actually, your Elo always changes, liek it does today (unless of course you have a really really high Elo and the opponent has a really really low one, then even in usual Elo your rating wouldn't change if you won).
The effects of the RE should be interesting. RE will, more or less, incorporate many of your suggested ideas. It would only change when you played people that are near your own rating. At the same time, by keeping current Elo, we also still allow people to play outside of the ranges and keep on doing what they already are. RE won't be very restricting and if it is then the allowed ranges can / should be changed by admin. By also allowing the usual Elo games we actually don't restrict at all.

At the same time I think RE would bring us something new in the stats. It would also, probably, bring us an overall better ladder which encourages players to play closer to their own skill set. In effect, again, it should lead to more people developing in Wesnoth and a somewhat higher quality of gaming on the ladder.

Speaking of which, the question will arise: Which rating will the ladder ranking use - Elo or RE, if RE is implemented? I don't have an answer at this moment. Personally I'd say RE, if it was up to me. I would however advise the admins to go with whatever a majority of the community says and wants.

Another question will be: "What happens with my current Elo? And do the already played games count towards my RE if/when it is introduced?" Answers are - nothing happens to your Elo, it will work like it does today. And yes, all already played games will count towards your RE. By looking at your game history you can probably get some crude picture if you are highly RE-ranked or not. Else you will notice when the time comes, if it does.

Feedback on the RE-concept?

(Reasons I don't want to go with a "you must beat x players above you"-strategy is that I believe that the RE suggestion is better, less restricting and doesn't lead as easily to "freeze-outs" as I've discussed in previous posts.)

Pelopidas
Posts: 18
Joined: November 5th, 2008, 7:42 pm

Re: Ladder Site Online...

Post by Pelopidas »

Eyerouge,

if I may shortly sum up some of your main concerns:
1. There is a proven inequality of points against relatively good players who do not play too many matches - true in the short run.

Well, I would take this seriously, but this is not solved by introducing a match cap - because then you prevent active players from playing. Adjusting a match cap to the median of players will directly reduce the happiness of all players above that median and it will indirectly reduce the happiness of those guys who do not find an opponent due to the lack of players who can play.
There is one solution and if you take your rating system from chess servers you should know this:
RD-value: An active player gets a slower RD than an inactive player. The RD directly determines how many points you win/lose. This means that if a very active player wins 10 points, a quite inactive player wins 20 points for the same match and a newcomer wins something like 100. I talked about this already above. It is the solution of the chess servers and I find it pretty appropriate - there nobody would come to the idea to put in a match cap.

2. The problem with match spammers (?)
WHO ON THE LADDER is a match spammer? I clicked through the list and did find those guys with the most matches played:
plk 303
KnightKunibert 295
Gallifax 238
Pelopidas 211
ArtemisX 189
Fosprey 168

The rest of players is near or below 100 played matches. I would NEVER ever say one of those guys is a spammer - all of them I have matched frequently and they never looked for playing only against newcomers.
It rather seems like you are fighting against shadows who are not there than being fighting real match spamming.
If there arised single cases one could still think about taking action.
AND in addition I would say that you are taking action against those guys who keep the ladder running: These 6 guys account alone for far more than 1000 matches on the ladder - that means that you are talking about one fifth of played games or the other way round - 1000 people would not have found an opponent without them (very frequently one does still not find an opponent on the ladder).
Going through the list I also cannot spot a too strong relation between points and matches played - that active players are indeed better should be no miracle.

3. Slowing down active players does not slow down the ladder?
Funny thing: First of all I recruited something like 30-50 players for the ladder. Of course some of them are not too active, but with a match cap I will certainly not waste my spare matches any more to a newcomer - granted that other active players recruit the same way by doing open ladder matches you will reduce the stream-in of new players.
In addition - as I already said. It is still not easy to find always an opponent and thus reducing the activity of active players (whom I highly esteem as opponents on the ladder) you are making the ladder die.

4. The match cap does not harm anybody?
IT DOES HARM. I am quite nervous about if I have to think about when I can play a match and when I cannot - I love the freedom about the ladder and that I am (or better was) free all the time to play ladder when I want to.
You should be aware of the high number of negative answers and in your place I would remove the match cap as soon as possible.

