Curse

Ctb · Oct 30, 2008, 03:50 PM // 15:50

Quote:

Originally Posted by Martin Alvito

Finding the flaw in this system is simply a matter of solving an easy math problem.

I'm not going to argue with you. I've been a software engineer my whole life, so I'll just say this: put your money where your mouth is then.

But let me know when you give it a go. I'll be placing a bet against you to make some easy money.

Martin Alvito · Oct 30, 2008, 04:20 PM // 16:20

Quote:

Originally Posted by Ctb

I'm not going to argue with you. I've been a software engineer my whole life, so I'll just say this: put your money where your mouth is then.

But let me know when you give it a go. I'll be placing a bet against you to make some easy money.

I use game theory on a daily basis in political science research. This is a simple mechanism design problem.

The essentials of the proof are in the first post. Assuming rationality, and assuming time has value, all players use their hourly ragequit. Reporting then has some risk of resulting in a loss of time. Without getting into the ugly details of sequential equilibrium and why (ragequit, /report) falls apart as a pooling equilibrium, I'll give you the intuition.

Even IF we could overcome the social choice problem, and everyone plays /report as a strategy, there's some risk of having someone screw up. Maybe the player is on the phone and playing, and the player doesn't /report like they should due to the distraction. Even if everyone agrees to /report every leech, the contract isn't enforceable. Further, there's a risk of time loss associated with using /report, which players don't like.

Since this risk exists, the rational strategy for everyone is (ragequit, not /report). The rational thing to do is let someone else pay the cost of using the /report function, and just play. The upshot is that you should expect massive underreporting in RAs.

This isn't such a problem in organized team play (because of lack of ragequits and the ability to organize your team to /report), but it's clear that the /report feature of the dishonor mechanism can't work in the format it was originally designed for. Using /report is costly and provides no observable benefit. It follows that most people won't bother, and that means that we have a flawed system on our hands.

Ctb · Oct 30, 2008, 04:51 PM // 16:51

Quote:

I use game theory on a daily basis in political science research.

Which would be relevant if the underlying problem here involved the people and not the fact that computers can only operate on a true or false basis given ANY set of starting criteria.

People think in nuance, computers don't. You cannot solve a social problem with a computer. At least not any we currently have.

This is why people don't like it when social theorists try to apply their work. They never have any practical understanding of why their grand theories don't work half the time in the real world.

Yea, I'm sure if we had enough time and enough money we could sit around and make a piece of software that could do this, but we don't, so the answer is, for all practical purposes: it can't be done.

Ellix Cantero · Oct 30, 2008, 04:57 PM // 16:57

Quote:

Originally Posted by Numa Pompilius

Specifically sloth and greed.

Welcome to the Guild Wars community, where negativity equals cool.

I think you meant "Welcome to teh Internets".

Martin Alvito · Oct 30, 2008, 10:04 PM // 22:04

Quote:

Originally Posted by Ctb

Which would be relevant if the underlying problem here involved the people and not the fact that computers can only operate on a true or false basis given ANY set of starting criteria.

You still just don't get it. The dishonor and /report systems are designed to solve problems of human behavior. Namely, ragequitting, spam and abusive commentary in local chat.

If we want to solve problems of human behavior, we should start from the presupposition that we want to build an institution that leads to the desired behavior. Here, you write a piece of code that creates the right incentives for the player base.

The current piece of code doesn't do that. It does a reasonable job of deterring ragequitting, but it has a flaw (underreporting) as well as an unintended consequence (if you're going to be stuck with this group of people for at least a match, entry spiking becomes more attractive).

Quote:

Originally Posted by Ctb

People think in nuance, computers don't. You cannot solve a social problem with a computer. At least not any we currently have.

Totally specious. Botting is a social problem. People in low-income countries run bots because they can make money IRL exploiting the wage differential by reselling the outputs at high-income country prices. So you write a piece of code to detect and ban the bots, and the problem is reduced.

Now, if the loot system were designed properly, such that accomplishments rather than drops determined access to rare skins, you wouldn't have the botting problem at all. It's hard to fault the designers of, say, Everquest, D2 or Acheron's Call for this failure, since that's the era when the problem was first confronted. It's much more reasonable to question why WoW and GW weren't built on different economic models than they were.

