Ladder Site Online...

[KnightKunibert]

By the way, I think there are some players on ladder who manipulate the RNG. I think this is a very serious. Maybe someone has an idea how to check reported matched for strange RNG results.

Regards,

KK

[eyerouge]

[Notice: I'm reading all input and considering it as it comes in. Keep it coming, I'll leave more elaborate replies once it has piled up and can be categorized.]

eye as admin
I don't have the time to administer the Ladder of Wesnoth, which also was one of the reasons I thought the community would be better suited to solve that itself. Administering the ladder and coding it is, as has been mentioned, two different things. Some of my questions to the community overlap these roles - you can't code ladder software if you're not aware of what would work or is wanted on a ladder by it's users and/or admin. An admin of the ladder doesn't have to know how to code or be interested in developing the ladder code at all. Such a person fills totally different roles, like for example, but not limited to: Revising rules, FAQ, helping newcomers, arranging events, solving conflicts, seeing to it that someboyd updates the site, writes relevant news etc etc.

rng cheating in wesnoth
What I've written earlier still applies to my knowledge: http://www.wesnoth.org/wiki/Competitive_Gaming#Cheating

Only(?) way you'd be able to see if somebody is cheating is to look at a very high number of that persons games and see how his/her rng behave, and of course, keep track of every one of these numbers. This is a very very boring, time killing and complicated task that's not worth it if my assumptions in the above link are correct.

Glicko
Elo sucks as a rating system where people can freely choose their opposition and has so many problems with that that it needs so much work that it would, if it was patched, probably become a brand new confused system. mr russ wrote a glicko2 implementation for the ladder but the code is currently lost in space.

Edit:

Admin on vacation
An admin can't have a vacation whenever it feels like it, at least not unless all of the 3-4 admins happened to get struck by lightning or other circumstances which would explain why none of them could a) foresee their own absence and/or b) delegate the work to a temp in case of absence and/or c) resign and officially open up for a replacement. The admins are there to serve, as they were elected to do, and see to it that whatever they're admins of works smoothly, as can be derived from the title "administrator". It's not even an admin but a group of admins we're talking about now. There should not and must ever arise a situation where over 100 people have nowhere to turn and where the simplest of tasks, like i.e. changing info in the FAQ that's wrong, can't be handled because nobody knows who's in charge/responsible/has access/interested in doing it. I probably should just keep quiet and role my thumbs as do many others, but it wouldn't suite my frank attitude or interests, so I rather have an open so-called "conflict" (as some may wrongly perceive it) and shout wolf now before the beast is actually out of the forest and already chewing at your leg... after all it's passed some time now and I'm interested in what the results are, if one might ask that question without being seen as too harsh or demanding.

[KnightKunibert]

[Notice: I'm reading all input and considering it as it comes in. Keep it coming, I'll leave more elaborate replies once it has piled up and can be categorized.]

eye as admin
I don't have the time to administer the Ladder of Wesnoth, which also was one of the reasons I thought the community would be better suited to solve that itself. Administering the ladder and coding it is, as has been mentioned, two different things. Some of my questions to the community overlap these roles - you can't code ladder software if you're not aware of what would work or is wanted on a ladder by it's users and/or admin. An admin of the ladder doesn't have to know how to code or be interested in developing the ladder code at all. Such a person fills totally different roles, like for example, but not limited to: Revising rules, FAQ, helping newcomers, arranging events, solving conflicts, seeing to it that someboyd updates the site, writes relevant news etc etc.

I agree totally with you

rng cheating in wesnoth
What I've written earlier still applies to my knowledge: http://www.wesnoth.org/wiki/Competitive_Gaming#Cheating

Only(?) way you'd be able to see if somebody is cheating is to look at a very high number of that persons games and see how his/her rng behave, and of course, keep track of every one of these numbers. This is a very very boring, time killing and complicated task that's not worth it if my assumptions in the above link are correct.

Well, maybe it is possible to write a program doing the boring job? I think this task is very important ...

