The Gini-Coefficient: Examining Fairness in PvP

Forum for discussing the Torchlight MMO Game

Moderators: Runic Games, Perfect World

The Gini-Coefficient: Examining Fairness in PvP

Postby Jerich » Mon Oct 05, 2009 7:41 pm

The Gini Coefficient Defined.
Tobold recently posted a fascinating article on his MMO blog here: http://tobolds.blogspot.com/2009/10/gini-coefficient-of-mmorpgs.html In it he talks about an economic concept known as the Gini coefficient. Basically, the Gini-Coefficient keeps track of the equality of wealth in a country. A Gini Coefficient of 1 would mean one person had all the money and power. A Gini Coefficient of 0 means that power and money are shared equally by all people in society. For example... a poor country may have a Gini coefficient of .6 wheras Europe has a relatively low Gini Coefficient of .31. The US is somewhere in between with a Gini Coefficient of .47. Any time a country votes in something that may modify their Gini Coefficient (i.e. Obama's health care plan), it causes a huge outcry.

The Gini Coefficient in MMOs
Tobald then goes on to talk about how the Gini-Coefficient can be applied to MMOs. This is difficult, because gear plays a huge element in calculating wealth. A raider with 0 cash may have more real wealth than someone who is at the money cap with only gold-bought items. Anyway, he goes on to say that WoW used to have a high Gini-Coefficient (The difference between the haves and have nots was huge), but it has been decreasing over time as they add more ways to get gear (i.e. welfare epics as raiders call them). Everquest had a much higher Gini constant than WoW in my mind. A game like Eve has a much higher gini index than even Everquest.

The Gini Coefficient in Torchlight
I think most of the community want the Torchlight MMO to have a lower Gini Coefficient than WoW. The main reason is that games like this appeal to the casual player. I think this is one reason that people are divided on micro-transactions. We do not want them to mess with the Gini-Coefficient too much and thus view them sceptically.

The Gini Coefficient Applied to MMO PvP
Interestingly enough, most pvp games strive to have as low a Gini-Constant as possible to promote fairness. A game like Starcraft for instance is striving to get the Gini-Coefficient as close as possible to 0. An FPS like Halo would also try for this. This got me thinking about PvP in MMOs. Everquest had an incredibly high gini coefficient. So did early World of Warcraft. People in top tier raiding guilds would blow away the competition even if the competition was unskilled. While World of Warcraft has added some ways for people to ramp their gear up, gear is still the primary decider of who will win a match versus who will not. The gini index in world pvp is even more unfair. Games like Guild Wars, on the other hand, strive to have a much lower gini index. Everyone has roughly the same quality of pvp gear and is the same level. I think this is why most people who experienced both will say that the Guild Wars PvP system was better than the World of Warcraft system.

Ideas to create a fair PvP Environment
I believe that most avid PvP fans want a level playing field. Only gankers and griefers typically want a huge advantage over their opposition. I believe that Torchlight should strive for fairness in overland pvp. It will make for a more robust and interesting PvP system. There might be several ways to do this, but I think one simple one is to utilize a system known as mentoring which is typically used to allow low level and high level people to play together. Here is my idea:
  • Use the mentoring system to make everyone in the overland the same level in regards to pvp, perhaps level 20. (They can still be different levels in pve)
  • All skills should be normalized to follow the level 20 damage curves.
  • Gear should all be normalized to have a level 20 item budget. This means a level 80 unique should give people the same advantage as a level 20 magic item. White items, should be given magic properties that the mods decide on.
  • People who are being teamed up on should get a slight bonus to their level to compensate.

I think using these methods would make for an extremely interesting pvp Shadowlands. Basically everyone would be a threat and there would be no Godlike people running around. I believe that any PvP system which makes skill a much bigger component to success than levels is innately more interesting and competitive.

What do you guys think? Would such a system work out for pvp or can you see some problem with it?

