forum.webdiplomacy.net

webDip dev coordination forum / public access todo list
It is currently Sat Nov 18, 2017 9:26 pm

All times are UTC




Post new topic Reply to topic  [ 26 posts ]  Go to page Previous  1, 2, 3  Next
Author Message
PostPosted: Mon Sep 22, 2008 7:09 pm 
Offline

Joined: Tue Aug 26, 2008 8:46 pm
Posts: 249
I'm sorry for the lack of a response. In short, schoolwork...


Top
 Profile  
 
PostPosted: Tue Sep 23, 2008 3:45 pm 
Offline

Joined: Mon Sep 01, 2008 9:04 am
Posts: 25
I've got a couple observations I figured I'd throw out there (small warning first, I haven't read all the posts on this topic but I've read enough to get the general gist of things).

First, you may want to look at what Blizzard has done with their multiplayer ladders and possibly the World of Warcraft Arena and (old) honor system. I believe they use a modify ELO system for their multiplayer ladders. Both their honor system and multiplayer ladders use a decay so that inactive players will slide off the top of the ladder after some time period.

Second, I'm not sure the suggested estimated value for taking over a CD country is realistic. For example taking over a country with 1 or 2 SCs in most cases will probably mean you're eliminated or at the very least don't gain any additional SCs whereas if you're lucky enough to take over a good situation with 10+ SCs you may very well win the game from that position. It may be that there's some SC value where you're expected to break even and above that gain SCs while below that point you're expected to lose SCs. Perhaps we could figure out a SQL query to investigate this (I'm not that familiar with the database to be able to create one currently - after a quick glance at the tables it doesn't appear to be possible since I don't see any differentiation between original players and ones taking over CD countries).


Top
 Profile  
 
PostPosted: Wed Sep 24, 2008 1:05 am 
Offline

Joined: Sun Sep 07, 2008 8:56 pm
Posts: 22
This said, the players who went CD needs to have lost the game with 0 SC's... The penalty needs to be WORSE than if you were eliminated in 1902.


Top
 Profile  
 
PostPosted: Wed Oct 01, 2008 7:39 pm 
Offline

Joined: Fri Sep 05, 2008 7:02 pm
Posts: 22
What's the status here? Are we any closer to implementing ELO?


Top
 Profile  
 
PostPosted: Wed Oct 01, 2008 10:24 pm 
Offline

Joined: Sat Sep 27, 2008 12:00 am
Posts: 44
Just wishing to advertise an improved proposal building on the ELO one (I thought a new thread would have been better not to hijack this one!)
viewtopic.php?f=12&t=75


Top
 Profile  
 
PostPosted: Thu Oct 02, 2008 9:37 pm 
Offline

Joined: Tue Aug 26, 2008 8:46 pm
Posts: 249
We are not really closer because of the start of my school term, but now we reach a middle point, we might consider continuing.


Top
 Profile  
 
PostPosted: Fri Oct 03, 2008 1:11 pm 
Offline

Joined: Fri Sep 05, 2008 7:02 pm
Posts: 22
Great! I'm still working on the icons so no rush. We might as well get it right the first time.


Top
 Profile  
 
PostPosted: Thu Nov 06, 2008 2:57 pm 
Offline

Joined: Sun Sep 07, 2008 8:56 pm
Posts: 22
I just want to interject a few points that may calm some of the worries out there. Unfortunately I want to add some questions and point out some more things that need to be considered.

The first big worry is implementation and changing user's built up measure of ego (currently points). I've been thinking about this quite a bit as my ELO ranking (in the recent test) put me much higher than my point stash (at it's highest). This problem is the reason why ranks (which seem to be simply an artistic project) are a crucial implementation. Points, again, are great and serve exactly the purpose they were meant for and must not be abolished. ELO, is a system that can be brought in to rank player in a better way that points can at the moment. The reason that ranks are important is that they will mask the numbers. A player should never see their raw ELO score or their numerical ranking. Instead, the ELO score is used to rank the player from 1 to 964. You then compare a player's ranking relative to the number of "rankable" players (this would exclude newbies) and establish their percentile. You then give them the appropriate rank from the decided percentile intervals.

This has a few advantages, players don't notice numerical gains and can't get picky and the number of ranks can be easily changed, as well as other.

An example:

I'm 29 out of 964 ranked players. Remove MarekP from the list to assign him a unique medal for the top player.

My percentile: .0301

Here's a table I prepared:

Rank Percentage Percentile
1 0.1 0.1
2 0.4 0.5
3 1.5 2
4 3 5
5 5 10
6 7 17
7 15 32
8 19 51
9 21 72
10 28 100

I would then be in the 4th rank, and get the appropriate medal.

I have some more points to get on, but I have to pay attention to class now...


Top
 Profile  
 
PostPosted: Thu Nov 06, 2008 9:03 pm 
Offline

Joined: Sat Sep 27, 2008 12:00 am
Posts: 44
Churchill, I agree with the general idea behind your post (but if you have read my Glicko-2 proposal and other posts you should know it by now). I think there are two features (and additional two issues) on the table though:
Features
1. - Ranking system
2. - Matching system
Issues
A. - Retaining newcomers to join tons of new games and CDing them.
B. - Discouraging people to perform multi accounting

At the moment the D-Points work as ranking system (not based on skill, but they still provide a rank people feel motivated to climb), matching system (there is a post to games that makes impossible for a number of people to join) and address problem A. Contrarily, D-Points encourage multi-accounting, as a multi-accounter will have tangible advantages in doing so.