Glicko
Elo sucks as a rating system where people can freely choose their opposition and has so many problems with that that it needs so much work that it would, if it was patched, probably become a brand new confused system. mr russ wrote a glicko2 implementation for the ladder but the code is currently lost in space.

What's the problem by establishing a rating system where the product between probability of wining or losing the game times the change in elo is constant for any combination of players ?

Admin on vacation
An admin can't have a vacation whenever it feels like it, at least not unless all of the 3-4 admins happened to get struck by lightning or other circumstances which would explain why none of them could a) foresee their own absence and/or b) delegate the work to a temp in case of absence and/or c) resign and officially open up for a replacement. The admins are there to serve, as they were elected to do, and see to it that whatever they're admins of works smoothly, as can be derived from the title "administrator". It's not even an admin but a group of admins we're talking about now. There should not and must ever arise a situation where over 100 people have nowhere to turn and where the simplest of tasks, like i.e. changing info in the FAQ that's wrong, can't be handled because nobody knows who's in charge/responsible/has access/interested in doing it. I probably should just keep quiet and role my thumbs as do many others, but it wouldn't suite my frank attitude or interests, so I rather have an open so-called "conflict" (as some may wrongly perceive it) and shout wolf now before the beast is actually out of the forest and already chewing at your leg... after all it's passed some time now and I'm interested in what the results are, if one might ask that question without being seen as too harsh or demanding.

I just want to recall that you were desperately seeking for ladder admins about one year ago. I think admins can have vacation like everybody. The ladder is no nuclear powerplant and doesn't has to be administrated every day. But I agree that FAQs should be updated at least in a monthly basis.

Regards,

KK

[Soliton]

By the way, I think there are some players on ladder who manipulate the RNG. I think this is a very serious. Maybe someone has an idea how to check reported matched for strange RNG results.

Where are those reports?

[eyerouge]

By the way, I think there are some players on ladder who manipulate the RNG. I think this is a very serious. Maybe someone has an idea how to check reported matched for strange RNG results.

Where are those reports?

KK. maybe you could take it in private with Soliton, and get his opinion about it all before going public with accusing people. I think it's the best way to go about it, but I really have no say in this as I don't represent any person(s) involved or the LoW.

Edit:

Well, maybe it is possible to write a program doing the boring job? I think this task is very important ...

I've already written a parser an eternity ago in php for save games which could work as the base for something like that, but the problem is that even with such analysis you can't be certain that a guy is cheating. Picture that he only cheats in some situations which he deems critical or happen to be so in that current game - no program can (easily) be made to understand that without writing an "AI" for the job.

What's the problem by establishing a rating system where the product between probability of wining or losing the game times the change in elo is constant for any combination of players ?

No, and I mean no ladder system is error free or exact, and all have their own specific problems. With that said, I still would want to understand what you just suggested before answering Please give me a real example, step by step, of peoples ratings and calculations involved. While I have some skills, I'm still sorry I'm not very bright when it comes to pure math and grasping such concepts.

I just want to recall that you were desperately seeking for ladder admins about one year ago. I think admins can have vacation like everybody. The ladder is no nuclear powerplant and doesn't has to be administrated every day. But I agree that FAQs should be updated at least in a monthly basis.

Yes, I was "desperately" seeking for a ladder admin because a) I was dictator on LoW, something I wasn't comfortable with. b) I had huge changes in my life which demanded that I go and do other things c) I rather code then admin any site d) I have too little time to play/undrestand MP Wesnoth enough to feel comfortable with acting as an admin.

As I wanted to democratize and give over the project to you guys, the community, I thought I'd go about it the only way possible and be "desperate" - to announce open elections. A handful of admins were elected as a result of the community's (un)interest in actually running it's own ladder.

I have never suggested that every admin in the admin crew must be active every day, nor have I hinted any other interval. It's not a matter about set times. It's a matter of set functions. If there is nothing for the admin to do at time x then there is no problem with an inactive admin at time x. I've also written that I can only know what I can see, meaning I don't know about the many hours of support etc that the admin has given in private to members and others.

