forum.webdiplomacy.net

webDip dev coordination forum / public access todo list
It is currently Wed Nov 22, 2017 10:11 pm

All times are UTC




Post new topic Reply to topic  [ 20 posts ]  Go to page 1, 2  Next
Author Message
PostPosted: Fri Mar 06, 2009 9:18 pm 
Offline

Joined: Wed Jan 28, 2009 4:51 pm
Posts: 31
Ok, I just spent some considerable time going through avinoam's complicated rating system and I think there was alot that could be taken from it. So I thought we could possibly implement a couple of more simple features that addressed some of his concerns, however inadequately. I believe TheGhostmaker made the point that perhaps we should salvage parts of his proposal to help solve some problems.

Something that really struck me from his post:
Some flaws with the current system:
* Reward surviving - the difference between 0 and 2 SC is too small
* Reward winning - the difference between 16 and 18 SC is too small
(We want to:)
* Reward surviving [L]
* Reward winning [L]
Pay-out => continuous generalisation of WTA and PPSC, that gives better separation of victory, survival and defeat

Ok. with some of his critical points that I found interesting thus isolated, without hiding behind any crazy code proposals, I think adjusting the WTA and PPSC system would be the most effective to meet these goals.

To get to the proposal: One can still create a PPSC or WTA game. However, they can also choose to implement a 'hybrid' system.
Basically the idea for a hybrid game is that it starts with a PPSC base and then rewards for certain behaviour can be added. So you select 'hybrid' and then boxes of things you can reward drop down. Although certain things for draws can be thrown in, lets keep it simple:

Reward for Winning the Game
Reward for Surviving the Game

You then get percentages (every tenth percentile) to apply to each reward. These will be taken out of the pot.
So for example: Let say we give Winning Reward (30%) and Surviving Reward (20%)
The end result of our 500 pot game is Centurian 18, TheGhostmaker 15 and Figlesquidge 1.
Under the current PPSC system we would be allotted 265,221 and 15 points respectively.
Under the Hybrid system only 50% of the pot is PPSC (132,110 and 7).
30% of the pot is allotted to the winner (150 to me)
20% of the pot is equally distributed to survivors (33 each).

Now the Distribution is 282, 143 and 40 respectively. Wow TGM really lost out.

So whats the big difference? Its an incentive issue. I just changed the difference between first and second from a measly 44 to a whopping 139. Although I had an incentive to go for the win before rather than just draw with Ghost, its just increased alot.
Figle, meanwhile, may have lost interest in the game, but staying alive becomes more of a priority for him, so he will submit moves.

Although I think most of us are firm believers in player self-regulation. If you feel the choice of different percentiles makes things too complicated, just make the hybrid a very simple half WTA and half PPSC, thus increasing the incentive to win, rather than a wimpy strong second or playing for a draw, which there have been alot of complaints about.

I look forward to hearing you comments. Hope it wasn't too long and its in the right place.


Top
 Profile  
 
PostPosted: Sat Mar 07, 2009 12:26 pm 
Offline

Joined: Tue Aug 26, 2008 8:46 pm
Posts: 249
I liked the system 'til I lost 78 points :P

Now, I agree that there are probably better "standard" systems than PPSC. Basically, we have, in my view:

Beginner-Style (PPSC)
Expert-Style (WTA)

WTA is great for a simulation of the game as played in FtF, but PPSC doesn't offer the sort of system we should want for beginners, where 16 centres is so close to 18. I would therefore replace the "standard" offering with one where winning is important, surviving is important and having SCs is important, as suggested here.

It does mean I have to create a new formula for ghost-rating though =(.


Top
 Profile  
 
PostPosted: Sat Mar 07, 2009 8:19 pm 
Offline

Joined: Wed Nov 12, 2008 8:48 pm
Posts: 29
Centurian I'm glad something could be salvaged from my proposal. :P

A minor quibble.
I don't think winners should get points PSC ever. I think it promotes game play that is vicious and I think that it's not in spirit with the original game.
PPSC should be changed to PPSCFS (FS = for survivors), with a fixed percentage going to the winner.