5. To the RE-system:
Well, a lot of bureaucracy - if you like this, o.k.. In all leagues and all ladders I know there is nothing equivalent to this and there are reasons for it. Your rating should be composed of MANY matches against opponents and not a selected small number with the proper rating.
All your concerns would already be solved by an RD (not RE)-dependend rating. Players quickly approach their proper rating and this is pretty independent of the number of matches played.

Fosprey
Posts: 254
Joined: January 25th, 2008, 8:13 am

Re: Ladder Site Online...

Post by Fosprey »

Yes as i said, right now there are quite a few players, the most active that can't play, and i have to say those are the one that host games, so i'm pretty sure the ladder activity will drop, and thus it can become a slippery slope effect.
I think this cap have more chances of killing the ladder than any of the things you mentioned.
as i said it's not only me that won't be playing until monday, it's players that won;t find me to play, and because a lot of ladder players join games but don't host themselves. Really, making the players that keep the ladder more active, unactive, seems very harmfull.

User avatar
eyerouge
Posts: 380
Joined: June 29th, 2007, 4:37 am
Location: wtactics.org
Contact:

Re: Ladder Site Online...

Post by eyerouge »

Pelopidas
1. There is a proven inequality of points against relatively good players who do not play too many matches - true in the short run.
I'm not sure I understand what you mean by that, but my main concerns can be described as one: Hard-core players that play many games, which set the tempo of the ladder, and makes it almost impossible for any normal player on the ladder to compete with/for ranking, which in return may have a negative impact on overall ladder popularity since people wonät join/compete in a place where you have no real chance of competing (again, since a majority of people can't play as many games as the minority can). I explain all this in greater detail in my previous post above.

You haven't addressed the plausibility in my reasoning, nor have you proven me wrong. Logic dictates, to some extent, that if I see a ladder, and I want to compete on it, I won't bother if I understand that I have no real chance of doing so due to me being an average joe (meaning, not being able to spend as much time on the gaming as a hardcore player would). What I want is a ladder open for the majority, where you don't ever have to feel the pressure that you must play all the time and many many games in order to be competitive. Without a cap, and with current Elo system, you are forced to adjust to the tempo set by the hardcore players. A majority of users can't do that. Hence, I draw the conclusion we have a situation that must be solved - yes.
There is one solution and if you take your rating system from chess servers you should know this:
RD-value: An active player gets a slower RD than an inactive player. The RD directly determines how many points you win/lose. This means that if a very active player wins 10 points, a quite inactive player wins 20 points for the same match and a newcomer wins something like 100.
(I'm not sure I followed what you suggested, so please nevermind this part of my post if I misunderstood you)

I believe that is one of the less attractive ways to solve the problem as it totally distorts/destroys the Elo rating: If you get different points for the same game, with the only difference being how active (meaning, how many games you have played within a time span) you were, then you clearly distort the Elo. How active anyone is has no logical effect or necessary connection whatsoever to that persons skills. Time in between games has no correlation to real skills. Why?

1. Because people are different and require different amounts of practice and learn differently fast etc etc.

2. Because people usually don't have alzheimers.

3. Becasue how active I am on the ladder doesn't say anything whatsoever how active I am in Wesnoth. Just because I appear to be low-active on the ladder it doesn't mean I have low skills. In reality I could play hundreds of games that were not reported, outside the ladder. Every now and then I would play a game on the ladder, and then, since I am marked as "low-active", I would get more points due to that. Totally unfair of course, and also generating distorted and unrealistic numbers, just adding to all other problems instead of solving them.
2. The problem with match spammers (?)
WHO ON THE LADDER is a match spammer? /../ I would NEVER ever say one of those guys is a spammer
In my previous post I answered this:
eye wrote: Who is and who isn't a so called match-spammer isn't important. The definition I use of a match spammer is however a version of the one Doc Paterson suggests: A person that gets his good rating mainly because he plays very many games compared to a majority of all other active players. I also include so called bottom-feeders in the definition since they tend to play many games, but, don't have to to be bottom-feeding. Let's not waste any more time and effort at pointing fingers on people and/or defending our playstyle. Nobody has question anyones reasons for a) playing many games or b) playing lower ranked opponents.