However, I mailed the official contact address found at the ladder footer on 27:th of June. That was not yesterday or so, it was over 2 weeks ago, and I also posted in here, the only thread dealing with the ladder. I have gotten no reply anywhere on the questions, more than from svek which hinted that he's not active and he doesn't have contact with anyone and knows nothing (translated: no answer).

Now, how can it take over 2 weeks for a group of 3(?) persons to answer one mail/thread? So what if they're on vacation? My whole point was that anyone in charge of anything and especially as admin, has a responsibility to see to it that whatever function he was filling is still filled if he goes on vacation or anything else for that matter (again, with the important exception for matters which would naturally prohibit a player from contacting community, like being in a car accident, dying, getting very ill, having serious personal problems, losing your internet connection and living in an area without other internet connections etc etc).

I'm aware it's not a nuclear powerplant, but I don't see the relevancy of the comparison. I honestly believe the ladder must and should have an admin that's there in some form whenever it's needed if possible. And I refuse to believe that it has been impossible to arrange so that it would be so right now.

[KnightKunibert]

Well Eye,

I guess you know best how to run ladder. I also hope you will find your dream admins ...

However, i am not interested anymore in endless discussions about ladder and RNG.

I am out of competitive wesnoth playing.

Have Fun,

KK

[eyerouge]

Well Eye,

I guess you know best how to run ladder. I also hope you will find your dream admins ...

Flattering, but I must reply that I don't think I would know how to best run a ladder, nor have I suggested I do. I only think it's healthy to have a discussion about how it's run, and I think a climate where very basic questions cant even be asked (like for instance - where is the admin, what has it done/is it doing/planning to do etc?) without people getting upset seems like a weird one to build a community on, at least from my personal perspective. It's not as much as about as finding "my" dream admins as it is about "finding the admins" to begin with.

However, i am not interested anymore in endless discussions about ladder and RNG.

I'm maybe slow to follow here, if you're not interested in the RNG question - why did you raise it in the forum in the first place? Also, if you're not interested in discussing ladder then you should of course not post here in the first place posing as if you were. After all I took the time (as did Solition) and tried to give back something that might help. I'm sorry you have taken it in a way suggested by your tone and briefness, but wouldn't apologize for helping you out.

I am out of competitive wesnoth playing.

That's off topic, but now that you mention it I can try to re-connect it to the topic: As the ladder stands now it's less competitive and accurate than it should/could be, which is what the discussion I'm trying to have aims to mend, among other things by me writing an integrated tournament system where Elo doesn't matter at all, and also in other ways suggested in the thread.

If you quit competitive playing because you suspect that people cheat with the RNG you are however probably in the wrong thread (either that or you should contact the ladder admins..oops... there are none.. my bad.... ) and I'm also uncertain how cheaters can be a real problem that's worth dealing with since, as I explain in the wiki post, they are very marginal.

In the end I really think you should do whatever you enjoy doing. If having a discussion makes you quit something then you'd probably better of elsewhere, and I'm sure there are many ways to play and/or compete without ever communicating with others about both the RNG and/or the ladder.

Best of luck.

[Yogibear]

A small essay about responsibility:

1.
20 years of project experience have taught me, that there is extreme danger of tasks not being fulfilled if there is no clear responsibility assigned. This means, that one and only one person is responsible for a certain task and that everyone is aware who that person is.

2.
Also, responsibility does not mean that the person in question has to do all the work by himself. Instead it means, that the work won't be done unless that person cares for it to be done. You can of course delegate work to someone else. However, since you are responsible, it means you have to check if the work is done. And take action if it is not.

3.
The aim of responsibility is to get work done. The aim of responsibility is not to have someone to blame if the work is not done. Unfortunately, the latter is what responsibility is used for most within companies. Let me be very clear here: It is not a shame if you are responsible and you fail (for whatever reason). However, it is a shame if you want to be responsible for something and don't learn from the mistakes you made.
In fact this is very natural. As with everything else, responsibility and what it means in practice has to be learned. And from my experience it is very hard to learn. Sadly, there are enough people who never learn it throughout their whole life.