- Jerich
User avatar
Jerich
 
Posts: 731
Joined: Sun Jun 07, 2009 8:29 pm
Location: Puyallup Washington

Re: The Gini-Coefficient: Examining Fairness in PvP

Postby Keft » Mon Oct 05, 2009 8:48 pm

I like the idea of limiting level difference to avoid greifing and improve fairness in pvp, however I think it might be better to implement it either only in about 50% of the areas with some incentive to go in the non "fair" areas so that there will still be high level villains (greifers) and heros (players who protect lowbies or get revenge for them). I have made many friends in other mmo either as a protector or protectee.

Another way it could be implemented that would allow all the area to be the same would be to use a sliding scale of pvp buffering. Each time you kill someone for pvp effectiveness you are lowered by 1/3 the difference between your level and theirs for maybe 10-20min. So say you are level 30 and you kill a level 10 for the next 10-20 minutes you are effectively level 20 for pvp. Kill them again and you drop the difference of your real level and their level in half again so now your level 10 effective pvp level. If a level 30 comes and gets revenge on you for them he will just stomp you since you are level 10 but then he will drop 1/2 the difference between your real level and his level 0/2 so he stays level 30. This way there can still be heroes and greifers but greifers are limited in how effective they can be long term and heroes are not.
Keft
 
Posts: 2
Joined: Mon Oct 05, 2009 8:36 pm

Re: The Gini-Coefficient: Examining Fairness in PvP

Postby Omnifas » Mon Oct 05, 2009 9:39 pm

Although a more fair pvp environment is pleasing, I still have a firm belief that strategy should be promoted over fairness, so:

People who are being teamed up on should get a slight bonus to their level to compensate.

Breaks my code because I believe no one should be protected from anything if they are already on even ground. If they are not on even ground, its a big maybe.

To be Honest, I think any Open PvP needs it fair share of gankers, and griefers. Crime Systems are just fun...Bounty Hunting, Knight vs Pirate(Bandit), good vs. bad. Its not a crime if everyone is equal. You really just need to make griefers(murderers), and gankers(psychopath) feel like they are criminals; no way to fix their "badness"(Negative Karma) except through PvP with a "Good"(Hero) player, They are open PvP in safe areas(no pvp areas). Basically they are barred from functions that are only available in towns like Auctions. Making the choice of being a Villain actually a logical choice, no "I wanna be a bad guy"s here, everyone that has negative Karma has it because they chose to live with the punishment, not because they felt like griefing or ganking.
Image
User avatar
Omnifas
 
Posts: 986
Joined: Fri Jun 05, 2009 7:43 pm
Location: California

Re: The Gini-Coefficient: Examining Fairness in PvP

Postby BN1 » Tue Oct 06, 2009 12:10 am

I've always been a proponent of how pvp was handled in GW. The gini coefficient as you describe it in Guild Wars is maintained by what is largely a seperation between PvP and PvE. If a person is interested primarily in PvP they can create a max level character with access to nearly all the skills and plenty of equipment options that are comparable to end game equipment. The reward to someone who is interested in playing the PvE in hopes of being a great PvPer is relatively minor. A player like that will get to unlock a small number of skills they wouldn't have access to if they simply rolled a PvP character, and will have access at end game to some items that aren't any better but are itemized differently. While the feasibility of implementing such a system with out it being one of the primary design goals of the game are questionable I think its a strong example of balancing fairness. Not that in the PvP environments there is no penalty for "ganging up" on someone, this is in effect one of the primary strategic decisions in the game. Which person on each side should the other team aim to most quickly eliminate, off set by a need to identify and protect which ever of your team mates is in the most danger.

I think PvP as implemented really really needs to be entirely concentual. Either through allowing servers with no world PvP, something done in WoW. Or through the more common practice of keeping it in isolated zones like battlegrounds/arenas or through a duel system which requires both people to agree (CoH, CO, Guild Wars, Diablo 2, and the other half of PvP in WoW that is available even on normal servers).

