 |
|
Oct 13, 2009, 03:04 PM // 15:04
|
#241 | |
|
Forge Runner
Join Date: Apr 2008
Location: Texas
Guild: Reign of Judgment [RoJ]
Profession: Me/
|
Quote:
For the third time, they're removing HB. It doesn't matter. We'll get one more RR day MAYBE. It's not going to asplode GW for there to be 1 more RR day. Let it go, people. |
|
|
|
Oct 13, 2009, 05:14 PM // 17:14
|
#242 | ||||||
|
Older Than God (1)
Join Date: Aug 2006
Guild: Clan Dethryche [dth]
|
Quote:
Behavior in the Hall of Heroes shows the impossibility of your approach. A similar tit-for-tat problem exists because the quantity of elite players is small. People gank for friends on long win streaks with the expectation of reciprocity. There's no way that you can prevent people from knowing who the holding team is, or the length of their win streak. You solve these problems by altering the payoffs. Strip the zkeys so that only players interested in playing HB for its own sake (scarce as they are) do so. Punish people for failing in Halls when there are three teams. EDIT: Quote:
Quote:
Quote:
Quote:
You've forgotten that gold is generated in addition to the faction. Quote:
Last edited by Martin Alvito; Oct 13, 2009 at 07:05 PM // 19:05.. |
||||||
|
|
|
Oct 13, 2009, 05:30 PM // 17:30
|
#243 |
|
Banned
Join Date: Oct 2009
|
I wouldn't even bother with HB if it wasn't for the zkeys.
|
|
|
|
Oct 13, 2009, 05:49 PM // 17:49
|
#244 | |
|
Jungle Guide
Join Date: Apr 2005
Profession: N/A
|
Quote:
Even if tit-for-tat could exist in HB, any tit-for-tat behavoir other than something that looked EXACTLY like RR (which is not tit-for-tat) would be non-equilibrium and thus worse off for everyone involved. |
|
|
|
|
Oct 13, 2009, 06:15 PM // 18:15
|
#245 |
|
Older Than God (1)
Join Date: Aug 2006
Guild: Clan Dethryche [dth]
|
1) Not resigning is NOT strictly dominated by resigning. If you cared about your rating, you'd always play. Your statement is true for most players on RR day, but not all.
2) This IS reciprocity on a mass scale. You map out because you can reasonably expect your opponent to map out when you draw blue. If you couldn't, it would make no sense to map out. Of course, you wouldn't be in the arena in the first place if you expected to have to fight. What's interesting is that the T4T punishment mechanism isn't necessary for the cooperative equilibrium to obtain. You expect to see 'tit' obtain only because of the threat of 'tat'. What appears to be happening is that the expectation of observing 'tit' in the future makes the arrangement self-enforcing. But that makes no sense. If the situation is as you stipulate and you never face the same player again, you're better off acting as though you want to play. The opponent will map out, your win rate improves with sufficiently low time investment to make doing so worthwhile, people eventually catch on and the cooperative equilibrium breaks down. That isn't happening (yet), which is interesting. Last edited by Martin Alvito; Oct 13, 2009 at 06:25 PM // 18:25.. |
|
|
|
Oct 13, 2009, 06:44 PM // 18:44
|
#246 | |
|
Grotto Attendant
Join Date: Apr 2007
|
I usually avoid threads like this because it seems they are inevitably filled with garbage. Imagine my delight to discover instead a discussion of game theory!
Quote:
1. The GW playerbase is not perfectly rational. 2. The GW playerbase is not perfectly perceptive. They may believe they will see the same player again. 3. The GW playerbase is not perfectly intelligent. Working out with certainty that defecting is a better idea is probably beyond the abilities of some and, while within the abilities of others, too much mental work given that RR works "well enough" ("satisfice" mechanism). 4. RR yeilds some psychic benefits that defecting does not -- the thrill of doing the forbidden and the chance to "send a message" to a-net. Perhaps these psychic benefits are more valuable than the marginal improvement in win rate of defecting over RR. |
|
|
|
|
Oct 13, 2009, 07:03 PM // 19:03
|
#247 |
|
Older Than God (1)
Join Date: Aug 2006
Guild: Clan Dethryche [dth]
|
All of those hypotheses are probably true. I'm not convinced that any specific one explains the observation fully. If the explanation is cognitive limitations, the equilibrium should be more fractured than it is.