Match-spamming could very well be a problem that is just theoretical. We could in the real life, on the ladder, at current time have zero match spammers. Even then the ladder system would still allow the problem with match spammers to become real one day, if and when such persons appear. Experience with people, internet and also game development tell me that the problem will arise sooner or later unless it's mended.

A match-spammer isn't equal to being a bad person or an idiot. It could be in some cases, but there is no real relation. It simply means that the person plays very many games. We could perhaps have used another word filled with less negative connotations, like hard-core competitor or whatever.

Lastly, it is also hard to measure who is one and who is not unless the person is very extreme. What I'm saying is that even if we don't currently have the problem, we are still inviting it over by not being prepared and/or by not working on a solution on it.
AND in addition I would say that you are taking action against those guys who keep the ladder running: These 6 guys account alone for far more than 1000 matches on the ladder
The cap doesn't target specific individuals. It treats all equal. I also fail to understand why they can't continue to play because of the cap. The only thing that has changed is their pace, in some cases. I also argued in my previous post that in the long term a cap can very well lead to, in total, more games being played since more users will be competitive due to the fact that we give them a chance to be so.
First of all I recruited something like 30-50 players for the ladder. Of course some of them are not too active, but with a match cap I will certainly not waste my spare matches any more to a newcomer


I'm glad you wrote that, as it's also a part of my point: By not playing newcomers, and letting newcomers play other necomwers/equally skilled players, you will contribute to an overall higher quality of the ladder, since you will now start focusing more on who you play, according not to me, but to yourself quoted in the above. From my perspective it's okey and a good thing. And even if you didn't, it would still be good. The cap can have the effect you mention, and in some cases it will, and in others it won't - it depends on what type of player we are talking about. Some will become more strategical, others not. In any case, that effect isn't primarily why the cap was implemented. Why it was implemented was explained in my previous post in this thread. Again, you, as everyone else this far, has ignored my main point, which still is "How do we create a ladder where as much people as possible can compete?". For that to happen we can't allow an environment where pace is set too high and where there is a relation between Elo and pace. If we allow that, the ladder will become a place only for those that are hardcore. While that is tempting for many, it isn't for me. I rather see a well-balanced ladder, in which there is room for the hardcore anyway - the cap doesn't hinder them from playing - it equalizes number of theoretical opportunities each player could play, slowing them down, so a majority of all the other players can catch up and thus, compete. I also stated all this in my previous post, but I haven't seen anything that refutes it. I'm not saying I'm right. I'm, just suggesting it might very well be like I propose and that it seems worth the consideration. This far I am the only one considering it, and "everyone" else seems upset that they can't play however many games they can in a timespan of their personal liking.
- granted that other active players recruit the same way by doing open ladder matches you will reduce the stream-in of new players.
If that's the main way ladder players are introduced to the ladder and the main way a majority of active players get to know it, then yes, you may have a valid point here. I am however not familiar with the facts and if it's true or not, so I can't refute it, nor will I accept it. Interesting perspective though, well worth a thought or two. Thanks.
In addition - as I already said. It is still not easy to find always an opponent and thus reducing the activity of active players (whom I highly esteem as opponents on the ladder) you are making the ladder die.
Probably depends on how a majority of players find opponents and plan their games. There are a number of ways - IM, game server, mail. In the future there will be built in challenges on the ladder site as well.

I do agree with you, Cap could cause that kind of problems, especially if people don't adjust to it or if it's set too low. In such cases it would of course have to be revised.
4. The match cap does not harm anybody?
IT DOES HARM. I am quite nervous about if I have to think about when I can play a match and when I cannot - I love the freedom about the ladder and that I am (or better was) free all the time to play ladder when I want to.
I have repeatedly stated that the cap affects a minority of hard-core players, and that it isn't designed to have an impact on the majority of average players (if it does, it needs to be higher). You may be one of them, I wouldn't know. Iäm sure you will adjust and that you won't be as nervous once you get used to the idea that the pace isn't up to you and there is a max limit for it built into the system. The freedom of the ladder is still there,. Actually, by limiting yours somewhat, we will potentially create an environment where more people will be able to compete - thus the sum of freedom will be larger, just as the sum of activity overall could become larger. (I'm nor saying it will or won't, just that it is plausible that it can.)