In my opinion, the problem with administering the ladder lies in the fact, that there is no clear responsibility assigned. Instead the responsibility is shared. If i am admin and i don't have time to do something, i can always say "oh well, np, there are 3 other guys who can care for it. I am sure at least one of them will do".
I am not opposed to have a group doing the work. But i think a "single point of contact" would be better. That way, one person is the contact person for the community. He gets all the communication and decides what to do with it. And he feels the need to care for a substitute if he is not available (note that there is no reason to do that with shared responsibility).

[eyerouge]

I second everything Yogibear wrote and believe (s)he might have done it both shorther and more elegant than I maybe did.

In my opinion, the problem with administering the ladder lies in the fact, that there is no clear responsibility assigned. Instead the responsibility is shared. If i am admin and i don't have time to do something, i can always say "oh well, np, there are 3 other guys who can care for it. I am sure at least one of them will do".
I am not opposed to have a group doing the work. But i think a "single point of contact" would be better. That way, one person is the contact person for the community. He gets all the communication and decides what to do with it. And he feels the need to care for a substitute if he is not available (note that there is no reason to do that with shared responsibility

(Admin probably has a single point of contact, it's the official e-mail found in the footer, and my guess is that only one person has access to that account.) I do agree with you about the notions about the shared responsibility and that it, in this case, seem to have made everyone non-active or less active instead of sharing the workload.

Sadly, I also have experienced that IRL: Whenever a group of people share responsibility 90% of them will not take any responsibility at all and throw the burden on whoever happens to have some ambition to get the tasks done.

Here I am involved indirectly: After the elections I, stupidly, assumed that the admins would organize them self in a way that was optimal for them, that work would be delegated, and that they would create their own roles and routines. From what I've understood of the hints from svek this probably has never happened. In any case, I'm not keen om speculating on what has been done or not, so let's just focus at the present and what's official.

At the present there are no admins, for whatever reasons. Period. It's a fact. I come to conclude this from a 14 day old e-mail with no answer and also no real answer on the issue that has been discussed in this thread, and also no news on the ladder site.

The questions I'd like to pose, and that are really about democracy, are:

1. How long should it be like this before somebody is entitled to act and call for re-elections? By my own standards it is already overdue, but I have come to understand that I seem to be hardcore and unrealistic, that it's summer, that people do stuff whenever they feel like it, if they feel like it etc etc.

2. Can re-elections be called without communicating with the current admin? In my mind, yes, they can, since I/we have already communicated and the response is nothing (again, for whatever reason, there might be good ones, but who would ever know...)

3. Who can initiate the re-elections and who should handle them? Again, I imagine that it would be the admin or somebody that's actually active on the ladder, ideally. I wouldn't start them myself as I'm not so and don't feel I'm entitled to do so.

With the account of current events I realize that a part of the problem here is that I should have established rules for all this before the elections, while I still was imperator. I really think the community should create them now anyways since there is a need for them, and, if it asked me to I would of course assist with my notions on the subject.

[grrr]

(OT with respect to the ladder administration discussion)

There was a thread (http://www.wesnoth.org/forum/viewtopic.php?f=6&t=26661) asking for randomly choosen maps, and Soliton mentioned the replace_map element. Well, together with a leader teleport things came together quickly. Also, it was recently proposed here by Q.

So there is that Random Default Map Chooser on the add-on server now, which currently only chooses among 1v1 default maps. Since there are 11 of them you play each map with a 9% chance. Given that many ladder player reported problems with HGB (it still has a significant FMA) one could perhaps exclude it from the random selection? What I imagine is that for +1800 players it should be mandatory to use the RDMC add-on for all their ladder games in order to reduce exploitive map choosing. But hey - it is not up to me to make up the ladder rules =)

Edit: split to http://www.wesnoth.org/forum/viewtopic. ... 91&start=0

[Soliton]

There was a thread (http://www.wesnoth.org/forum/viewtopic.php?f=6&t=26661) asking for randomly choosen maps, and Soliton mentioned the replace_map element. Well, together with a leader teleport things came together quickly.

