Greedy Goblin

Thursday, November 18, 2010

To leech or not to leech?

Commenters found a contradiction among my posts. On the one hand I inspired an addon to kick leeches from WG. On the other hand I suggested to leech honor as it has the highest honor/hour. Am I for, or against leeches?

I'm for maximum benefit. Let's see the standard prisoner's dilemma, where the players can choose to "cooperate" or "defect":
  • both cooperate: both of them get 3 points
  • one cooperate, one defects: cooperator gets nothing, defector gets 5 points
  • both defect: both of them get 1 point
Let's translate it to WoW PvP:
  • both playing well: win, lot of honor for effort
  • one is leeching: lose, medium honor for effort to the player, medium honor for no effort to the leech
  • both are leeching: graveyard camped, close-to-zero honor for no effort
The traditional prisoner's dilemma studies has simple solution: defect, as no matter what the opponent does, you get better results with defect than cooperation. However in the iterated prisoner's dilemma cooperation, is favored as the other player can punish defection (the optimal strategy is tit-for-tat, the player opens with cooperation, cooperate if the other cooperates, but defects after the other did).

So the answer to the question: you shall defect (leech), if you cannot be punished by other players. Please note, that assuming a symmetrical situation (which is more or less given in WoW), if you cannot be punished, you can't punish the other one either, so he will defect anyway. This is the situation in a random BG. The others can't punish you, as flagging you to AFK only forces you to make one hit into an enemy to remove the flag.

On the other hand if you can be punished, you shall cooperate, not only in fear of punishment, but also because it's a proof that other leeches will be punished too! In a guild premade leeching can lead to gkick. In a random pre-made, you can get to the ignore list of many and never be invited. Same with the WGClean addon.

Next step of the thought. If you are unpunished and leeching, you shall continue until there is anyone left to leech on. After you reached the "everyone defect" state, you shall actively seek punishing methods to enforce cooperation. Seeking methods before is counter-productive, as you are trading your 5 points of defect-cooperate for the 3 points of cooperate-cooperate. On the other hand if the defect-defect state is reached, you'll replace your 1 pt for 3, so go for it!

Final, but most important step: if you are cooperating for moral reasons, when others are defecting, you are harming the system, as you allow the defection exist. If you would defect too, the others would be more motivated to implement a punishment system against defection. So if others are defecting and you can't punish them you shall defect too to make the World better!

Pilgrim's bounty is coming, don't forget to sell spiced bread, simple flour, mild spices.


Squishalot said...

The long-term strategy in an iterated prisoner's dilemma scenario for asocials is for both prisoners to defect consistently, not tit-for-tat. Tit-for-tat only works when you bring in forgiveness. If you consider revenge and griefing, then it most definitely gets pushed to a 'both defect' scenario.

I disagree with your arguments on the motivations of leeches. The reason people leech is that they maximise their benefit-to-effort ratio.

"Next step of the thought. If you are unpunished and leeching, you shall continue until there is anyone left to leech on. After you reached the "everyone defect" state, you shall actively seek punishing methods to enforce cooperation. "

This sounds like the story of your life, as described by you a few posts ago.

Anyway, that all works, provided that the reward gained by actively seeking enforcement of cooperation is worth the 3 points you get from cooperation. In reality, it's not, and so, the effort-to-reward ratio is still very poor. Easier for them to to leech and farm than it is for them to drive improvement.

Note on the end point - if you cooperate for moral reasons, you are creating the system, not harming it. The system of defecting and subsequently driving cooperation occurs *because of* such people.

Gevlon said...

@Squishalot: the benefit-to-effort ratio is not optimal in the defect-defect situation. It's DC < DD < CC < CD.

No, the leeching is always started by the M&S who don't decide to leech, they are just merely useless.

Squishalot said...

@ Gevlon - no, I was talking specifically the leeching being caused by benefit-to-effort, not the defect-defect scenario in prisoner's dilemma. Prisoner's dilemma has no difference in effort between each choice, only outcome, so maximising outcome = maximising benefit ratio.

By comparison, in WoW, if I can leech honor by pressing a key every minute or so (honor from losing random BG), and I am able to use that time instead to do other things that I want (eg, posting messages on blogs / forums), then my benefit-effort ratio is much better than cooperating. I get half the honor, but spend significantly less than half the 'focused' time. This is the reason people leech, not because it's an optimal 'net gain' strategy - you yourself note that a leeching person with someone who is cooperating will earn less honor than if they both cooperate, unlike the prisoner's dilemma.