You can still decide yourself when you want to play your ladder games, there is no timer or timezone control or such a thing. The only thing you can't decide currently is how many games you can squeeze in within 7 days once you reach 14.
5. To the RE-system:
Well, a lot of bureaucracy - if you like this, o.k.. In all leagues and all ladders I know there is nothing equivalent to this and there are reasons for it. . Your rating should be composed of MANY matches against opponents and not a selected small number with the proper rating.
RE has nothing to do with "bureaucracy". It is just plain Elo that sees to it that people play other players around their own skill class. Nothing complicated really. I argue that it a) makes the ladder more interesting for many competitive players (many have already in this thread given expression they want some version of such a system) b) is way more accurate than the Elo we use now c) heightens the quality of the average ladder game d) just maybe even leads to player getting skills faster.

I don't care the least what is and isn't implemented in all other ladder and league systems. From what I have seen most of them suck for several reasons (as do the one I code on, but again, that's why we have this discussion and why I'll continue to code on it and have done plenty of it already). As I have stated in my previous posts in this thread: How many that think a thing says nothing about how correct it is. I'm not saying that I can't learn from what's out there. I sure can, and I'm doing it right now on this forum among other places. What I'm saying is that what you have seen proves nothing in this case. Also, ladder admins might have different values: For some maybe the sum of games played on the ladder is the most important thing, or that there are some players on it that play hardcore. For others it maybe is something else. There are many explanations for why people choose to create a certain system, and none of them would be any good on it's own.
Your rating should be composed of MANY matches against opponents and not a selected small number with the proper rating.
Yes and no: Elo, which we use, is broken when we have a system where people can more or less handpick their opponents. This is a known and uncontroversial problem to anyone who has the slightest idea of the problems associated with Elo. This in itself is actually enough to motivate a change of the current system. Ideally the system would suggest random opponents and people would find each other and play. In real life, it would probably not work well or at all, hence I suggested a system close to Wintermutes/chains, which even seems way smoother.

I have also proven to you in my previous posts how Elo can be exploted by minimizing risk and maximizing profit. It's a mathematical fact and nothing any of us should waste breath on debating. The effects of that exploit would be less of a problem with RD.

User avatar
eyerouge
Posts: 380
Joined: June 29th, 2007, 4:37 am
Location: wtactics.org
Contact:

Re: Ladder Site Online...

Post by eyerouge »

Fosprey wrote:Yes as i said, right now there are quite a few players, the most active that can't play, and i have to say those are the one that host games, so i'm pretty sure the ladder activity will drop, and thus it can become a slippery slope effect.
I think this cap have more chances of killing the ladder than any of the things you mentioned.
as i said it's not only me that won't be playing until monday, it's players that won;t find me to play, and because a lot of ladder players join games but don't host themselves. Really, making the players that keep the ladder more active, unactive, seems very harmfull.
Yes, it might become a slippery slope effect. Or, it might be that few players are active since mos players sooner or later realize that they don't stand a chance competing against the most active players, as I have reasoned in other posts above. But, I agree with you that it might have the effect you suggest. We'll see. Game activity was around 100 games per week for a very long time. Now, the latest weeks, it was above - 200 and even 300 games per week. If it falls very much below 100 games per week and you guys don't "strike" on purpose ;) we'll have proof that what you say is true. If so, cap must be set to a higher level, but, if we do that, then it will eventually lose it's only meaning, so we could might as well remove it all together. I guess time will tell.

The reason for why you can't play for a couple of days is that you played plenty of games before the cap was introduced, so now you are forced to wait an extra long time because of the introduction. In the future it won't affect you that way, it can't. So this is a one-time thing. In the future you'll know that the cap is there and can choose how to distribute your games within the timespan.

Also, I refute that the cap makes the active players non-active. The cap makes the super-active players less active. That's a very huge difference.

Zephyrus
Posts: 8
Joined: November 6th, 2008, 7:13 pm

Re: Ladder Site Online...