Something systematic (generalizable?) appears to be at work here. |
|
|
|
Oct 13, 2009, 08:14 PM // 20:14
|
#248 |
|
Guest
Join Date: Dec 2006
|
Wow.. have any of you ever read any game theory? You all bring up good points but somehow you fail to agree on a common model.
Here's my take on it: 1. Rational behaviour. This is basis for any model. This point should be unnecessary to debate IMO, it is something you can generally accept as given. The playerbase is large enough to expect rational behaviour on a general scale. 2. Complete/total information. This is definitely a crucial variable in economic/game analysis. It does influence how RR is played out, but as I will show the ladder element of HB will for all practical purposes remove this variable from the RR Game. 3. External effects. Chthon mentions "psychic benefits" which is one such external effect. I will get back to this as it does influence how RR is played out, however I can say already at this point that it's likely working in tandem with the reciprocal effect of the main reward. (title+faction) In their conclusions I find traversc to be correct, and Martin Alvito wrong. (though he brings up good points!) RR HB Game Analysis: 1. Introduction and problem layout. AFAIK this phenomenon did not exist until about a month ago. The interesting thing is why this didn't emerge earlier and what makes it sustain itself and even grow proportionately. If you think this post is growing out of proportion skip directly to point 4. 2. Transaction costs. Some activities may be favorable but are not set in motion because the entry threshold or the transaction cost is too high initially. In this case this is because no one was resigning, everyone played to win. Being the only one to resign would simply not help one bit. Someone(soldat and his friend I believe) became aware of how the rating system and pairing system worked(ladder). By losing a number of games in a row they would lower their rating substantially("tanking the rank") increasing their chances of playing each other and other low-ranked players by a significant margin. They extended their idea and invitation to GWGuru and a small base of people were effectively farming each other in the low end of the ladder. Even with waiting times and only winning 49% of matches this method proved quicker than playing "legit". In other words they had by sharing information and paying the initial transaction costs set up a favourable system for themselves. 3. A simple game. To quickly show how rational behaviour can be generalised into a model I'm adding a simple game. This shows how acting in pure self interest will always be favourable in an unregulated system. My action: Selfish Altruism(solidarity) The action of others: Selfish Altruism(solidarity) I would add this into a table at this point but I cba  Basically it sums up to show that when acting individually it's always favourable to act in self interest(selfish). If everyone is altruistic(sharing the fruits of their labor for instance) I can still reap more benefits by not sharing the products of my own work. If everyone is being selfish it would still pay off for me to act in self interest, ie. why should I share when no one else is? If we assume(and we can) that everyone is rational and sets up this argument everyone will end up acting selfish in this system. This is a "law" and is the main driving force behind market theory. (Adam Smith and all that)4. Why RR is a market model. How does this translate to RR? Simply because the RR system that has been set up works the same way a market does. When everyone is acting in their own self interest it will be to the benefit of all. This is due to how the matching system in HB works. The system will try to pair people who have a similar rank(based on your rating), this is done to reach as many fair matches as possible(hence being more fun for the community, and it's a good thing). The winner will receive rating and the loser will lose rating based on a formula I don't really have present in mind. Instead of actually playing games it is much quicker to resign half your matches and win the other half, the RR system lets you do just that by setting a common rule that "Red Resigns". This is not arbitrary, because red is always the player with the highest rating him resigning ensures equilibrium. If blue was to resign the higher ranked