I think scaling all PvP is a fairly good idea, however I might suggest it instead scales to a bracket based on the arena the players are playing in. For instance in early zones the pvp offered via BG/Arenas would scale item stats and skills of high level players down while in higher levels the available PvP zones would primarily scale the stats and skills of lower level players up. The technology behind a fair scaling system either way could be problematic. It would be easier to scale player stats and skills downwards (just remove the skill points and stats in the order they were taken on level up) while items it would arguably be easier to scale upwards. Afterall if my sword with three abilities is going to lose two in a pvp match how do i know which two until I enter the match. I also think a system, that allows matches scaled in both directions is a good thing because it allows players to pvp at relatively equal footing while still having the option to experience pvping at different stages of development. Low level scaled matches would have fewer skill/power options while high level pvp would have more, such a system would allow players to play whichever they prefer.
User avatar
BN1
 
Posts: 45
Joined: Thu Sep 24, 2009 11:14 am

Re: The Gini-Coefficient: Examining Fairness in PvP

Postby Jerich » Tue Oct 06, 2009 7:46 pm

Keft wrote:I like the idea of limiting level difference to avoid greifing and improve fairness in pvp, however I think it might be better to implement it either only in about 50% of the areas with some incentive to go in the non "fair" areas so that there will still be high level villains (greifers) and heros (players who protect lowbies or get revenge for them). I have made many friends in other mmo either as a protector or protectee.

Another way it could be implemented that would allow all the area to be the same would be to use a sliding scale of pvp buffering. Each time you kill someone for pvp effectiveness you are lowered by 1/3 the difference between your level and theirs for maybe 10-20min. So say you are level 30 and you kill a level 10 for the next 10-20 minutes you are effectively level 20 for pvp. Kill them again and you drop the difference of your real level and their level in half again so now your level 10 effective pvp level. If a level 30 comes and gets revenge on you for them he will just stomp you since you are level 10 but then he will drop 1/2 the difference between your real level and his level 0/2 so he stays level 30. This way there can still be heroes and greifers but greifers are limited in how effective they can be long term and heroes are not.


I think this system is better than the one I first proposed. I agree with the need for villains and heroes in a world pvp. It does make the game more interesting, and with a game where you can solo instances, there is probably no real reason to make everything completely fair. I like the idea of the karma system you proposed. I think it would make the game more interesting for both heroes and griefers.

Omnifas wrote:Although a more fair pvp environment is pleasing, I still have a firm belief that strategy should be promoted over fairness, so:
Jerich wrote:People who are being teamed up on should get a slight bonus to their level to compensate.

Breaks my code because I believe no one should be protected from anything if they are already on even ground. If they are not on even ground, its a big maybe.


I think this is more needed if there is faction based pvp than if there is only FFA pvp. If one faction severely outnumbers the other, pvp is not as fun for anyone. Wow did this with tenacity in Winter grasp and I think that was an improvement to the system.

Omnifas wrote:To be Honest, I think any Open PvP needs it fair share of gankers, and griefers. Crime Systems are just fun...Bounty Hunting, Knight vs Pirate(Bandit), good vs. bad. Its not a crime if everyone is equal. You really just need to make griefers(murderers), and gankers(psychopath) feel like they are criminals; no way to fix their "badness"(Negative Karma) except through PvP with a "Good"(Hero) player, They are open PvP in safe areas(no pvp areas). Basically they are barred from functions that are only available in towns like Auctions. Making the choice of being a Villain actually a logical choice, no "I wanna be a bad guy"s here, everyone that has negative Karma has it because they chose to live with the punishment, not because they felt like griefing or ganking.

Your system of Karma is interesting. Do you think your system could be long term and Keft's system could be short term? I think that would make for an interesting PvP System.

BN1 wrote: think PvP as implemented really really needs to be entirely concentual. Either through allowing servers with no world PvP, something done in WoW. Or through the more common practice of keeping it in isolated zones like battlegrounds/arenas or through a duel system which requires both people to agree (CoH, CO, Guild Wars, Diablo 2, and the other half of PvP in WoW that is available even on normal servers).