Post by Zephyrus »

On a practical note:

I am currently on the server and there are two open ladder games waiting for opponents. But I can't play because I've hit the cap. Note that my TOTAL number of ladder games is 14 (all played in the last week).

User avatar
Doc Paterson
Drake Cartographer
Posts: 1973
Joined: February 21st, 2005, 9:37 pm
Location: Kazakh
Contact:

Re: Ladder Site Online...

Post by Doc Paterson »

If wonder if RD could be applied retroactively (I'm guessing probably). Regardless, I think it's a nice idea.

On a completely unrelated note, it seems like whenever I go to take a look at the ladder site, I see something like this in the spotlight:
spotlight wrote:Comment by KnightKunibert:

Drakes vs Undead. This time Tondo was unlucky with the combination of races. Funny game again :-). Thanks.
Comment by Tondo:

Bad luck continued... this time by giving me drakes vs undead. Hopeless combination, disastrous outcome.
I want to tack a response onto a message like that, but arg, it's not possible! I also sometimes see messages that were cut off- For example, Leocrotta's comment on his recent match with Wintermute. I don't think it would be so bad if the ladder had a forum. :hmm:
I will not tell you my corner / where threads don't get locked because of mostly no reason /
because I don't want your hostile disease / to spread all over the world.
I prefer that corner to remain hidden /
without your noses.
-Nosebane, Sorcerer Supreme

Pelopidas
Posts: 18
Joined: November 5th, 2008, 7:42 pm

Re: Ladder Site Online...

Post by Pelopidas »

you possibly misunderstood me, eyerouge - there are indeed some players (whom I know well and was told by them) not playing any more because of the cap and I myself will have my decision if I want to play in such a restrictive ladder.

Best
Pelopidas

User avatar
eyerouge
Posts: 380
Joined: June 29th, 2007, 4:37 am
Location: wtactics.org
Contact:

Re: Ladder Site Online...

Post by eyerouge »

Zephyrus wrote:On a practical note:

I am currently on the server and there are two open ladder games waiting for opponents. But I can't play because I've hit the cap. Note that my TOTAL number of ladder games is 14 (all played in the last week).
1. They could perhaps... ehrm.. I dunno.. play each other? lol... pardon my sarcasm here, it just strikes me as a waste of life (in many cases) to open a game when there already is one around. Unless of course you have some interesting motives.

2. Your total number of games isn't relevant for the cap, as you have noticed now. Cap regulates the pace, and treats all equal. Including newcomers. There might perhaps be a point to give them more space while they are still new... I'm not sure, and I think we're already discussing very many notions at the same time so I dare not think of it at this point.

3. You will be able to play tomorrow. Guaranteed.
Doc Paterson wrote:If wonder if RD could be applied retroactively (I'm guessing probably). Regardless, I think it's a nice idea.
If RD = something that changes how we rate the games depending on how active you are then yes, it wouldn't be a problem: We'd just code it and then re-rank the ladder, as we did the last time with the provisional player changes. It would all be 100% accurate (rather, as good as the Elo and patches on it that we apply).
Doc Paterson wrote:I want to tack a response onto a message like that, but arg, it's not possible! I also sometimes see messages that were cut off-
Agree it would be nice. It is somewhere on the list, but very low prio. Have a [censored] of other stuff to deal with first. :( Cut-off messages depend on a cap on maximum chars allowed in a comment is set to 500. We could, probably should, allow more of them. I actually thought the text-box wouldn't let you type more chars when you reached the limit...I just tried it out and it allows you to type, which really sucks and has to be fixed. I must have dreamt that I already did this with java once.. hrm...
Doc Paterson wrote:I don't think it would be so bad if the ladder had a forum.
I suggested that a subforum in here should be for the ladder but it was denied, and we're also not allowed anymore threads than this one. Personally I'm totally against creating a separate forum. It takes 3 min for me to setup phpBB but I won't do it unless the community really believes it's a good idea. I don't do so, because it would split up the Wesnoth community and centers for forum communication, and that seems as a bad idea for several reasons: Multi accounts, multi sites and a lot of info players would miss out on which they'd reach/see easier if they were all in one place. I also don't think that splitting up in group after group after group is a good idea when it comes to communications and overall community activity. Unity, and within that, diversity. Trying to have a common meeting ground is usually a good choice. I don't think the ladder would be "more official" just because it got it's subforum in here (each subforum has it's description - it could even say in there that it's not a part of the official site/forum and lent out space), nor that regular mods would have to babysit it since ladderers could do so. In any case, I'm not taking sides and don't care much about that specific detail - just venting my thoughts on the subject.
Pelopidas wrote:you possibly misunderstood me, eyerouge - there are indeed some players (whom I know well and was told by them) not playing any more because of the cap and I myself will have my decision if I want to play in such a restrictive ladder.
Well, I'm sorry to hear that, especially since it's irrational: They could still play if they wanted to, but clearly they don't think the cap is high enough since they changed it to 0 games within all days, of their own free will.