Did you read my newer thread on my system?
http://forum.phpdiplomacy.net/viewtopic.php?f=23&t=137

I don't think the code changes are that significant. You don't need to change the data structures or have extra overhead.
You have an update function at the end of every game just as you do now.
You have a function calculating who can join just as you do now.

Also, if you're making minor code changes.
If you're staying with the current rank system, when it comes to rankings please calculate based on total points (including those invested in games). They don't even have to be adjusted for current status in games. Currently your system penalises players for participating in games since they're ranked by their left over points. This is annoying and is a minor deterrent to play.
Unfortunately this might require a change in the data-structure, including total points, not just left overs.


Top
 Profile  
 
PostPosted: Sat Mar 07, 2009 11:25 pm 
Offline

Joined: Wed Jan 28, 2009 4:51 pm
Posts: 31
I see the point system as only a semi-ranking system, with good players and high points being loosely correlated. The combination of elo and points makes a good cocktail I think.

Its my belief that if a winner gets 0ver 18 centres they still only get 18 centres worth of points. This discourages the behaviour you mention.

I agree with the active points thing, but thats a bit off topic.

Would it be easier if we just split hybrid games 33% Ppsc, 33% winner, 33% survival? This could just simplify things and make it more uniform.


Top
 Profile  
 
PostPosted: Sun Mar 08, 2009 1:15 am 
Offline

Joined: Wed Nov 12, 2008 8:48 pm
Posts: 29
The way I handled the PPSC/WTA question was to take the winner out of the equation.
We both agree that the winner's reward should not depend on his SCs.
We only need to set what percentage of the game value is his reward. (W=100% = WTA)
Then the question of PPSC vs Hybrid is then resolved by setting the percentage of the remaining points that should be rewarded for survival, regardless of SCs. (S=0% = PPSC).
This is precisely what I described in my system, so the problem is already solved. :P

The question of leftovers and rankings is off topic indeed, but relevant nonetheless if you are talking about half-measures.
I agree that the ranking is only a correlation, but the better this correlation the more this system works as an incentive.


Top
 Profile  
 
PostPosted: Sun Mar 08, 2009 9:59 am 
Offline

Joined: Tue Aug 26, 2008 8:46 pm
Posts: 249
There is one problem that I can see with this method:

A 15 centre survival with a 19 centre winner is better than a 15 centre survival with an 18 centre winner and 1 1centre survivor.


Top
 Profile  
 
PostPosted: Sun Mar 08, 2009 4:01 pm 
Offline

Joined: Wed Nov 12, 2008 8:48 pm
Posts: 29
TheGhostmaker wrote:
There is one problem that I can see with this method:

A 15 centre survival with a 19 centre winner is better than a 15 centre survival with an 18 centre winner and 1 1centre survivor.


What exactly is the problem?


Top
 Profile  
 
PostPosted: Sun Mar 08, 2009 5:31 pm 
Offline

Joined: Tue Aug 26, 2008 8:46 pm
Posts: 249
Basically, that it is not only your strength that matters, but also the number of other people who you face. point farming would lead to "elimination play" whereby strong seconders seek no longer to win, nor to maximise the size of their country even, but rather just to make sure that there are as few other countries as possible, by eliminating other small countries.


Top
 Profile  
 
PostPosted: Sun Mar 08, 2009 6:29 pm 
Offline

Joined: Wed Jan 28, 2009 4:51 pm
Posts: 31
There isn't anything wrong with finishing someone off if you can do it. It would never be over alot of points anyways.


Top
 Profile  
 
PostPosted: Mon Mar 09, 2009 9:56 pm 
Offline

Joined: Wed Nov 12, 2008 8:48 pm
Posts: 29
Just to point something out.
Point farming is impossible in my system since it has relative buy-in.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 20 posts ]  Go to page 1, 2  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