Ladder Site Online...

[Noy]

I’d just like to chime in here, as an interested party who has observed the ladder from afar. As I’ve stated to a number of people before, I don’t think the ladder’s best contribution to wesnoth has been some sort of perfect ranking system for players. I’m not saying the system can’t be improved, as some of the suggestions you have here should be expanded on. But there will always be problems with such a system, as with any other type of competition because people will dispute what makes a "true" winner. I’m not here to get into an argument about this, as its not central to the point I'm making.

Instead, I would argue the best part of the ladder is providing interested 1v1 players a system that allows them to find competitors who adhere to basic standards of conduct and skill. Part of this as to do with the controlled competitive element it introduces. I’d encourage you to think about that when crafting new regulations on the ladder. Curtailing the number of ladder games one can play would directly undercut the main strength of the ladder, which should be discouraged.

[ElvenKing]

Here are my thoughts on the issue. Match spamming will always be a problem to some degree. It's nature of how the ELO ranking system works. Since you get points for winning each game; the people who play the most games will be able to get the most points.
Even a top level player could match spam against people who were of inferior skill and get a much higher rating than one who was an equally skilled top level player and only played a few games against high level opponents.

The goal of any changes should be to lower the effect of inflated ratings on the ladder. This can be done in three ways; one as Yogi suggested, by lowering the points you get for beating an opponent that is ranked far below you, or two as Wintermute suggested; requiring you to beat opponents above you before you are able to attain a higher position on the ladder, or three, as is currently the case by limiting the number of games you can play.

I think both Yogi's and Wintermute's have merit, but as they are currently proposed; Wintermute's is much better as it doesn't completely take away the incentive for a higher ranked to play a lower ranked player, which in my opinion would be a bad idea, as it can be a great learning experience for a lower level player to play a higher level player.
Reducing the amount of points a higher level player gets for beating a lower level player is perhaps not a such a bad idea; it's just zero points is going to far in the kind of match ups that you would need to change to limit problems caused by match spamming.

A match cap is a pretty bad idea in my opinion, as Noy said it goes against the major strength of the ladder.

[Eskon]

People who want to play a ladder game against each other must not be kept from doing so for any reason. No matter what problem it would solve. The ladder is to ensure that players find proper opponents among the ladder community; but the ladder players are not there to ensure that the ladder gives an accurate ranking - "Don't ask what the players have to do for the ladder, ask what the ladder has to do for the players." In other words, we have to find ways to solve the problem in a way that does not restrict players, or at least that restricts them in the least amount possible. The game per week cap is simply not it.

The thing we have to do is make "playing 30 games a week against weak players for 1-2 points each" or similar a non-workable strategy. Interestingly, the recent change that made it harder for high-ranked players to lose points against low-ranked ones made this a much better strategy than it probably once was, because the hit you'd take from losing has become lower. Reducing the stakes of a match encouraged grinding behaviour more than it reduced it, I think. Maybe simply undoing that change would help inhibit match spamming behaviour already?

Besides, match spamming behavior would hardly be eliminated by the cap. If it is a good strategy to beat weak players over and over for 1-2 points each, then it still is a good strategy even if you can do it only 14 times a week. It would take longer for it to take effect, sure, but the effect would still be there, and the benefit of the strategy too.

We need to make that strategy bad, either by imposing a hard cap after which you don't get points for a victory and can only LOSE points (say, 500 points better than the opponent or something), or by increasing the stakes for the more skilled player. But trying to prevent people from using this strategy in the way instituted is throwing the baby out with the bathwater.

EDIT: This is not supposed to offend anyone. I can only stand in admiration as to how well the ladder system works already, and the constant effort to improve it only increases this admiration for all the people involved in that effort. I'm just voicing my personal view on things.

Incidentally, regular tournaments that would count for the ladder would be a terrific thing to have. I know I'd play.

[ElvenKing]

Besides, match spamming behavior would hardly be eliminated by the cap. If it is a good strategy to beat weak players over and over for 1-2 points each, then it still is a good strategy even if you can do it only 14 times a week. It would take longer for it to take effect, sure, but the effect would still be there, and the benefit of the strategy too.

Ah, yes I forgot to mention that in my post. For the match cap to be really effective you would have to make it something like 5 games or possibly even lower. That would simply be unworkable.

Interestingly, the recent change that made it harder for high-ranked players to lose points against low-ranked ones made this a much better strategy than it probably once was, because the hit you'd take from losing has become lower. Reducing the stakes of a match encouraged grinding behaviour more than it reduced it, I think. Maybe simply undoing that change would help inhibit match spamming behaviour already?