I disagree with your point about who starts it - the useless people who do not leech don't exist in the model. The leeches act only because there will be people to leech from (i.e. those cooperating).

Anonymous said...

What you describe is not a prisoner's dilemma. In fact, no matter what another player is doing in a battleground, you should *cooperate* as it always give you more honor.

1. If your partner defects and you defect, you get 0 honor.
2. If your partner defects and you cooperate, you get moderate amount of honor.

Conclusion: if your partner defects, you should cooperate.

1. If your partner cooperates and you defect, you get moderate amount of honor.
2. If you partner cooperates and you cooperate, you get a lot of honor.

Conclusion: if your partner cooperates, you should cooperate.

So the benefit (honor gained) is maximized by always cooperating. It also leads to maximum effort, though.

Gevlon said...

@Anonymous: it's not honor that matter, but honor/hour (or you can claim that standing in the middle of Icecrown and killing 3 random allies every day is a valid honor farming strategy).

The defector get more honor/effort than the cooperator. "Effort" usually measured in active time. A leecher can log in, press one button every 5 min while watching TV.

Riptor said...

I get the Prosiners Dilemma, but how dos this translate to WG? Is WG, if leeched, better Honor/Hour than a BG?

I'm not really into PvP so I wouldn't know, but even if leeching in a BG is the best Option, don't the Goals of Wintergrasp make leeching obsolete?

What I basicly want to know, does the Prisoners Dilemma still apply when the Goals go beyond farming Honor?

Andru said...

Gevlon, we're talking about absolute time here, not relative.

Let's build an hypothetical scenario.

Say it takes 24 hours of leeching for 4k honor. Say that in these 24 hours, you're effectively pressing buttons 5 minutes. Which brings your honor/hour farm according to your method at 33k honor/hour.

Fair enough.

Only that 4 k honor can be earned in 3 hours of active playing. Your method places that at 1.3k honor/hour, or a lot less effective.


The second method allows the person to gear up a lot faster, despite absymal perceived h/hr. So, next time when person 1 will try to get into arenas (where leeching doesn't work), they will get roflstomped by person 2 with vastly superior arena gear

Despite #1 being an optimal honor earning strategy, it's an actually suboptimal gearing strategy.

Gevlon said...

@Andru: what you miss is that everyone in the BG get nearly the same honor, the guy who plays well alike the leecher.

Ðesolate said...

I have another point of not leeching. Usually when I´m in BG I´m there with my arenamates or guildmates. When the BG is running fine and everyone cooperates it´s usually a opportunity to train for rated bgs wihtout voice-tool.

If the BG is running on leechers / no-brains (AB 1 base in our hand and everyone is running around at stables (I´m horde with main)) then we seek x vs x situations to train for arena / open pvp. At that point you can do that the best way by attacking a base / a tower or something in general "cooperating".

If I´m alone and leechers / no-brains got their upper hand I usually try to do some missing achievements or will be more attracted by TV or maybe some university related simulations (deflecting).

In general I have more profit from cooperation because it generally increases my performance and my knowlege of certain x vs x situations (no matter of class or numbers). By leeching you usually never get really interesting input.

When I´m in a BG where the enemys are extraordinally bad I usually get bored and start to defect. Why? Because I get nothing in extra out of a fight without challenge.

Maybe I´m an odd kind but I value experience very high.

ardoRic said...

I'm not sure I'm convinced by your WG-prisoner analogy for two reasons.

Firstly, iterated prisoner's dilemma assumes you have the same opponent, and that the opponent (and you) have memory. In WG your opponents will not remember you, and you probably won't remember your opponents (most of the time). So WG resembles single prisoner's dilemma more than iterated.

Secondly, tit-for-tat is the optimal strategy when you don't know your opponent's strategy. In this case, you claim you know your opponent's (the M&S) strategy: they'll always defect.

Bobbins said...

'Pilgrim's bounty is coming, don't forget to sell spiced bread, simple flour, mild spices' with level 1 cooking

+ recipes + tracker food + raw food for beyond 350 cooking

But are people going to be distracted by cata events?

Bobbins said...

Forgot to mention does leeching relate to the cata event as I think you do not have to participate in the clearing of the cities to get the access to the bosses if the elementals are cleared?
This would be the best current example of the 'to leech or not to leech'

Ðesolate said...

Addition: I don´t think the PD system can be compared with BG. Of course your Honor / efford is higher by succesfull leeching. How actually scales minimum efford (sitting at the pc watching that you don´t get afk.flagged and watching TV)?

That depends on the group and on you.