I suggest that you, and they, run for ladder admins. You have the whole 7:th of november (today) to mail me and tell me you want to run for the admin posts especially if you haven't done so already. I know a couple players have, but really, all are welcome. I will, as announced much earlier and before the cap, quit as admin and only work on the project as coder. On the 8:th you guys will either have 3 new admins, or elections for those 3 positions (if more than 3 persons applied for the posts). In either case, as admins you could decide to do whatever you want with the ladder and with the cap. So... ;)

Until then though (1 more day), I'll do the same with it, and do it in a fashion I think is the best for the time being.

Eskon
Posts: 184
Joined: August 12th, 2008, 2:21 pm
Location: Esslingen, Germany

Re: Ladder Site Online...

Post by Eskon »

Many of us are arguing this on principle, in that you shouldn't keep players from playing on the ladder if they want to. It appears you simply don't agree that this is the gospel truth - that there is a constellation in which allowing a player to play a ladder match is harmful to the ladder overall. I think I see your point now. But I think that any solution that doesn't restrict the freedom to play of anybody is preferable to one that does. If you think there is a problem with the overall pace of the ladder because some play at a higher pace than others, you should reduce the impact of the pace, but you shouldn't try and stop people from playing at their pace unless absolutely, positively, bold-letters-for-emphasis-requiringly necessary. The ladder needs to accomodate all players, regardless of their pace, as inobtrusively as possible.

Unless you can conclusively show that the level of necessity has reached bold-letter status, so to speak, you will of course face opposition, and rightfully so, I think. Other methods, preferably ones that don't put restrictions on players and only fiddle with the way score is calculated, should be considered and tested, and if they help reduce the problem, they should be chosen over the match cap.

As of now, I doubt the match cap has had or will have the positive effect you desire. Statistics that show that positive effect might be able to convince some.

(Again - I'm merely voicing my opinion here. I can say that of all the changes made, the one to reduce the impact of losses of higher-ranked players has discouraged me more from playing than any other change will be able to undo. I certainly am not encouraged to play more games by the match cap - it doesn't affect me personally, and... seriously, why should it make it any kind of difference to me or any other newcomer? There's still a buttload of players above me in the ladder. Place the match cap too high, and it will make no difference except for the ones who suffer from it. Place it too low, and you discourage players from making the effort to get into the upper ranges of the ladder, as well as have even more players suffer from it. Me? I'll wait for ladder tournaments. THOSE would encourage me to play more, and I somehow think I'm not the only one. Tournament situations are what I'm used to from chess, and it is what ELO is designed for after all.)

Zlodzei
Posts: 44
Joined: January 6th, 2007, 10:31 am
Location: Belarus, Minsk
Contact:

Re: Ladder Site Online...

Post by Zlodzei »

Some feedback: this evening I tried to play some ladder games and was unable...because C4rlos (who was the only decent player online) could not play anymore :)

I do not see how it helps ladder ;)
I can see you!...

Pelopidas
Posts: 18
Joined: November 5th, 2008, 7:42 pm

Re: Ladder Site Online...