I think scaling all PvP is a fairly good idea, however I might suggest it instead scales to a bracket based on the arena the players are playing in. For instance in early zones the pvp offered via BG/Arenas would scale item stats and skills of high level players down while in higher levels the available PvP zones would primarily scale the stats and skills of lower level players up. The technology behind a fair scaling system either way could be problematic. It would be easier to scale player stats and skills downwards (just remove the skill points and stats in the order they were taken on level up) while items it would arguably be easier to scale upwards. Afterall if my sword with three abilities is going to lose two in a pvp match how do i know which two until I enter the match. I also think a system, that allows matches scaled in both directions is a good thing because it allows players to pvp at relatively equal footing while still having the option to experience pvping at different stages of development. Low level scaled matches would have fewer skill/power options while high level pvp would have more, such a system would allow players to play whichever they prefer.

I think the Torchlight PvP will probably be consensual. You will either roll a pvp or non-pvp character. Non-PvP characters will still be able to take part in pvp, however. I like your idea of variable scaling. I think this would be especially important in the E-sport type arenas. As the other people posted above, there may be some reasons to allow a higher Gini-Coefficient in the over world, but anything that approaches a battleground etc is much better if the playing field is even. Here is how I envision item scaling working....
  • Each type of effect has a certain item value cost. Thus, by adding up all the effects of an item its total item budget can be calculated.
  • Every level would be set to have a set item budget.
  • When someone was scaled down or up in levels, all their items would be scaled accordingly until their items were at that budget.
  • Note this would also effect people who had epic gear that had a higher item budget than a normal item for their level. It would be scaled down to the proper value.
  • Skills and stats would also be modified so that they matched the level skill / stat damage.
  • You would not lose skill points... instead your skills would be modified to do the appropriate damage for that level.
I think this method, if implemented correctly would create a low Gini Coefficient without destroying player uniqueness. I also believe that it might be a good idea to make a certain skill act one way in pvp and another in pve. That way pve players don't feel punished by pvp balancing and vice versa. I am not sure if this creates too much of a developer headache, however.

- Jerich
User avatar
Jerich
 
Posts: 731
Joined: Sun Jun 07, 2009 8:29 pm
Location: Puyallup Washington

Re: The Gini-Coefficient: Examining Fairness in PvP

Postby BN1 » Wed Oct 07, 2009 5:11 am

Jerich wrote:•You would not lose skill points... instead your skills would be modified to do the appropriate damage for that level.

I understand what your getting at but wouldn't this always give an advantage to higher level people. Consider the following:

Lvl 1 player has a skill that enters a cooldown state of 10 secs upon use and does ranged damage
a Lvl 5 player has this skill as well as a second skill with a cooldown state of 15 secs that also does ranged damage (likely different type or different scale)

Even if both players are rebalanced so their skills are doing the same amount the lvl 5 player has an advantage as they have an additional skill they can use while the other is coming off cooldown. This would effectively increase their damage per second.

Similarly if a lvl 1 player has a skill that allows them to remove a damage over time effect on themselves and temporarily raise their resistance to toxic and a lvl 5 player would have this skill and another skill they can use to raise their resistance to fire wouldn't a lvl 5 player even assuming the skills are all scaled similarly have a signifigant advantage because they could adapt to more situations because they have more options.

Similarly on item budgetting:
Jerich wrote:•Each type of effect has a certain item value cost. Thus, by adding up all the effects of an item its total item budget can be calculated.
•Every level would be set to have a set item budget.
•When someone was scaled down or up in levels, all their items would be scaled accordingly until their items were at that budget.