That was quite rightly instituted becuase of the effect that luck has in wesnoth. Maybe it might cause more match spamming; I doubt it though, most players would have done it just as much before the change I think. The thing is when most people match spam it's not that they particularly want to have an inflated rating; it's they want to play games of wesnoth and the ones they find the most fun are the ones with ranking points on the line. People are often limited by who is available to play with. If their aren't any high level players around; they have no choice but to play the lower ranked players.

This is why I particularly like Wintermute's idea of having to beat some people above you before you actually move up the ladder. It would require people to seek out and play against higher level players a few times at least, otherwise they would get stuck at a certain spot on the ladder with a higher ranking than people above. This would easily identify them as people with inflated scores, but it doesn't punish them for wanting to play.

[Zlodzei]

For a lot of them , i'm easy points, but they don't come to play me. I was suprised yesterday when Pagan asked me for a game. I told him that maybe after i finish, he told me the game would end shortly. I won, i looked for him in the lobby, and he wasn't there.

Was around 2-3 AM and game went bit slower than I expected.

However, to my mind there is actually no problem with so-called "point-farming". 1st thing is that even if you are playing well you won't automatically get to top. You still need to play some games to get to your actual rating and changing of elo coefficient INCREASED IT VERY MUCH. It is absolutely natural that those who plays more active get their rating FASTER (not getting overrated, they just get it faster). And I would like to underline that nobody of very active top players (like Fosprey, Pelopidas, plk2) never ever refused to play because they were afraid of loosing points. I can see absolutely nothing wrong in how it is currently going and absolutely against any game caps. The only solid decision is in auto-assigning rivals, but wesnoth is not ready for it (and won't be ready anyhow soon).

[Doc Paterson]

nobody of very active top players (like Fosprey, Pelopidas, plk2) never ever refused to play because they were afraid of loosing points. I can see absolutely nothing wrong in how it is currently going and absolutely against any game caps.

As far as I can tell, that's not the issue. The issue (I think) is that you take someone like plk2, above, who, if you look at his stats, has an average gain of 1.21 per match. Now that does not mean that he is playing players that are so weak that he usually gains something even close to that in any given match. That would be bottom feeding, and I really don't think that any of the so-called spammers are willfully doing that. I think it means that he's playing practically everybody, and is losing a real lot, and is winning a real lot. Imagine a sequence of +10,-8,+9,-5, etc. etc. Now naturally, playing 300 games is going to give him an inflated rating (relative to players who gain maybe 6-10 on average, per match, and have played maybe a quarter of the matches he's played), because there is that very small average increment of gain per match.

I don't know what the answer to all of this is, really.

As for the match cap though, it would be interesting to see some data on how many ladder players have even exceeded the 14 per seven days, historically. Would it even be that many? Maybe it would then be interesting to see how many had ever exceeded 21 in 7 days? I can't imagine it being more than 4 or 5, in all of the 1,000+ membership.

Btw, I thought Wintermute's idea was very interesting.

[Cernunnos]

To Doc Paterson :

Gallifax, 18 won games from 2008-09-10 to 2008-09-16.
Svek, 15 games from 2008-05-09 to 2008-05-15.
Fosprey
Loser
c4rLOs8
Pelopidas

6 out of the ten top players.

[Gallifax]

I am pretty sure I exceeded the 14 games per week quite a lot . Especially when I was playing bfw stil quite excessivly.

I remember days when I played 7 games on ladder on a single day.

In average though if you really want to have a cap I think 20 games might do.

Edit: Hmm thinking about it again. Even back then I remember I would have liked to play more:P But lack of opponents stopped me from doing so. So maybe 25?:D

[Doc Paterson]

To Doc Paterson :

Gallifax, 18 won games from 2008-09-10 to 2008-09-16.
Svek, 15 games from 2008-05-09 to 2008-05-15.
Fosprey
Loser
c4rLOs8
Pelopidas

6 out of the ten top players.

Is it only these 6? I don't know and haven't looked into it.

(They are good players, but your saying that they are "six of the top ten" doesn't mean too much, because it stands to reason that good players who play so much would be towards the top, because of the nature of the point system- what I was talking about earlier.)

[Cernunnos]

@Doc Paterson,

I counted only in the ten top players of the ladder, i won't check for the 1228 players.

But if you want an approximation, let's say those who played more than a 100 games do that quite often, that's 18 players.

If you count only the active ok, you go to 11, but that's 11 among the 110 total active players, which means 10%.

Think you can add some for people who did that once or sometimes only due to their schedule and won't play the week after like said pelopidas (if i remember well) so that their total number of games is medium and you can't see them at first sight.

Well, here we are with stats, the problem, if it ever has been one, is still unsolved.

Thx, see ya.

