Forum
suggested change of distribution algorithm
|
Ryan wrote
at 1:24 PM, Friday December 29, 2006 EST
Currently dice are distributed randomly over your territories. The effect of this is that on average fuller territories get as many dice as empty territories. This makes the probability of fuller territories becoming full higher than less full territories.
The strategy that comes out of this effect is to wait with a full territory until its maxed and then move one by one most likely always getting filled up each turn. The reason this is bad is because if someone gets a large stack at the begining they have quite a bit more power than players with even stacks. This is a lot of power due to the random algorithm. THE SUGGESTED CHANGE: Instead of distributing dice one by one to a random territory number the algorithm should randomly distribute to an available dice spot. So if a territory only has one dice it has 7 available spots. If a territory has 7 dice it has one available spot. The less full territory is 7 times more likely to get a dice. When it gets one it is 6 times more likely...etc. The effect of this is that on average dice will even out across your territories. Its important to note that the distribution is still random. This change I believe would make a fairer game and I'm posting here to get some feedback first. |
|
|
Ryan wrote
at 4:10 PM, Friday December 29, 2006 EST I want to avoid scoring based on dice. The dice are the tools, the territories are the goal.
The best scoring I have so far is: attack wins - attack losses/2 |
|
|
Ryan wrote
at 4:17 PM, Friday December 29, 2006 EST Scenarios for the above scoring:
A player takes 10 territories and then gets conquered. His score is 10 - 10/2 = 5. A player loses 3 territories and sits on one for the rest of the game. His score is 0 - 4/2 = -2 A player gets all territories and during the process loses 10 attacjs. His score is 40 - 10/2 = 35. A player has a 50 round game going back and forth with 100 wins but 100 losses. His score is 100 - 100/2 = 50. The winners score is 100 - 70/2 = 65. |
|
TheGrid wrote
at 4:29 PM, Friday December 29, 2006 EST Sounds Incredible! Very Smart!
|
|
|
TheGrid wrote
at 4:34 PM, Friday December 29, 2006 EST Oh no, I see a problem. What if the last 2 players team up, and go back and forth forever to get their scores in the sky? Since 1000 - 1000/2 = is still 500 for the winner.
|
|
|
TheGrid wrote
at 4:35 PM, Friday December 29, 2006 EST Maybe these points should be used just for ordering the players positions, and then adjust scores as supplied before?
|
|
|
Ryan wrote
at 4:40 PM, Friday December 29, 2006 EST Yes, the score would just be used for ordering. The finish order would be compared to the ELO estimate and your rating would be adjusted.
|
|
JKD wrote
at 4:40 PM, Friday December 29, 2006 EST er, didn't think you guys would talk about the scoring here, I still like my score system suggestion in Ryan's original ELO topic better than what i see here... anyway, a simple way to deal with people in the corner is 8 dice max for 1,2 territory (or maybe less!), 9 dice max for 3, 16 for 4+. Those are bad example numbers but they look cool because of the square root stuffs. ;)
Combined with how hard it is to rebuild large stacks with the suggested distribution something like this could work I don't want people who insanely attack as much as possible to be rewarded. imo it's too complicated for the scoring system to look at strategies instead of just the results (like it currently does) @grid, people already ask about the distribution even though it's explained beside the game and is totally random. Making it into slots would cause frustration without a simple info link. Then scoring info and basic stuff could also be elaborated there; people only stopped asking about scoring so much because it's more neutralised so doesn't really matter. The finishing hand chart could be brought back to gpokr so they'd both have an info link. Sorry if this isn't constructive, all the best! |
|
|
TheGrid wrote
at 4:48 PM, Friday December 29, 2006 EST JKD, the scoring system looking only on "results" is bad for places 2-7. The result of the winner is clear. The "results" of place 2-7 are only random, often felt unfair, e.g. the corner sitter. If you get a 2nd place, it should be one that just didnt make the 1st, not one that sat in a corner all the time, or was considered away and therefore not a thread. etc.
Oh, and I don't like the idea much the max dice be determined by number of connected countries, since it a) very fast nominates a dominator in the game, with others haven few to nothing chances. b) makes the game complicated. I can already imagine the newbie sitting in the corner... "ah why dont I get more dices anymore, but you are... this is sooo unfair" Ryan: Used this scoring for ordering then ranking -> perfekt in my humble opinion. Maybe you should make a seperate thread about that? ;) |
|
redrock wrote
at 4:53 PM, Friday December 29, 2006 EST Ryan --
It would be interesting to get some players of various skill levels to try out a server where the attacker wins on draws...I think people turtle alot because there is a simple mathmatical advantage to letting the other guy go first and a pretty big chance that you are going to lose when attacking with even forces. This means that until you are 7 territory and full, a losing attack with a big stack will basically end your game. it would be interesting to see how the dynamics changed if you found those big stacks to have a target on them.. |
|
|
TheGrid wrote
at 4:56 PM, Friday December 29, 2006 EST redrock, the attacker has already an advantages, since if he looses, he looses only 7 dices, if the defender looses, he looses 8 dices.
|
