« Mobile MMOs | Main | Meet Mr January »

Jan 22, 2005

Toward A Cheater Predictor

[Ed: Posted on behalf of David Reim, [email protected]]

I imagine that most mmog, VW, and online game companies would agree that cheaters aren’t welcome. The definition of what constitutes a “cheater” is an essay in itself, but for this thread let’s focus on the cheaters who can be characterized as premeditative, chronic, and self-aware (i.e., “I’m going to do this; I’ll do it again if I get a chance; I know that it is cheating.”). The effect of this type of cheating ranges from disruption in game play issues to the infrastructure exploitation (and potential damage) that can be characterized as hacking. Since this is an industry-wide problem, it seems to me that there might be an industry-wide interest it advancing our understanding of these issues. And, I believe that our academic brethren might be in the best position to get the ball rolling on addressing these challenges.

As in all issues of security, it seems like the problem can roughly be divided into two parts: preventing it and responding to it. In my humble opinion, it would be a fascinating exercise to define the online cheater, construct a taxonomy of known methods, and develop a comprehensive approach to minimizing this element of the gaming world (even if it’s theoretical at this point). Perhaps some graduate student wasting away her days playing a mmog while she tries to think up a dissertation topic will step up to the challenge. ;-)

Without making this post to long, I would like to mention one interesting area of this problem. With no academic research or theory to back up this statement, my guess is that the online game cheater probably has personality “markers” that can be screened for. If so, a good study would produce 4 to 6 questions that have a certain percentage chance of predicting a cheater personality. There are proof points in a wide variety of similar non-gaming issues ranging from loan applications to medication compliance. Now what is done with such a screener is a whole other topic but again, perhaps constructing this model is a challenge for our academic colleagues.

on Jan 22, 2005 |

| | | |

Comments

1.

If so, a good study would produce 4 to 6 questions that have a certain percentage chance of predicting a cheater personality.

While this is probably true, wouldn't a cheater simply 'cheat' on this test to avoid being labelled as one?

Posted by: GSH | Jan 22, 2005 at 23:58

2.

While this is probably true, wouldn't a cheater simply 'cheat' on this test to avoid being labelled as one?

As I said, it's not immediately clear how such a profiler would be used. Your question assumes the answers are either made available ahead of time or can be guessed on the fly. While this might be true in some percentage of cheaters, it is probably not true for all cases. If you got their credit card number before they were shown the profiler, that credit card would be an invalid payment method (creating a barrier to reentry). Of course all of this could be worked around by a motivated cheater but since it is relatively low cost to implement, what might a company be willing to impose to reduce cheating? A corollary is, “What would non-cheaters be willing to put up with to reduce cheating?” These issues are addressed in the credit card industry, for example, all the time.

Posted by: David Reim | Jan 23, 2005 at 01:19

3.

For a start, it would be nice if a mechanism existed that allowed developers to share information. Blizzard banned some people for cheating. As a developer, it would be nice to get a head start and put a watch on any account that one of those folks signed up for in a different MMOG.

As an aside, it's not enough to say "Never trust the client" or don't ship with bugs. Developers are limited in what we can legally install and check for on a clients machine. What are other folks doing to harden/encrypt communication protocols, check for debuggers, macros or simply monitor for hacking?

Posted by: Bill O'Neill | Jan 23, 2005 at 01:42

4.

David Reim > The definition of what constitutes a “cheater” is an essay in itself,

Actually I should be working on a book chapter on this very topic right now in stead of using TN as a work avoidance strategy. More specifically I’m working on two angles – the first (initially presented as ‘It is wrong to cheat online’ at AoIR 0.5) very theoretical which positions cheating in respect of the Magic Circle (the explanatory force of which I find I have to defend as the metaphor has come under attack a lot recently) and then suggests that if cheating has a moral component then it is one that derives from the idea of trust / promise breaking. The second paper attempts to find parallels between one supposed duty to obey laws and ones possible duty to obey ‘rules’ (the notion of a rule is something I look at quite closely as rules tend to be defined only within certain limits and are always negotiated).