- Using a Bot would be the lowest scale = doing nothing at all.
- Avoiding pure game-mechanic AFK-Flag = hopping / walking every 10 minutes.
- Avoiding reported Pig-Debuff = Run into the enemys every 5-10 minutes to get rid of the zero-honor debuff.

I consider that botting is of no debate here. If you consider that I´d you´ll better be gone.
The second method would be the one with most benefit.
The thjird one would be that innefective that you could better zerg or cooperate.

Method 1 gets you 20
Method 2 gets you the named 5
Method 3 gets you 3-0

I´ll also mention that when somebody cooperates he hardly gets 0. You would scale it as an 0 til 1 depending on how succesfull the cooperating are.

Of course loosing by doing maximum efford is near to zero because you can´t get less...
...declaring it as an PoV zero point because you are at the potentual least profitable situation is valid, but its near to method three.

Camo said...

@Riptor: The point is that you can no longer be punished (kicked from the raid) in WG.
The new system removed tenancy and introduced a 1:1 ratio with a 20 minimum cap (?). I'm not sure if it's easier to win if the other faction has more players and a higher chance to fill their slots with M&S. I don't know if you can choke the enemy if you show up with a small team and then pull the rest of the players once you advanced and speculate that the other faction won't sit there and try to queue for 5 mins leaving them with only the starting players and a few late-comers.

Squishalot said...

I think Andru is assuming that the presence of the leecher will mean that everyone gets a reduced amount of honor (a loss). So we're only looking at the marginal leech, the one that would tip a side from victory to defeat.

Gevlon said...

@Ardoric: WGClean added exactly that, memory to the system.

Ðesolate said...

Again addition..:
I think the basic of PD is missing some extra variables. Such as the number of player and the fact that you can decide anytime durige the game (BG). So you have in a BG infinite possibilities to decide.

At the numbers:
you have 15 per BG.
4 decide to deflect 11 decide to cooperate:
deflecters earn marginally "10" (win without efford)
cooperators get marginally "3" (win with efford)

now we have a opposite 11 deflect 4 cooperate
deflectors earn maginally "5" (loose without efford)
cooperators get "0" (loose with high efford)

now just imagine that you get the time variable into this situation with different strategys that change the playstyle every second potentally.
So you can do the "Tit for tat"-tactic deciding every second. And yes this will affect each other. You get an interesting complex system with the following playertype and their strategy:

M&S: Generelly the defecting one (may be a started grim trigger).
"Social": Generelly the cooperating one (mostly trying a Collusion but often failing).
"Selfish" or "Asocial": Tit for tat. If the others seem to deflect, they will also. No efford for minimum profit.
Dislogic: Random. They wage profit in different ways than the first three. It is also possible that their kind of efford is defecting. I´m adding this to get the bit of chaos. (I´d consider myself as such as long as I seem to take profit from personal experience)

You could programm a simulation witch includes all this, maybe the outcome is interesting but I think such a program should already be excisting. But you can already forecast the most likely outcome of every situation.

Oh and yes the opposite side in the BG also affects your side and the whole "progress".
Happy thinking.
(sorry took some time to focus on the given simulation)

Ulsaki said...

"I'm not sure if it's easier to win if the other faction has more players and a higher chance to fill their slots with M&S."

I would say yes, it is. On my server, WG has gone overnight from being dominated by the Alliance 90% of the time to the Horde winning constantly.

The most recent WG I participated in was a complete failure. There were morons standing on and watching as the siege were attacked and destroyed, and there were idiot kids making catapult after catapult, and completely ignoring the guns.

My theory is that a few competent Alliance players did most of the work. But now it's almost impossible for them to get in, and Alliance side is filled almost exclusively with retards, causing the Horde win ratio to approach almost 100%.

Wilson said...

There's a flaw in using Prisoner's Dilemma as a model for BGs. In PD, points are inherently valuable in and of themselves. But in WoW, Honor Points are only valuable if you use them to purchase gear which improves your chances if you are actively trying to win. If you're leeching, then you might as well be naked - your accumulated points have no value. If you always leech, then you win nothing of value, and you would have been better off not playing at all.

Gevlon said...

@Wilson: there is some truth in your point, however gear can be used elsewhere:
- moron boosting his gearscore
- moron believing that he needs gear to "pwn", and plans to start playing seriously after he is geared
- arena player who consider BG boring but needs honor for buying start arena gear

Anonymous said...

How long would it take a leech to gain all of the honor they would need? You would assume they have a goal as honor has a cap and at some point leeching would be pointless. Once the goal is reached gear or honor cap does the leech start to contribute or vanish from the BG?