player would eventually climb out of the "RR area" and get into the "serious area" where people who play for fame and glory rather than faction alone prevail. This means that for both players it is rational to have red resign. Any other way of doing it will be unfavourable for both, the blue player will lose(bad) and the red player will only get an immediate reward. He will have to "tank" his rank back down again by losing about the same number of matches to get back into the RR area. So as long as this state of affairs has been set up it is indeed reciprocal and will remain so for as long as players find it rewarding. If the rewards are removed or rewards can be obtained more easily elsewhere the system will collapse and might not easily be started up again. (Reference to 2 about transaction costs) 5. External effects. The main model can be modified by adding in other variables. However, I can't identify any variables that strictly oppose how RR works. For one as chthon mentions we have psychic effects. I would speculate that one effect would be the feeling of belonging to a community. When someone "is kind" and resigns to you you will want to pay the favour back in kind increasing your chances of resigning next time you are red. I believe this is true for the majority of players, it's human nature to repay good with good. Of course you have other elements like people who are upset about "cheating the system" and similar who will refuse resigning, maybe verbally abusing others etc. However, these are in a clear minority and the majority still finds it rewarding to partake in the system. Also chthon(wow dude, hardest nick ever to spell?) brings up lack of intelligence. This is true, some players out there(you don't know who you are! hehe) will not understand how the ladder works and will think they are being really smart by never resigning. Of course they will initially reap all the benefits but will ultimately bring themselves out of "friendly RR zone" and end up facing more people of their own kind (people who don't resign for whatever reason). I think I've covered the most important aspects, and... long post is long I guess. Also I don't know why i use british spelling not american.. don't flame on that account, I'm just as likely to write american english in another post. Edit: Screwed up my numbering.. Last edited by Sankt Hallvard; Oct 13, 2009 at 08:25 PM // 20:25.. |
|
|
|
Oct 13, 2009, 09:34 PM // 21:34
|
#249 | |
|
Krytan Explorer
Join Date: May 2005
|
Quote:
This isn't anything new. |
|
|
|
|
Oct 13, 2009, 09:56 PM // 21:56
|
#250 | |
|
Banned
Join Date: Oct 2009
|
Quote:
|
|
|
|
|
Oct 13, 2009, 09:59 PM // 21:59
|
#251 | |
|
Guest
Join Date: Dec 2006
|
Quote:
Last, but not least /rps command was removed. I think the /rps system was more similar to a tit for tat-system where agreements were made based on the assumption that people would play by the rules. The RR system does not require people to play by any rules. |
|
|
|
|
Oct 13, 2009, 10:26 PM // 22:26
|
#252 | ||
|
Jungle Guide
Join Date: Apr 2005
Profession: N/A
|
Quote:
Also, if you take the fact that most people aren 't completely rational actors and a percentage would rather punish others for trying to "beat the system," then there is quite a bit of downside to staying on as red. Regardless of what your opponents do, it is better to both leave as red and to leave as blue if your opponent doesn't. Quote:
Last edited by AtomicMew; Oct 13, 2009 at 10:28 PM // 22:28.. |
||
|
|
|
Oct 13, 2009, 11:21 PM // 23:21
|
#253 | |
|
Hall Hero
Join Date: Aug 2005
Profession: E/
|
Quote:
|
|
|
|
|
Oct 14, 2009, 12:27 AM // 00:27
|
#254 | |
|
Guest
Join Date: Dec 2006
|
Quote:
I made a rather elaborate post on this earlier in the thread. I'm fully aware that HB has been farmed by other methods before. The new feature here is the "game theory" aspect of it, fundamentally different from the other concepts. |
|
|
|
|
Oct 14, 2009, 12:28 AM // 00:28
|
#255 | ||||
|
Older Than God (1)
Join Date: Aug 2006
Guild: Clan Dethryche [dth]
|
Quote:
Quote:
Quote:
The fact that red is always the higher rated player is interesting (and explains both why red resigns and why you get punished in the limit for deviating). It's unnecessary to equilibrium. If the blue/red assignment were random, either color could be the designated resigner. You'd still win and lose half of your matches on average, and the rating problem would even out (usually) by chance. EDIT: Sankt, you're missing the forest for the trees. Hawk is correct that there is no difference between /roll resigning and RR. It's functionally the same game. The only difference is that you're not relying on chance to take care of the rating issue. Quote:
Last edited by Martin Alvito; Oct 14, 2009 at 12:40 AM // 00:40.. |
||||
|
|
|
Oct 14, 2009, 12:55 AM // 00:55
|
#256 | |||
|
Guest
Join Date: Dec 2006
|
Quote:
Quote:
Quote:
That's all I'm saying. I know it sounds very abstract and might not be so easy to see the direct impact on practical play. In short RR is a vastly more efficient way of farming and in all likelihood increasing both the players involved and the total output of zkeys compared to any other method(by a lot). |
|||
|
|
|
Oct 14, 2009, 01:10 AM // 01:10
|
#257 | |
|
Hall Hero
Join Date: Aug 2005
Profession: E/
|
Quote:
|
|
|
|
|
Oct 14, 2009, 04:00 AM // 04:00
|
#258 | |||||
|
Jungle Guide
Join Date: Apr 2005
Profession: N/A
|
Quote:
However, the fact that the system hasn't broken down and won't any time soon shows a flaw in your line of argument. Sankt's post is basically spot on. Quote:
 )Quote:
Quote:
Actually, I'd say that /roll grew much faster than RR did. Given more time, the "range" would have developed naturally, and similarly, cheaters would push true /roll-ers down in rank, simultaneously, they would rise in rank and have to tank to get back down. As martin pointed out, you don't need rank information to approach an equilibrium. You just need a way to make the game non-symmetrical, which color designation and /roll both do. The difference I think is mostly psychological. With /roll, it feels like you always have a chance to win the game, even if it's always 50%. It doesn't feel so much like you are just giving up, as with RR. This could explain why it took so long for RR to catch on. Last edited by AtomicMew; Oct 14, 2009 at 04:31 AM // 04:31.. |
|||||
|
|
|
Oct 14, 2009, 05:26 AM // 05:26
|
#259 | ||
|
Forge Runner
Join Date: Oct 2006
Profession: E/Mo
|
Quote:
The removal of /roll was probably one of the most hilarious things in the history of Guild Wars looking back at it. Anet introduces people to things that make them WANT to exploit the game, and instead of removing this, they make a sad attempt at removing the exploit. The problem? As long as these things exist, people will always find a way to exploit your game to get them. Quote:
They introduce a problem into their game (titles), and instead of removing the problem (because too many people like it), they try to fix everything AROUND the problem (obviously to no avail). Then they introduce another problem (HB), and instead of fixing the problem (because they can't), they remove it from the game (because not as many people like it). It is a bit humorous to think about in this way. |
||
|
|
|
Oct 14, 2009, 05:50 AM // 05:50
|
#260 | |
|
Academy Page
Join Date: Dec 2007
Profession: E/
|
Quote:
Every time you win, your rating increases. The higher your rating, the higher your likelyhood of facing non resigning enemies. => The more you win, the more avarage time will you need to win the following match. The opposite is true as well. The more you loose, the less avarage time will you need to win the following match. Every time a normal Player faces a "good" RRer, his chances of winning are ~100% Every time a "good" RRer faces another , his chances of winning are ~0% Every time a "good" RRer faces another "good" RRer, his chances of winning are ~50% The two groups seperate rather fast, leading to RR nearly always facing other RR. In the bottom regions, you will find the diehard RR players, zoning 10-15 seconds after loading, before the start of the match, sometimes even if they are blue, as waiting would cost you another 15-35 seconds to see wheter or not the opposite player may be afk. Those are the 600-800 crowed that only farm titlepoints, they are active even without the z quest and have been for month. |
|
|
|
 |
|
«
Previous Thread
|
Next Thread
»
| Thread Tools | |
| Display Modes | |
Linear Mode
|
|
All times are GMT. The time now is 12:50 PM // 12:50.
jQuery(document).ready(checkAds()); function checkAds(){if (document.getElementById('adsense')!=undefined){document.write("_gaq.push(['_trackEvent', 'Adblock', 'Unblocked', 'false',,true]);");}else{document.write("