Nice work!
While advancing to the right scenario would be even nicer for the standard maps it should be fine.

Given that many ladder player reported problems with HGB (it still has a significant FMA) one could perhaps exclude it from the random selection? What I imagine is that for +1800 players it should be mandatory to use the RDMC add-on for all their ladder games in order to reduce exploitive map choosing.

I agree that HGB should be removed from the random selection and in general from ladder matches until it's fixed. (If there is no agreement on that maybe you can just provide several random selections in your addon?) Making random selection mandatory might indeed be nice as well.

[Yogibear]

I second everything Yogibear wrote and believe (s)he might have done it both shorther and more elegant than I maybe did.

For the record: It's "he" .

[Cackfiend]

So I was thinking about this and did a search on the forums to see if its already been discussed but couldnt find it.

Ladder started before the ability to choose a random ToD in game and I was wondering what the consensus was about using it for Ladder matches. I think the argument I heard once before was that some maps were balanced for a dawn start ToD but I don't believe this to be true. All it does is force you to change your tactics and may alter your strategy to the random ToD.

Is it ok to have a random ToD in a ladder match right now if both parties agree with it? Any possibility it could go up for discussion and be polled to possibly become a rule in all ladder matches? Am I posting this in the correct place?

I think it would make the game more fun and just better. Doing the same old thing over and over again due to a static ToD will make some ladder players get bored quickly. (for example, every time I play orcs I get the same feeling of "yay im going to get rushed as always" rather than if the ToD started out at say, Afternoon, then both players would have to alter their strategy significantly)



Also, I think adding the feature of a random 1v1 map selection (and even a random 2v2 map selection) option available is a great idea. Though im not sure it should be mandatory and you should be able to select which maps are up for the random pick.

[Tesafilmchen]

1. Agreed.

With the account of current events I realize that a part of the problem here is that I should have established rules for all this before the elections, while I still was imperator.

Well, you have my vote to be the loney imperator again till a new admin is elected.

Maybe you could announce a thread with suggestions who it should be/is maybe able for the job and a bit more specific requirements, so we can start throwing names into the pot.

2. In my opinion the current setting is fine.

3. I prefer 1xp and think 100points are a to small range. First it will reduce the choosable players alot (if you want some points). Additional there is the RNG-factor which makes a 1500 points player not exact 1500 point worth depending on the daily mood of his Holyness the Almighty RNG. Random race is also a thing which makes it more fuzzy.
So i vote for a bigger range to the 1xp cap. Lets say 300-500. It leaves more player in the list to play against and the fact that i lost lot games vs underrated players too (not only 100 points).
The 1xp-thing. In my opinion the higher rated player should gain at least 1 point. Because you learn alot from them (at least i did). If its 0 why should they take the risk to loose a lot but gain nothing?

...4. We should announce a prize for the first "God" as it will be the evidence that (s)he is a totally Wesnoth-Addict

gruß tesa

[eyerouge]

[Status on eye & admin issues: Pelopidas was friendly to get in touch with me on the 18:th of August via e-mail, answering some of the questions, which I'll also discuss further, so a dialogue is under way. ]

HardCap dismissal
In recent posts I've announced that the code now supports various hard caps and limitations that regulate how large gap in Elo there can be before the winning and better rated player gets for example 0 points as reward for beating the poorly skilled victim (or, "poor unknowing exploited frakker" if I was to put it in my borrowed words). I won't go into the details and reasoning behind this more than I already have. Sufficient to say it's a crude, but optional, feature that can be used if needed. While it solves some problems with the current LoW it also introduces some others that mainly depend on activity and the low number of players on the LoW, as such, the feature is probably best not uses and I've come to change my opinion to not recommend that it is so, if and only if the following fixes to the LoW are applied instead.