But this is only one small and highly theoretical take on things, there is a lot of academic work going on in this area right now, including some taxonomies, behavioural and motivational stuff, we have also covered Grief a few times here on TN

In may I posted Good Grief (There are no grief players, only players that grief – discuss):
terranova.blogs.com/terra_nova/2004/05/good_grief_.html. More recently Nate added a news update Grief News: terranova.blogs.com/terra_nova/2004/12/grief_news.html) including a link to a piece about “John Suler's classification of griefers in terms of their intent: antisocial types challenging authority; and those out to getchya, personally” Inflicting pain on 'griefers': news.com.com/Inflicting+pain+on+griefers/2100-1043_3-5488403.html?tag=cd.top.

There are also a bunch of papers out there:

Grief play motivations (2004). C.Y. Foo and E.M.I Koivisto in Other Players conference 7.12.2004.
www.itu.dk/op/papers/yang_foo.pdf

Defining Grief Play in MMORPGs: Player and Developer Perceptions (2004) C.Y. Foo and E.M.I Koivisto in ACM SIGCHI International Conference on Advances in Computer Entertainment Technology, Singapore. June 3-5 2004.
www.itu.dk/op/papers/yang_foo_koivisto.pdf

At AoIR 5.0 last year there was a storm of work on Grief, Cheating and Normative stuff (links are abstracts only):

Mia Consalvo Get your cheat codes here: Online help, offline play, and the structure of the “support side” of the digital game industry
gsb.haifa.ac.il/~sheizaf/AOIR5/270.html

Tim Jordan Online Play: cyberpower, heidegger and mmporgs
gsb.haifa.ac.il/~sheizaf/AOIR5/307.html

Thomas Malaby Ethical Code: Engagements and Evasions in the Design of Virtual Worlds
gsb.haifa.ac.il/~sheizaf/AOIR5/201.html

Daniel Pargman Law and order in online games
gsb.haifa.ac.il/~sheizaf/AOIR5/406.html

Ren Reynolds Is it wrong to cheat on-line ?
gsb.haifa.ac.il/~sheizaf/AOIR5/177.html

Drew Ross How 'woof woof woof' and 'quack quackity quack' create pseudo-community: online gaming without the fringe benefits
gsb.haifa.ac.il/~sheizaf/AOIR5/295.html

Jonas Smith Playing dirty – understanding conflicts in multiplayer games
gsb.haifa.ac.il/~sheizaf/AOIR5/71.html


And in 2003 Mia also presented a paper on norms:
Consalvo, Mia. Control Versus Community: Taking "The Sims" Online

Also everyone should look at the categorisation of player types in Salen and Zimmerman’s (Rules of Play)

And lastly, I run a mail list for posts on new research in cheating: cheaters AT groupcare DOT c0m, email me (ren AT aldermangroup DOT com) if you want to be added.

Posted by: ren | Jan 23, 2005 at 08:12

5.

To be less meta about this subject, I wonder whether such a system would work for the good of a game in the long run. Suppose we have some psychological profiling that is very good at detecting potential cheaters and that the cheaters don’t cheat at the test (getting someone else to take it for them, identity swapping etc etc).

Might it not be good for the game to have some people that do cheat, as it might be the case that for many player types (define in Bartle, S&Z or what ever terms you like), some of the individual in that type are likely to cheat at some point but we need the mix of types including those at the edges (if we suppose that it is those at the edge that are more likely to cheat but also taking into account that cheating acts may be done by any / all at some point) for the game to be good. Yes, this is the ‘are criminals an essential part of society’ argument – paging all sociologies.

Posted by: ren | Jan 23, 2005 at 08:30

6.

I am sure a good profile can be created with 5-6 question, but for game's sake should we start this arms race? Do we want this process to turn into a hacking game.

From another angle with cheat codes common to traditional single players, why should cheats be unacceptable in MMOs (just a theoretical question)?