A Lvl 1 player has a short sword of fire, normal damage + minor fire damage
a lvl 5 player has a short sword of fiery strikes, normal damage + minor fire damage and + small amount of deterity

If these are scalled to a similar budget you have three outcomes:
1. The low Lvl player's sword would do more fire damage than a sword by a higher level player because it only has one stat, as apposed to two
2. The lvl 5 player's sword would lose one or the the other ability and possibily have the remaining ability buffed slightly (making a transparent system so players can tell which abilities their weapons would or would not have in this case would be difficult)
3. The scaling would still result in the higher level player having a better weapon because even when the abilities are scaled similarly the higher level player has more abilities on their weapon

Yes to some extent this can be avoided (see the badge purchassed Heirloom items, in world of Warcraft) but frankly I'm not sure how to extrapolate such a system to be used with a system that generates random assemblages of loot stats (heirloom items works by having preset stats for every level in the range, between 1 and max lv, no random component to the statsl) and maintain a system where players can predict what the stats of their items will be.
User avatar
BN1
 
Posts: 45
Joined: Thu Sep 24, 2009 11:14 am

Re: The Gini-Coefficient: Examining Fairness in PvP

Postby Jerich » Wed Oct 07, 2009 3:25 pm

My idea presupposes a mentoring scaling already in the game. PvP scaling would piggy back the existing system. If there were no mentoring system, I do not think it would be worth adding one simply for pvp. That being said, I don't think the scaling issues are a huge problem The goal is to get the Gini-index closer to 0. Not set it to zero. I don't think setting it to zero is possible with a deep item/ skill point leveled game like Torchlight.

BN1 wrote:
Jerich wrote:•You would not lose skill points... instead your skills would be modified to do the appropriate damage for that level.

I understand what your getting at but wouldn't this always give an advantage to higher level people. Consider the following:

Lvl 1 player has a skill that enters a cooldown state of 10 secs upon use and does ranged damage
a Lvl 5 player has this skill as well as a second skill with a cooldown state of 15 secs that also does ranged damage (likely different type or different scale)

Even if both players are rebalanced so their skills are doing the same amount the lvl 5 player has an advantage as they have an additional skill they can use while the other is coming off cooldown. This would effectively increase their damage per second.

Similarly if a lvl 1 player has a skill that allows them to remove a damage over time effect on themselves and temporarily raise their resistance to toxic and a lvl 5 player would have this skill and another skill they can use to raise their resistance to fire wouldn't a lvl 5 player even assuming the skills are all scaled similarly have a signifigant advantage because they could adapt to more situations because they have more options.

I don't it is possible to make the two players completely and still allow diversity and level gain. I think it is ok if the higher level player retains a slight advantage. The goal is to remove the crushing advantage so that a skilled lower level player can still beat a mediocre higher level.

BN1 wrote:Similarly on item budgetting:
Jerich wrote:•Each type of effect has a certain item value cost. Thus, by adding up all the effects of an item its total item budget can be calculated.
•Every level would be set to have a set item budget.
•When someone was scaled down or up in levels, all their items would be scaled accordingly until their items were at that budget.

A Lvl 1 player has a short sword of fire, normal damage + minor fire damage
a lvl 5 player has a short sword of fiery strikes, normal damage + minor fire damage and + small amount of deterity

If these are scalled to a similar budget you have three outcomes:
1. The low Lvl player's sword would do more fire damage than a sword by a higher level player because it only has one stat, as apposed to two
2. The lvl 5 player's sword would lose one or the the other ability and possibily have the remaining ability buffed slightly (making a transparent system so players can tell which abilities their weapons would or would not have in this case would be difficult)
3. The scaling would still result in the higher level player having a better weapon because even when the abilities are scaled similarly the higher level player has more abilities on their weapon

Yes to some extent this can be avoided (see the badge purchassed Heirloom items, in world of Warcraft) but frankly I'm not sure how to extrapolate such a system to be used with a system that generates random assemblages of loot stats (heirloom items works by having preset stats for every level in the range, between 1 and max lv, no random component to the statsl) and maintain a system where players can predict what the stats of their items will be.


The same applies to this case. It is ok if the higher level player has a slight advantage as long as it is not a crushing advantage. I believe the converse would cause players to revolt, however.

- Jerich
User avatar
Jerich
 
Posts: 731
Joined: Sun Jun 07, 2009 8:29 pm
Location: Puyallup Washington


Return to Torchlight MMO Discussion

Who is online

Users browsing this forum: Google [Bot] and 3 guests