Much of the following is the result of discussions and analysis by me and chains.

Results in diff Systems
Yes, what would an Elo ladder discussion be without these frakking numbers? I present to you the result of various outcomes Elo wise between different players and using different ladder settings.



Table Breakdown
Let's explain the table briefly before we begin. Read this is you don't happen to have played plenty of rates chess or are acquainted with what happens in the table, else this post is of zero use to you except making you generate irrelevancy later on in this discussion if you wish to participate (which I invite you to do):

Case: Shows the pair of players that vs each other. Each case is composed of 2 players and each case is unique. See these as "scenarios" and ask the question - "What would happen if I beat player x and the rules were like this?"

Elo: Displays the specific players Elo rating at the time when he played the other player in the case. (Yes, it's the rating before they played the game, not afterwards)

Prov. Player: Indicates if a player is stil a provisional or not. A provisional player on current LoW is defined as a player that has played less than 10 games. Being provisional is thus translated to being new on the ladder, but necessarily into the world of Wesnoth.

Win/Loss: Shows number of Elo points the player would win or lose if he won or lost the game against his opponent in his case.

Current: This means the current LoW settings are in use. They use Provisional System 1. Current LoW also uses a factor of divisor of 800 in the Elo forumula. The higher the divisor is, the easier life is on the better rated player when he plays a player that is not near his own Elo range.

Provisional System 1:
Means that whenever a non-player playes against a provisional player, and loses against him/her, he/she would only lose 25% of the Elo he/she would otherwise lose if the same game was played against another non-provisional player. Translated: ProvSys 1 protects "veteran" players from losing much when they play provisional players.

Provisional System 2:
Is a new and alternatice way of treating games with provisional players in them which can replace system 1. In Sys 2 whenever at least one player is provisional then both players only get or lose 25% of what they would normally get or lose in a game that had no provisional players in it.

Suggested: In my suggested LoW settings I believe that these settings are relevant to solving the more serious parts of the rating inaccuracy problems in current LoW. My suggestion is that We use the second provisional system and that we use a normal Elo divisor of 400. That is also the "normal" value used in chess.

In the following I will look at the only my suggestion and compare it with the current rules at LoW and comment very sparse, as I believe the numbers speak for them self.

Comparing current formula with Suggested formula
Now, look at the table. Open it up into another window or whatever so you can easily read and watch.

Case 1 - tsr vs Dauntless:
Both are normal players (meaning, non-provisional), and there is an Elo difference of > 600 to tsr:s disadvantage. On current system Dauntless would win 3 points for beating a player thats more than 600 Elo below him. In the new he would gain only 1, due to that huge difference. On the other hand, in the new system tsr is slightly more rewarded for beating Dauntless than in the old, and Dauntless is also more heavily punsihed for losing against a player with so much lower rating in the new system.

Case 2 - 3vil vs Dauntless: In the current system the provisional player 3vil gets 53 points(!) for beating Dauntless just once. On the other hand, if Dauntless manages to win over the provisional 3vil, then he would get 4 points in that system. Here is one of the problems as we percieve it: A player should not get 53 points while being provisional. Why not? Because several reasons, from which I'll only mention these:

a) It's just one game and question is if a game should ever have such a huge impact on the rating.

b) Because only players that can win 53 p are the provisional ones in the current system, giving them a very huge advantage when competing for rank.

c) With very little effort a brand new player on the ladder can have superb rating after only 10 games. His rating would at that point be misguiding, heavily inflated, since it was pumped up while he won those games as a provisional player.

d) The previous trick can be used by anyone by registering a new account and trying to play 10 games and win them all against as highly rated players as possible. If he fails, he can re-register a new account to get a new shot at it. In the end and with little patience he will end up with very high rating after only a few games, surpassing plenty of others which have more accurate and less inflated ratings. I'll call this problem and such strategies "the Rocketeer". It can only exist if the provisional players K values are inflated, which they are in the current sys, but not in the suggested sys.