Anonymous said...

It seems like the readers on this blog have gotten dumber recently, or at least the ones who support Gevlon have stopped commenting. Gevlon seems clearly right here... but I guess like everyone else I can't be bothered to type out the paragraphs needed to argue with the people commenting.

If the goal is to argue with those people -- Gevlon is cooperating, so I defect.

Martin said...

"I would say yes, it is. On my server, WG has gone overnight from being dominated by the Alliance 90% of the time to the Horde winning constantly."

I have exactly the same experience. I don't have explanation for this. I assume that it is the same for Agamaggan.

Read the above, you might be doing mistake of seeing correlation in WGClean and patch 4.0.1

Gevlon said...

@Martin: the huge change is made by 4.0.3, not 4.0.1, analysis of it on Monday.

Dàchéng said...

Now that Blizzard has put an end to your scheme for world - I mean, Wintergrasp - domination, are you going back to leeching?

In fact, virtually nobody is in Wintergrasp to maximise their Honor per hour. They are there mostly for two or three reasons:

1. To collect enough Honor to get PvP gear. Maximising Honor (to a particular cap) is their goal, not Honor/hour.
2. They enjoy the battle, and are maximising their happiness/hour by participating in it.
3. They are farming (herbs, nodes, fish) and just make use of the Battlemaster's teleport to get them there quickly.

I suspect that you are not there for reasons 1 and 3, and therefore will continue to visit Wintergrasp for reason 2.

Anonymous said...

@Wilson & Gevlon:

Also, the goal behind Prisoner's Dilemma is to amass more points than your opponent. That's why tit-4-tat is an optimal strategy and "always defect" is not. If you always defect, you will always lose.

In WoW PVP, the goal of an honor point leech is to simply gain honor for little effort invested. It does not matter to the leech that the opposition may very well gain MORE honor for less work. If you always defect, you will always gain at least 1pt, guaranteed, and it does not matter what the other guy/team/faction gained.

Anonymous said...

I assume the leech does not care about how much honor is gained as long as honor is gained. If a leech started to trend 0 honor gains you may see a change. Otherwise even 1 honor for doing nothing is a gain. So if out of 10 players 9 were leeching and one was active a single kill by that player would satisfy a leech.

My thoughts on it at least.

Anonymous said...

I see a lot of arguments here are based on calculations of hypothetical (ie, made up on the spot) honor amounts. Has anybody bothered to, say, do ten BGs using each method with careful measurement and controls?

Here's what I'm imagining: two players on the same faction and battlegroup run the same BG (but different instances of it) at the same time of day. One leeches and one plays hard to win. Each records the amount of real time, 'active' time and honor gained in the course of doing 20 BGs. Increment across all of the different BGs. Publish results.

I think this would lay to rest a lot of the arguments here. Depending on an individual's relative value for actual time and 'active' time, it would be possible to consult the table and find the best source of honor.

Here's my guess at the outcome: cooperate in AV comes out ahead for actual time, leeching in EotS or WSG wins for 'active' time.

Maarten said...

Tit for tat was shown to be superior only in a simulation together with other strategies. This does not mean that it is always the optimal solution. It is quite likely that there are better strategies (wikipedia mentions "tit for tat with foregiveness") depending on circumstances/nature of other strategies present.

Crombach said...

As pionted out PD is not the best choice to apply to the leeching problem. Whith honor in a BG we look at a 'common good'. I think there is an application for this in an extended PD version which includes groups and is not restricted to two players. Question than is not 'Do I defect or cooperate with the other player' but 'Is my cooperation necessary for the production of the common good'. Even punishment goes to the next level. It may be not necessary to punish but to reward the one who does the punishing. That would be applicable to your behaviour in WG (and WGclean). If like you can research this kind of PD to get a better theoretical grasp on leeching, I guess. The keyword(s) would be social dilemma (e.g. Hechter (1987):Principles of group solidarity)

Anonymous said...

I believe that the most important question here is:

Why on God's Earth would someone queue for a battleground in order to leech it so that he can buy pvp gear?

Yes, I've read the comment before, and asked many people but I still don't understand.

F.e. I dislike raiding. There's nothing on this world that would force me into ICC. What on earth would I need pve gear then for?

So, why do people do things they dislike? (Leeching is a byproduct of that).

Let's say the gear part is nullified. (blizz sends all newly dinged epic set in mail). Who would then play battlegrounds? Is the only reason people play anything in Wow a shiny reward at the end?

Just some random thoughts...

Soge said...