In the case of the language problem: as noted, we know that NSA uses code to sift through the massive web of global communications to find desired information. The same principle applies here; program a computer to flag accounts when players input things you don't want them saying too often, then save the transcripts in question and have humans review for potential action.

Whether that's cost effective or not is another matter entirely...but it'll solve the problem. Judging by everything I see scroll by in local chat in PvP, I'd say that /report isn't getting the job done.

Quote:

Originally Posted by Ctb

This is why people don't like it when social theorists try to apply their work. They never have any practical understanding of why their grand theories don't work half the time in the real world.

Got news for ya: I spent five years employed by one of the big banks in this country. The lesson in the real world was the same as the one in academe: if you want to figure out why the people around you do what they do, put yourself in their shoes. Then you can predict their behavior. Further, if you change the incentives that make them do what they do, you can make them do different things. (That last observation came from managing people at the bank.)

When we (and everyone else in the industry) went over to commission-style incentives on the retail side, I left. The coming catastrophe in financial services was obvious if you were actually booking the loans.

Sir Seifus Halbred · Oct 31, 2008, 01:12 AM // 01:12

Quote:

Originally Posted by baltazar knight

in snowball arena the same guy reported me 10 times in a row for leeching....

and i didn't get dishonorable

That is impossible-you can only report someone for leeching once. If you were in the same match as him and he reported you for leeching, he can only do so once.

baltazar knight · Oct 31, 2008, 11:51 AM // 11:51

Quote:

Originally Posted by Sir Seifus Halbred

That is impossible-you can only report someone for leeching once. If you were in the same match as him and he reported you for leeching, he can only do so once.

i mean we won 10 times and he reported me each batle

Ctb · Oct 31, 2008, 02:46 PM // 14:46

Quote:

You still just don't get it.

No, you still don't get it. If you think you can do it, then go do it. A lot of game makers would pay you a lot of money to design a cost-effective method of automating the reliable identification of bad player behavior. Until you do that, you can run your mouth until you're blue in the face, I'm just going to tell you to prove it.

Not only do I not believe you could effectively model this, I KNOW you can't program a fix for it.

I'm done arguing with you. Now I remember why I put you on ignore.

Koudelka · Oct 31, 2008, 02:58 PM // 14:58

Talking about imploding threads...

The Knights Templar turned into a cesspit of flaming, and "imploded" just like you said, and much to our confusion, was closed down for months. I remember getting a "Come back! We miss you!" E-Mail, so I started using them again. At this point I was accused of necro-posting, which was rubbish, because there were literally no new threads as it had been closed for months... hence I came to Guru.

And look what I found! More crying and more flaming and more rudeness! Not to mention the horrible evangelicalists... but say no more!

On topic: I remember the good ol' days when to report someone you used to have to fill out a ticket, took forever, so a lot of people didn't...

Ctb · Oct 31, 2008, 05:55 PM // 17:55

Quote:

Originally Posted by Koudelka

More crying and more flaming and more rudeness!

pfft. You think this is flaming and rudeness? This is nothing compared to what this place was like a year or two ago when it was almost all "hardcore" players.

Welcome to Guild Wars: the worst gaming community on the internet.

Numa Pompilius · Oct 31, 2008, 06:00 PM // 18:00

Quote:

Originally Posted by Ctb

pfft. You think this is flaming and rudeness? This is nothing compared to what this place was like a year or two ago when it was almost all "hardcore" players.

Pretty much true. Although at least the first years there at least existed positive threads also, in addition to the flaming and whining. Haven't seen a positive thread in months.

Quote:

Welcome to Guild Wars: the worst gaming community on the internet.

Inherited from battle.net, the previously worst gaming community on the net.

Martin Alvito · Nov 01, 2008, 12:56 AM // 00:56

Quote:

Originally Posted by Ctb

Not only do I not believe you could effectively model this, I KNOW you can't program a fix for it.

All I'm going to say is this: the last time we had a protracted argument on here, it was about a problem that I SOLVED that you could not solve. Instead, you QQ'd because you were unable to solve it.