Due to 3vil being provisional and the difference to Dauntless in Elo is so huge, nothing much happens in the suggested system.

Case 3 - 3vil vs aphex:
In current system where both players have virtually 1500 p and both are provisional, they both get or lose 32 p vs each other. In the suggested system they would only get or lose 4 p. Reasoning for prefering the suggested system is the same as in case 2, no need to repeat anything here more than the fact that it looks even more wicked to allow brand new players on a laddet with equal ratings to get so many points for playing each other.

Case 4 - Dauntless vs Coriolan:
No changes worh mentioning between the two systems: This matchup is between two non-provisional ratings with almost the same skills according to the rating. It's the definition of the "normal" game.


Summary
The new suggested system solves the following problems which the current LoW has:

1. Rocketeers: Problem with players just joining and rocketing into the top as a result of inequal and inflated K-values: The normalizing of the provisional players K values totally solves this. Such a normalization if a part of the new suggested system, but could ofc be adopted even in the current system.

2. Cowards, Startegists & Exploiters: Problems with some cowardly players that are supe protective about their ratings, picking only opposition they are almost guaranteed to win over, slowly but securely gaining more and more Elo points, and passing way better players which are on the ladder and which play anyone without disciŕiminating. This problem exists in any Elo system where players can pick their opposition them self, because Elo was created as a system for random pairings, something which leads to the Elo rating beeing way more accurate than on a ladder where pairings aren't forced at all. These risk-minimizers-point-maximizers are a problem because they bottom-feed and make the ladder lose credibility/function. Their activity is mainly possible because of the rocketeer problem, and later when they get out of provisional, it's made possible by playing newcomers on the ladder that are provisional. So, whenever the exploiters win they usually win plenty of points even though they beat a player that is much lower rated than them (this is a fact because of the divisior that is set to 800). And, whenever they lose, they lose almost nothing at all - because they are protected to only lose 25% (this is how provisional system 1 works). In the suggested system this is not so.

3. Gap problems: With players getting poinst for beating opposition which they are very superior over rating wise: This exists in the current system as a result of the divisor being set to 800. Example of this is best seen in Case 1 where a player with more than 600 elo still gets 3 points for beating the one that was 600 below him. The suggested system uses a divisor of 400 and thus shrinks the gap that the ladder will tolerate between players. With the suggested settings a game where the winner has a rating of 600 more he/she'd only get 1 point for that win.

All in all, I'd recomend upgrading the code to the latest with the suggested settings, meaning Provisional system 2 used with normal k-values for provisional players and also a divisor of 800.


Re-rank

If we would restart the ladder and re-rate every game from scratch accoridng to the suggested system, and include most players (by making them active) we'd end up with the result below (using the database from the 15:th of August 2009 in the sumulation below):



1) Most renowned players are still near their previous rank.

2) The whole ladder is deflated since the current LoW is wrongly inflated.

3) The more a player has played against provisional players, the more deflated his/her rating has become.

4) Also, the more players a player has beaten that are much further away downwards from his/her own rating, the more that players rating has been deflated in the re-ranked ladder.

I'm not saying that a re-rank should or must be done in order to impement the suggested settings which I've argued for in this post. But if you were to rerank it you'd end up with the results above. I do however believe a re-rank should be avoided, and that the new settings should just be implemented and announced as such - without touching the evaluation of how people have behaved in the past. Thus, the full power of the new settings can't easily be determined or understood by looking at the ladde re-rank since the players in the ladder didn't know about the suggested rules and played with different rules. The power of the suggestion is seen in the table screenshot, and the re-rank shot is only included as a reference and not as something that's really relevant.

Edit: Please feel free to replace the names of any player with x y z or whatever suits your eyes. They were selected for no particular reasons except for their ratings on August 15:th. Nothing if aimed or directed towards a specific player, and nothing is suggested about anyone of them as persons, nor their skills.