Post by Pelopidas »

detailed reply:
You are concerned by players getting points from playing too many matches.
Obviously you did not really get the crucial basics about Elo and other systems right.
Players EXCHANGE points in an elo system, which means, that as soon as all players reached their fair rating you can only rise to a point where your rating is fair, too (+ some random walk around this point).
The only problem is that players reach their fair rating relatively slowly in the ladder, which is indeed a source of distortion. By increasing this adaption, e.g. by introducing the RD-variable you get this problem done.

Since you did not grasp the meaning of the RD-variable, I will try to explain this in a more mathematical way:
A normal rating system assigns points due to some formula like:
deltapoints = K*(W - E)
where: deltapoints is the change in points from a match
W is the win pointage, i. e. win: W=1, loss: W=0;
E is the expectation value for a win against a player at a certain point difference, i. e. if the player is 200 points lower than you, you look up in your magic table and see that a player with 200 more points wins at 90% probability. Thus the bracket gives 0.1 for a win and -0.9 for a loss.
This also ensures that you cannot gain points by match spam - your expectation value for points will always be 0, if your rating is fair (below 0 if your rating is too high).

The thing that now comes into play is the constant K. K gives the opportunity to control the speed of adjustment. The reason for letting this vary is:
1. Players are definitely wrong rated when entering the ladder.
2. Players' ratings get more and more equal to their skills if they had several matches (random noise problem).
3. Players' ratings go wrong due to unnoticed changes in their skills if they are inactive.

That's where the RD variable enters the scheme. You assign every player an RD variable, which in fact contains the information how precisely his skill level is determined.
A new player has a very high RD and a player with many matches has a low RD. For the determination purpose you add a small increase of RD with time, so that the RD always continues to increase. It is also appropriate to set a minimum RD, so that still significant points come out of a normal match.

Now we need to work that into the point constant K.
The dependencies we need are plainly clear - if you match a new player or a player whose skill level is uncertain due to long inactivity the information about your skill is low - vice versa for a well-determined player as opponent. Thus the higher the opponent's RD, the lower K should be for you in this match.
The other way round - if your skill level is not well determined (your RD is thus high), K must be high, since by the outcome of this match, the relative increase of information about your skill level is high.

Thus in the end, you need K increasing with your RD and decreasing with opponent's RD. For example the setup:
K = k0*(yourRD)/(opponent'sRD)
does such a job (k0 = 10 should be fine)
Play against a newcomer with RD = 300.0, your RD = 30.0, then K is pushed down to 1, while against an active player it is around 1. On the other side for your opponent K is 100 if he is a newcomer.

Getting back to
deltapoints=K*(E-W)
and assuming that your opponent is a newcomer 200 points lower than you (we assigned 90% chance of win) the points given will be
if he wins:
you: -0.9*K = -0.9 points, he: +0.9*K = 90 points
if he loses:
you: 0.1 points, he -10 points.

For this setup float points would be appropriate - one still shows the integer values, but accounts for the player's ratings in float, to avoid unjust changes of points.

About the RD - a fine setup might be:
start with RD = 300.0 and for every match played you divide the RD by a factor of say 1.4, so with the following matches, RD will decline subsequently as: 300.0, 214.3, 153.1, 109,3 and so on.
A minimum RD of 20 or so is appropriate, so RD cannot fall below 20.
Every day the RD of a player reincreases by 1, so after 10 days of inactivity, RD has risen by 10. This gives more rating weight to matches of low activity players to still keep them adjusted with changes in the ladder level or changes in their skills.

I hope this is not going unread nowafter I invested a lot of time in this post.

One remark in the end: The fact that you receive many negative comments about a cap and virtually no positive comments should make you think about if this is a proper thing for the ladder. You want to be democratic, so take the democratic vote of the posted comments serious!

Best,
Pelopidas

Yogibear
Retired Developer
Posts: 1086
Joined: September 16th, 2005, 5:44 am
Location: Hamburg, Germany

Re: Ladder Site Online...

Post by Yogibear »

I think Eskon raised a very good point, indicating that with a match cap we are rather addressing the symptoms than the root of the problem.