From yet another angle, why not develop a behaviorial analyzer to find makers of cheats based on past behavior? Behaviorial interviews seem to be better than Q&A interviews, so wouldn't this be a better and easier path? Credit Card companies don't interview people about credit card fraud, do they?

Posted by: magicback | Jan 23, 2005 at 09:27

7.

For a start, it would be nice if a mechanism existed that allowed developers to share information.

Absolutely. I too have been impressed with the aggressive stands taken by Blizzard and Valve lately. Such sharing could include names, like a credit report, but it would also be useful just to share techniques, like CERT does in the area of computer viruses. One could argue that the lack of such a facility means that cheaters have not caused the industry enough pain yet. One might also argue that the ability to detect and eliminate cheaters might be viewed as a proprietary advantage of a company and therefore there would be little incentive to share.

Posted by: David Reim | Jan 23, 2005 at 09:28

8.

Yes, this is the ‘are criminals an essential part of society’ argument – paging all sociologies.

Again, it depends on what type of cheater. I don’t think that any developer would agree that someone that builds or uses a tool to intercept and manipulate the data stream is helping the game at all. On the other hand, some people consider powerleveling “cheating” but I would argue that this falls into the range of acceptable behavior. The gray areas are around things like bug exploitation. However, if I was running a world, I’d do whatever I could to minimize the number of criminals. After all, what good is being God if you can’t optimize! ;-)

Posted by: David Reim | Jan 23, 2005 at 09:42

9.

From another angle with cheat codes common to traditional single players, why should cheats be unacceptable in MMOs (just a theoretical question)?

Cheat codes are put in place by developers with the intention of players using them. If the developers didn’t want those debugging codes in there, they would sweep them prior to code lock. In my opinion, using cheat codes is not cheating because the developers intended you to use them. They are, in fact, a game feature.


why not develop a behaviorial analyzer to find makers of cheats based on past behavior?

Yes the first step would be to determine what type of profiler has the best chance to accurately predict cheaters. There are at least three types of profiling approaches: (1) demographic (also called functional) – used in setting car insurance rates (i.e. how old are you and what are you driving); (2) behavioral – used in issuing credit cards (i.e. do you have a history of bad payments); and attitudinal – used to help select stocks investment (i.e. how risk adverse are you). To find out which small number of questions are the best predictor you need to start out with a large number of questions that draw from each of these three approaches and then use an analysis technique (usually regression) to find the “golden questions”.

Posted by: David Reim | Jan 23, 2005 at 09:59

10.

Here are some additional papers on cheating/griefing:

Kuo, Andy. “A (very) Brief History of Cheating.” (2001) Student Paper (Stanford: STS 145). 16/01/05 http://shl.stanford.edu/Game_archive/StudentPapers/BySubject/A-I/C/Cheating/Kuo_Andy.pdf

Kücklich, Julian. “Other Playings - Cheating in Computer Games.” Other Players. Eds. Jonas Heide Smith and Miguel Sicart. Copenhagen: IT University of Copenhagen, 2004. 05/12/04 http://www.itu.dk/op/papers/kuecklich.pdf

Pritchard, Matt. “How to Hurt the Hackers: The Scoop on Internet Cheating and How You Can Combat It.” (July 24, 2000) Gamasutra. 08/01/05 http://www.gamasutra.com/features/20000724/pritchard_pfv.htm

Truthseeker, Arios. “An Essay on D00dism and the MMORPG.” (2000). 11/02/04 http://www.kanga.nu/archives/MUD-Dev-L/2000Q4/msg00166.php

The last one (a classic...) makes an interesting argument for using the cheaters [or "D00Ds" as the case may be...] as a resource to identify weaknesses in a game. Unfortunately the MUD-DEV archives appear to be down - dunno if anyone else has a live link...

Posted by: Peter Edelmann | Jan 23, 2005 at 16:16

11.

David Reim> "I don’t think that any developer would agree that someone that builds or uses a tool to intercept and manipulate the data stream is helping the game at all."

Why wouldn't they?

Intercepting and manipulating the data stream is a neutral action. It is what you *do* with that data that makes it cheating.