@19:07 Anonymous

You can use Honor to buy other things than PvP gear, such as money. BG AFKing a quite profittable Gold/hour activity. Launch a BG and go watch a Movie, play some other game, read a book, clean your house, work, or whatever. Check back every minute or so to see if you were reported for being AFK, and if then just go hit someone or die.

Vesoom said...

"Here's what I'm imagining: two players on the same faction and battle group run the same BG (but different instances of it) at the same time of day. One leeches and one plays hard to win. Each records the amount of real time, 'active' time and honor gained in the course of doing 20 BGs."

I would love to see the results of this test and definitely encourage anyone that wants to do it. It's always nice to have actual numbers to discuss and argue over. This won't solve the PD question, but may shed some light on how closely PD is reflected in WoW BGs.

I would also be interested in seeing a discussion of the methodology before anyone attempts the test. Perhaps the Goblin Society for Scientific Research can come to a consensus on the methodology for this test.

Anonymous said...

To Soge:

There is an interesting thread on forums, in which I wish for "no honor for loss" battlegrounds.

Then the gold/gear/what not part would be non-existent and people would play them only in order to try to best the other team. (And, when win, to buy gear).

Not so much different than raids today. If you don't kill a boss, you get nothing (yes, there is random stuff, some low gold, a ring if you get exalted on trash; however, if you constantly wipe on Marrowgar you get nothing!)
So, why is that allowed in battlegrounds?

Wipe on Marrowgar= 0 points/emblems, 0 t10 gear
Fail in Battleground = enough honor to buy wrathfull weapon in 1 day (done it on alt).

So, the system is flawed?

Anonymous said...

Working hard/cooperating always gets you more honor and more honor/clock-time than leeching. So unless you are a social offended by justice (leechers getting the same as you), then you work hard regardless of whether there are 1% or 100% people cooperating.

Pressing one key every few minutes is going to get you more honor/keystroke than any cooperation.

So I do not see the "dilemma" Unless you are swayed by your emotions, if you are trying to maximize your honor per hour of the clock, then you always cooperate. If you are trying to maximize your honor per effort/keystroke you always leech.

Yes, I assume most pvp leeches, at least the ones not driven by the sociopathic appeal of leeching, are doing it for the gems/gold as opposed to pvp gear. Although the 264 ring, neck, cloak, bracers were better than anything before ToC; certainly helped start the heroic grindfest. Counting WG marks, commendations, and HK, the weekly WG was nearly two cardinal rubies.

Anonymous said...

"Working hard/cooperating always gets you more honor and more honor/clock-time than leeching. "

It may all be true, but let's all ask one simple question: Why do we play this game??

I work because no work means no pay means no food. And I kinda need to eat.

I join a bg in a game that is supposed to be my leisure time and all I see are mid field fighters. They can give me millions in gold and honor, but I'm not staying, participating or contributing to that battleground.

15min debuff is better spent playing with my dog.

I don't understand why we even have such debates. Imagine what would happen if on Lich King 9/10 players went afk, alt tab, autoshoot, run in to die...

I do not leech. I do not contribute. I either fight with players or leave.

Anonymous said...

Anonymous (21:35) says:

"Working hard/cooperating always gets you more honor and more honor/clock-time than leeching."

The haughty tone and lack of evidence in this post does not impress me. Try again with real numbers.

Kuckuck said...

@annon 21:35

How long does each keystroke take on average?

Deepcut said...

Here's a good example:

Boxington said...

There's a lot of confusion about Game Theory here.
This is not an iterated prisoner's dilemma, as another commenter posted. Such a game would require playing with the same player each time. In reality, you are playing against a N different players who follow some distribution of strategies. The key point here is that your strategy only has a 1/N effect on the distribution of strategies. This is studied in the field of Equilibrium Dynamics in game theory, and finds results VERY different from 2-player games.

For instance, depending on the parameters, there may be stable equilibria at Coop-Coop AND Defect-Defect. However, whichever equilibrium happens first then becomes "sticky" insofar as the least rational thing to do is to play the opposite of everyone else (i.e. exactly what Gevlon said is optimal). Obviously this depends on the honor/hour/activity ratio, but that hasn't yet been established here.

And yes, it is important to note that Gevlon is not correct when he claims that tit-for-tat is the optimal strategy. A.) He's talking about the wrong game type, B.) the findings were that tit-for-tat was the best performing strategy over other strategies in practice (not in theory as the word optimal suggests), and C.) Tit-for-tat's basis for success is premised on having effective punishment, which is definitely not the case when you are only 1 person on a thousand+ person server.