Eyerouge, you said that a match cap slows down the pace of individual players and therefore puts a hurdle for match spamming. I completely agree with you about that.
eyerouge wrote: Again, you, as everyone else this far, has ignored my main point, which still is "How do we create a ladder where as much people as possible can compete?". For that to happen we can't allow an environment where pace is set too high and where there is a relation between Elo and pace. If we allow that, the ladder will become a place only for those that are hardcore.
Not the pace itself is the problem, but instead the coupling to the Elo. Once that coupling is no longer there, pace can be chosen completely individually without any negative impact at all. In my opinion, "Real Elo" (RE) does exactly that.

Actually, the Elo system itself has a built in RE already. It is just so that the range is so high (670 points) that at the moment it covers almost the whole ladder. No match spammer would indeed play an opponent more than 670 points below him.

Edit: I think the magic number was 670, i can't find any reference at the moment, though. If is not, just replace mentally that number in my text :wink: .

I dare to say (without proof but it makes sense to me), that if we had not 100 but maybe 1000 players, the ladder would spread out much more (say from 1000 to 3000). Which at the same time means, that players who are like 500 points below you are still relatively much closer to your skill than they are at the moment. Which in return means that because of that you would eventually lose some games and match spamming becomes pointless.

However, we have not reached that point yet (and maybe never will). Until we do, i think shrinking those 670 points range to something ensuring better that players with equal skills play each other is one good way to address the root, not the symptom of match spamming.

Many have stated that even without limitations it is hard to find opponents. I can second that from my own experience. As well as the fact that there are few people that host games. A smaller Elo range won't make that better of course. We *may* find, that this solution is not feasible for a ladder of 100 players. If that is the case, i think we should head for other ideas (like Wintermute's for example). But i think it is worth a try.
Smart persons learn out of their mistakes, wise persons learn out of others mistakes!

User avatar
Wintermute
Inactive Developer
Posts: 840
Joined: March 23rd, 2006, 10:28 pm
Location: On IRC as "happygrue" at: #wesnoth-mp

Re: Ladder Site Online...

Post by Wintermute »

eyerouge wrote:
Doc Paterson wrote:I don't think it would be so bad if the ladder had a forum.
I suggested that a subforum in here should be for the ladder but it was denied, and we're also not allowed anymore threads than this one. Personally I'm totally against creating a separate forum. It takes 3 min for me to setup phpBB but I won't do it unless the community really believes it's a good idea. I don't do so, because it would split up the Wesnoth community and centers for forum communication, and that seems as a bad idea for several reasons: Multi accounts, multi sites and a lot of info players would miss out on which they'd reach/see easier if they were all in one place. I also don't think that splitting up in group after group after group is a good idea when it comes to communications and overall community activity. Unity, and within that, diversity. Trying to have a common meeting ground is usually a good choice. I don't think the ladder would be "more official" just because it got it's subforum in here (each subforum has it's description - it could even say in there that it's not a part of the official site/forum and lent out space), nor that regular mods would have to babysit it since ladderers could do so. In any case, I'm not taking sides and don't care much about that specific detail - just venting my thoughts on the subject.
As someone who pushed for the ladder to have it's own subforum (in the moderator debate that took place over the issue), I can say that your request was not ruled out flatly by the development community - there were quite differing opinions on the matter. I wish there were a subforum, and I imagine that there may be at some point, but IIRC there was not a solid concensus one way or the other, and as a result no action was taken.
"I just started playing this game a few days ago, and I already see some balance issues."

Zephyrus
Posts: 8
Joined: November 6th, 2008, 7:13 pm

Re: Ladder Site Online...

Post by Zephyrus »

2. Your total number of games isn't relevant for the cap, as you have noticed now. Cap regulates the pace, and treats all equal. Including newcomers. There might perhaps be a point to give them more space while they are still new... I'm not sure, and I think we're already discussing very many notions at the same time so I dare not think of it at this point.
Right; I didn't mean to say that not letting me play after 14 games was a bug; just that it was ironic, given the purpose for which the cap was introduced.
3. You will be able to play tomorrow. Guaranteed.
Great!

I like Pelopidas' ideas, although I don't know how much they differ from the current approach since the actual formulas are not completely detailed in the FAQ. Do they appear somewhere else?

Are we allowed to start ladder-related threads (besides this one) here in the MP Dev forum? If so, perhaps we don't need a ladder forum.

Post Reply