The client, I will remind you, is a device for intercepting and manipulating the data stream. The developer obviously feels building and using that tool is justifiable, as they build it themselves and request that users use it.

I think such third party hacking should be encouraged. As a player, it ensures that the developer will be forced to deal with clunky UI. People whined about UOs targetting from the early days. Cries of "Let us target the health bars!" were raised quite early. It is somewhat instructive that UOAssist was only officially sanctioned once the most important ability (last target) was rolled into the main client.

The fact remains, your users are lazy. They are only going to seek out and install third party tools if they give an advantage. This implies your client is not efficient for playing the game, or your server is buggy. In both cases, you can solve the problem by improving your client and fixing your server.

Sure, this means stupid gameplay mechanics like: "You can't alt-tab, its a feature!" and "There are no maps, you can't tell which way you are facing, and we have this neat skill which tells you which way you are facing, provided you have hit the button 50,000 times" can't be created. As a player, I'm not particularly upset. These are largely Sim-Ant features anyways. "Let's see how players adapt if we give no maps. Hopefully, they'll make extensive hand drawn maps and find great joy in locating landmarks, etc." Ignoring, of course, the possibilities that users write ShowEQ. Bad ants! You optimized the wrong variable!

- Brask Mumei

Posted by: Brask Mumei | Jan 23, 2005 at 19:35

12.

I think such third party hacking should be encouraged.

And when you create a mmog you can encourage it. In the mean time most mmogs are tightening down on their policies. As for manipulating the data stream, WoW is very specific that players can not http://www.worldofwarcraft.com/policy/exploitation.html#dsmdm”>alter the data stream. On the other hand, one could envision a more “open source” approach to a mmog where the data stream and the relevant APIs are published so that anyone can work with them (legally). In fact this is done all the time by the mod builders of other games.


This implies your client is not efficient for playing the game, or your server is buggy.

The way for players to respond to this is to cancel their subscriptions, not violate EULA or other stated policies.

Posted by: David Reim | Jan 23, 2005 at 23:05

13.

> If so, a good study would produce 4 to 6 questions that have a certain percentage chance of predicting a cheater personality.

Somehow reminds me of the visa application you have to fill out to enter the United States: "Are you, or have you ever been a member of the communist party?" and so on. The purpose and the "good" answer to give are so obvious, that the questionaire never catches anybody.

In my opinion to catch a cheater, you would need two things: A rule saying that everything you can do *in* the game is not cheating (which obviously requires good design with no exploitable bugs), and a good encryption of the client/server datastream which is able to automatically detect if people tampered with that.

Anything else means that you probably do more harm with your cheat-prevention than the cheaters did in the first place. You can't ask people whether they ever used cheat codes in a console game before, label everybody who answers yes as cheater, and ban them all.

Posted by: Tobold | Jan 24, 2005 at 08:49

14.

I found the "D00dism" essay in the google cache. The link is really long but it is easily found by entering this into the google search [ MUD-Dev-L 166 of 639 ] (everything between [])

Posted by: Dee Lacey | Jan 24, 2005 at 15:43

15.

FYI
>>Truthseeker, Arios. “An Essay on D00dism and the MMORPG.” (2000). 11/02/04 http://www.kanga.nu/archives/MUD-Dev-L/2000Q4/msg00166.php

Is now a dead link, the text can be found at:
http://www.deinonych.com/archives/deinonych/000032.html

Posted by: ren | Jan 27, 2005 at 08:39

16.

Have you seen this before? It's a number guessing game: http://www.amblesideprimary.com/ambleweb/mentalmaths/guessthenumber.html. I guessed 32038, and it got it right! Pretty neat.

Posted by: Merideth Carleton | Nov 16, 2005 at 19:34

17.

Have you seen this before? It's a number guessing game: http://www.amblesideprimary.com/ambleweb/mentalmaths/guessthenumber.html. I guessed 32038, and it got it right! Pretty neat.

Posted by: Merideth Carleton | Nov 16, 2005 at 19:36

The comments to this entry are closed.

Search

Recent Comments

Powered by Postrank