My personal take on the four above is:

The ranking system should be a system that is "pondered" (in the mathematical sense), i.e. that gives the chance for all players to have the same chance to reach the top, regardless to the amount of games played or to the time since they have been enrolled in phpDip. D-points dramatically fail at this. I personally would use a ranking system based on skills (so that the best players are also the top players) but I would find acceptable to have a system based on something else, provided that all players would be given the same possibility to climb the ladder to the top.

The matching system should be strictly based on skills. This is based on my perception that players are gratified by winning games that challenged their skills, and are fair across the gameboard, rather than by games in which the final outcome is based by some players having had a dramatically quick expansion due to a totally inept player, and another player having been locked in-between two masters...

Preventing newcomers to flood games and then CDing can be achieved with "ad hoc" measures. In another post I presented a simple idea to implement, where the number of games one can join is determined by the number of games finished and by the number of games "CD'ed".

Discouraging people to perform multi accounting can also be achieved by "ad hoc" measures. One simple idea (to be explored) is to limit the percentage of games two player can play together. However if the ranking system would be based on skills, multi-accounting would bring little advantage, as repeatedly beating a "mock player with a low rank" would minimally increment the "real player" ranking score.


Top
 Profile  
 
PostPosted: Fri Nov 07, 2008 1:05 am 
Offline

Joined: Sun Sep 07, 2008 8:56 pm
Posts: 22
mac wrote:
Churchill, I agree with the general idea behind your post (but if you have read my Glicko-2 proposal and other posts you should know it by now). I think there are two features (and additional two issues) on the table though:
Features
1. - Ranking system
2. - Matching system
Issues
A. - Retaining newcomers to join tons of new games and CDing them.
B. - Discouraging people to perform multi accounting


Yes and no, details below

Quote:
At the moment the D-Points work as ranking system (not based on skill, but they still provide a rank people feel motivated to climb), matching system (there is a post to games that makes impossible for a number of people to join) and address problem A. Contrarily, D-Points encourage multi-accounting, as a multi-accounter will have tangible advantages in doing so.


DipPoints are not an effective ranking system, which is why there's such a positive side to bringing ELO into the game. What the DP are good for is their original use, matching. DP serve their original purpose faultlessly. We shouldn't take points away, especially as the current inflation allows players to move higher and higher as newcomers and those who lose add to the pot.

Quote:
My personal take on the four above is:

The ranking system should be a system that is "pondered" (in the mathematical sense), i.e. that gives the chance for all players to have the same chance to reach the top, regardless to the amount of games played or to the time since they have been enrolled in phpDip. D-points dramatically fail at this. I personally would use a ranking system based on skills (so that the best players are also the top players) but I would find acceptable to have a system based on something else, provided that all players would be given the same possibility to climb the ladder to the top.


Points can be accumulated if a new player plays well quite easily. This is a subtle strength of the inflation, as new points get added, they will never be scarce. ELO should still not be disabling, a measure such as a raw score floor of 60 might be introduced to ensure a player never regressed further than where they start. This would create an inflation in raw ELO, but since the idea is that no-one will see it, it will all be relative.

Quote:
The matching system should be strictly based on skills. This is based on my perception that players are gratified by winning games that challenged their skills, and are fair across the gameboard, rather than by games in which the final outcome is based by some players having had a dramatically quick expansion due to a totally inept player, and another player having been locked in-between two masters...


This is where points are good, as they allow a player to choose how much they'd like to wager. ELO serves the function of rewarding depending on opponent strength. Again the marriage of the two systems is by far ideal, as it will give rank incentive to play with the big-leaguers, but the pots can still be whatever the players choose.

Quote:
Preventing newcomers to flood games and then CDing can be achieved with "ad hoc" measures. In another post I presented a simple idea to implement, where the number of games one can join is determined by the number of games finished and by the number of games "CD'ed".


The phrase "Preventing newcomers" is never the start to a point we like around here. Although the spirit of your argument is the spirit behind points. Don't use ELO to match games (or have it as an option), continue to use points. If you want to make sure that someone has put some effort into a game before they join yours, make a game with an entry fee of 101.

Quote:
Discouraging people to perform multi accounting can also be achieved by "ad hoc" measures. One simple idea (to be explored) is to limit the percentage of games two player can play together. However if the ranking system would be based on skills, multi-accounting would bring little advantage, as repeatedly beating a "mock player with a low rank" would minimally increment the "real player" ranking score.


Again: bad. Don't prevent players from legitimate behavior of playing with someone else. We have IP checks, etc... that can be added to a list of suspicious behavior regulated by moderators. You are correct on the rank issue, except that the difference would have to be very large for the expected result to be >18 SC's; you're also forgetting that other player's rank could be in the mix.


All in all you worries are quite valid. Most of your issues can be answered by keeping the points system around (which is my full suggestion). I'd encourage an optional rank matching feature (both max and min), so that if Rait wants to play a 50DP game with fellow attachés (suggested example rank) he can, rather than putting his ELO rating at risk of a bunch of n00bs (of course he's likely to win against novices). Some might say that this will create groups that players cannot earn themselves into, it's not a new problem. DP are no better, and we also have pasworded games, closed groups of experienced players have not been formed yet, so I'm sure they won't be formed in the predictable future.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 26 posts ]  Go to page Previous  1, 2, 3  Next

All times are UTC


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Group