[Wintermute]

@Doc: I fully agree that a game cap only impacts a few players (it would never affect me, for instance - I'm arguing against it on principle ). But it seems unfortunate that good players who want to play a lot are being punished for what seems to me to be an issue that does not bother that many people. In addition to that list, I know that Becephalus (when active) might play 10 games a day sometimes, and he was looking for games with 1600+ rated players.

Though at the same time, he was frustrated that he NEEDED to play that many games in order to keep his rank. So he might also be arguing for some modification to the ranking process (to prevent the need to play so many games to keep a top rating).

Personally, I am on the ladder far more to find good games, and the rankings are an interesting and fun bonus for me.

[Doc Paterson]

@Doc: I fully agree that a game cap only impacts a few players (it would never affect me, for instance - I'm arguing against it on principle ).


(Becephalus) was frustrated that he NEEDED to play that many games in order to keep his rank. So he might also be arguing for some modificat

I think that the idea you posted earlier was a rather nice one.

[Wintermute]

I think that the idea you posted earlier was a rather nice one. [/quote]

...not to be confused with any of my OTHER crazy ideas...

[Zephyrus]

I hope you all will let a newcomer to the ladder add his two cents. After reading the conversation above, here are my thoughts:

1. We already have a ladder statistic that captures precisely what is being called 'match spamming'. It is the average points gained per win (NOT per game, as was pointed out above). Looking at the top 20 players, nobody has this value lower than 7 (most are much higher). So it appears there is not a problem. I really think we don't need to worry about a purely theoretical problem, though I know others feel differently. One thing that could be done without changing any rules is to list the average points gained per win for each player on the main ladder page. Then one could list players in reverse order of this stat and see quickly if anyone is 'spamming'. You could even make a player's name show up in a different color if it fell below, say, 3 pts. (or whatever is deemed reasonable) as a mark of shame). Of course, it is expected that top players will have lower pts. gained per win (call it PGPW), since that's how the system is designed. Thus it is important to make sure that this doesn't discourage top players from at least occasionally playing low-ranked players.

2. The only thing not captured by PGPW is the effect of wins against new (provisional) players. One could play many games against new players whose 'true' skill level puts them around, say, 1200, but one still gains significant points since their rating is 1500 or so. A solution (and a general improvement to the system) would be to calculate all the point gains/losses for matches against provisional players based on their rating when they finish their tenth game (that is, the provisional player would gain/lose points in their first ten matches as now, but their opponents loss/gain of points would be as if they had played them at their post-provisional rating). I hope that makes sense. It would complicate things somewhat, but I think would be worthwhile even independent of the potential for 'spamming'.

3. When starting out on the ladder, it's nice to be able to play many games in a week in order to (a) learn; and (b) find out how good you really are (i.e., gravitate toward your 'true' rating).
Just my thoughts; I realize most of you have much more experience with both Wesnoth and the ladder, but I thought I'd throw them out there.

[Wintermute]

I hope you all will let a newcomer to the ladder add his two cents. After reading the conversation above, here are my thoughts:

1. We already have a ladder statistic that captures precisely what is being called 'match spamming'. It is the average points gained per win (NOT per game, as was pointed out above). Looking at the top 20 players, nobody has this value lower than 7 (most are much higher). So it appears there is not a problem. I really think we don't need to worry about a purely theoretical problem, though I know others feel differently. One thing that could be done without changing any rules is to list the average points gained per win for each player on the main ladder page. Then one could list players in reverse order of this stat and see quickly if anyone is 'spamming'. You could even make a player's name show up in a different color if it fell below, say, 3 pts. (or whatever is deemed reasonable) as a mark of shame). Of course, it is expected that top players will have lower pts. gained per win (call it PGPW), since that's how the system is designed. Thus it is important to make sure that this doesn't discourage top players from at least occasionally playing low-ranked players.

Newcomers are always welcome. And you are quite correct that average points per game is one measure of so-called "match spammers". However, it also goes down as players move to the top of the ladder, where they are playing more and more players below their level (fewer to none remain above them), so the highest value of average points per game comes when a good new player joins, and then it will go down as they reach the top. Thus it is not a perfect indicator. Also, it says nothing about the true skill of a player. Just because a player is playing lots of games vs. lower ranked opponents it no indication that the player is not any good - so it may not be fair to punish them simply because they have a low value. Furthermore: many of the top 20 players are below 7 (several are below 3), so you may want to look into that a bit more.

Not sure if I was fully clear there, but as I understand it the value of average points per game is really only a good indicator of whether a player is playing (and beating) other players at thier own level. In other words, a high value indicates that a player is moving up the ladder "quickly" still, whereas a lower value indicates that a player is probably pretty well ranked (winning about many games as they are losing vs. their peers, or otherwise maintaining their current rating).