Feel free to search for the relevant thread.

You're absolutely right that I can't code a program to fix it. I spend a fair amount of time teaching statistical packages like STATA to do what I want them to do, and I have quite the time teaching them to solve complex tasks. Coding isn't an area of expertise; I just don't think that linearly. However, it's a demonstrated fact that I can induct from RL data to locate solutions to problems you consider intractable.

My point is simple: we know that solutions to this search routine exist. If NSA can solve it (on a VASTLY larger source of data), why can't anyone else? Provide incontrovertible cost data, and I'll concede the point. Ex: In testing I determined that the last brute force STATA program I wrote would take about 1.6 years to solve the task at hand. In response, I streamlined the code and the solution took three days. Had no streamlining been possible, I'd have thrown up my hands and found a new research project.

Show me that kind of data and I'll believe you. Until then, I have every reason to disbelieve anything you tell me, because you've demonstrated that you're willing to believe that the possible is impossible (without cheating).

Quote:

Originally Posted by Ctb

I'm done arguing with you. Now I remember why I put you on ignore.

The fact that I tell you what you don't want to hear, despite the social costs, is the strongest possible reason you could ever need to listen to me rather than ignore me.

Rhamia Darigaz · Nov 01, 2008, 01:53 AM // 01:53

i love when martin and ctb argue about quantum physics, way more interesting than gw.

Numa Pompilius · Nov 01, 2008, 04:12 AM // 04:12

Quote:

Originally Posted by Martin Alvito

My point is simple: we know that solutions to this search routine exist.

Excuse me, but what search routine?

As far as I can tell you guys are debating how to stop "bad player behavior" and I think you'll find that the biggest hurdle to solving the problem is comprehensively defining it. You guys could try to start there.

If you're specifically talking about botting (I really can't tell), then I would say that it's a safe bet that ANet already have daemons searching for behavioral patterns indicating automation. And, of course, that the botters have tried to counter that by introducing variable timing and a degree of randomness in the bots actions.
If you can design statistical tools to reliably tell such "camouflaged" bots from real players, then I'd urge you to do so: I'm pretty sure both NCSoft and Blizzard would be willing to pay big bucks for such tools.

Avarre · Nov 01, 2008, 04:16 AM // 04:16

Quote:

Originally Posted by Ctb

Welcome to Guild Wars: the worst gaming community on the internet.

There are definitely worse communities in terms of manners - though possibly not many worse in terms of understanding their own game, hence why people complain in every way about everything.

FlamingMetroid · Nov 01, 2008, 04:18 AM // 04:18

Quote:

Originally Posted by Ctb

Welcome to Guild Wars: the worst gaming community on the internet.

wuts a korean FTP MMO

Evil Neato · Nov 01, 2008, 04:46 AM // 04:46

Quote:

Originally Posted by Ctb

pfft. You think this is flaming and rudeness? This is nothing compared to what this place was like a year or two ago when it was almost all "hardcore" players.

Welcome to Guild Wars: the worst gaming community on the internet.

No wai.

Not enough Brazilians.

br?
br?
br?
food plx
br?
teleport plx
craw craw
hunted soul of zelda

Try playing Tibia.

Chicken of the Seas · Nov 01, 2008, 06:45 AM // 06:45

I got reported when I died and got rezzed at the shrine, and didn't do anything because it was clear we were not going to die and everyone else had resigned anyway. We were like 4/18 and there was no possible way we'd be able to catch up anyway. Only the one idiot reported me though, not the rest of the team. Reported as being a leecher, mainly cause the guy was probably pissed that we wouldn't win.

snaek · Nov 01, 2008, 06:51 AM // 06:51

reporting leechers is a fail tactic

even when i manage to get everyone on my team to report a leecher
it doesnt do anything...

why?

because leechers r afkers
and how does an extra 10mins of afk time harm an afker?

Abedeus · Nov 01, 2008, 07:00 AM // 07:00

Quote:

and how does an extra 10mins of afk time harm an afker?

Shows you don't know mechanics. 10 points of dishonorable = 1 hour of afk. This doesn't harm him? Hehehe. Yeah, right.