Recommend
 
 Thumb up
 Hide
18 Posts

BoardGameGeek» Forums » Board Game Design » Board Game Design

Subject: Game Theory rss

Your Tags: Add tags
Popular Tags: [View All]
Dhaval Mistry
Canada
flag msg tools
Hey guys,

I was wondering if anybody has heard of a thing called "Game Theory"? Game theory in a nutshell is a computer program where it tests all the possibilities of the given rules. It can be programmed to output pretty much any stat of your game. For more information, check out the link below.

https://en.wikipedia.org/wiki/Game_theory

I am looking for a programmer who can build this for me. I am designing a board game and I have the rules ready. I need to collect data about the game and see if any of the rules are broken or can be exploited.

If you are interested then please pm me your rate and contact info.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
J C Lawrence
United States
Campbell
California
flag msg tools
designer
badge
Avatar
Nope, that's not what it is. I think you need to read that page you linked.
21 
 Thumb up
0.02
 tip
 Hide
  • [+] Dice rolls
Nathaniel Grisham

Indiana
msg tools
mb
I believe game theory is a field of behavioral study for analyzing how people will behave when put in a certain situation. One popular example is the prisoner's dilema (https://en.wikipedia.org/wiki/Prisoner%27s_dilemma) I'm sure that various studies will model some situations programatically, but the program does not make the study.

I'm not sure I understand the scope of what you are asking. You want someone to program a model of your game, complete with AI players, to see whether the AI stumbles upon any combos that are too powerful? You might (read: will) get faster/cheaper results by showing the game to a group of seasoned Magic players.

If you want someone to model your rules to see whether there is anything is missing, or just doesn't work, then I would still recommend observing some playtests of your game. That is where the questions will come up.

If you are set on getting someone to program this for you, then I think you're more likely to get someone to bite if you can better explain the scope of what you want. I think that this is a much larger project than what I think you think it is.
13 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Dhaval Mistry
Canada
flag msg tools
Yes I understand that it falls into psychological studies but the concept can be applied to making a game model and analyzing the programmed outcomes.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Geoffrey Burrell
United States
Cedar Rapids
Iowa
flag msg tools
Is this the same Game Theory proposed by economist John Nash, the subject of the movie "A Beautiful Mind?"
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Adam P
United States
Seattle
Washington
flag msg tools
Avatar
mb
dhavalmistry wrote:
Hey guys,

I was wondering if anybody has heard of a thing called "Game Theory"? Game theory in a nutshell is a computer program where it tests all the possibilities of the given rules. It can be programmed to output pretty much any stat of your game. For more information, check out the link below.

https://en.wikipedia.org/wiki/Game_theory

I am looking for a programmer who can build this for me. I am designing a board game and I have the rules ready. I need to collect data about the game and see if any of the rules are broken or can be exploited.

If you are interested then please pm me your rate and contact info.


Test all possibilities of a rule set? Aside from IBM's Watson I believe computers still have a very hard time with linguistics.

You may be able to create an intermediate programming language that is used for the ruleset, then a computer could graph limited permutations. You would still lack the path of "creative play" where players find interesting ways to use the rules to achieve strategic paths. An example would be the starvation strategy from Stone Age.

Even then, the amount of permutations reaches a point that it takes years to compute. Think about the game "go" which ONLY NOW through machine learning was able to go up again top human players.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Nathaniel Grisham

Indiana
msg tools
mb
adamredwoods wrote:

Even then, the amount of permutations reaches a point that it takes years to compute. Think about the game "go" which ONLY NOW through machine learning was able to go up again top human players.

And even though a computer can play the game at such a high level, the game is not solved. We simply don't have the resources to map out every possible play. It's one of the reason's why there is so much research going into quantum computing, but I digress.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Cheb
United Kingdom
flag msg tools
Avatar
mbmbmbmbmb
Depends how simple the game is I guess! (And it'd have to be simple..)
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jeremy Lennert
United States
California
flag msg tools
designer
Avatar
mbmbmbmbmb
dhavalmistry wrote:
Yes I understand that it falls into psychological studies but the concept can be applied to making a game model and analyzing the programmed outcomes.

Despite the tantalizing name, "game theory" has very little to offer to game designers.

Game theory analyzes simple interactions where players make a simultaneous decision. You could use it to analyze, say, rock-paper-scissors (or a variant of it).

You could not use it to analyze Tic-Tac-Toe, because players take turns making a series of decisions. For that, you'd want the similarly-named combinatorial game theory.

Combinatorial game theory has a lot more relevance to most board games, but it's just not advanced enough to handle games of the complexity that most people want to play. It's generated some very interesting results for, e.g., dots and boxes. And it helps us build some abstract concepts that can be used to describe and discuss game positions, such as temperature. But it doesn't reveal much about the strategy of, say, Chess.

(And that's not an accident. If we had the mathematics to easily analyze a game and prove the optimal strategies, people would probably stop playing that game for fun. If you ever found a simple mathemtical theory that fully explained the strategy of your game, that would probably not be a good thing for you.)

Even if you did plug your game into a big supercomputer and it spit out a gigantic game theory matrix combining together every decision in your game, I don't know what you think you'd do with the output.

It sounds like you've fallen victim to a buzzword, and have optimistically assumed that it will solve the problem you're interested in without taking the time to really understand what it does or how it works.
10 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Was George Orwell an Optimist?
United States
Corvallis
Oregon
flag msg tools
Abdullah Ibrahim - Water from an Ancient Well
badge
Avatar
mbmbmbmbmb
The trick in applying game theory for use with boardgames isn't in programming the selection, it's in evaluating the relative worth of potential outcomes. Once you've decided on values, it's a simple matter to lay out a matrix and roll a die to make a selection.

I have an old book that's ideal for gamers, titled The Compleat Strategyst by J.D. Williams (and yes, that is the correct spelling for the title). Mine is hard cover, copyright 1954 by McGraw-Hill, but I'm pretty sure it was reprinted as a paperback at some point. If you're interested in this subject, go find a copy.
3 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Dhaval Mistry
Canada
flag msg tools
Sphere wrote:
The trick in applying game theory for use with boardgames isn't in programming the selection, it's in evaluating the relative worth of potential outcomes.


This is what I was actually looking for. I am designing a co-operative game which involves several choice of actions on any given turn. I just wanted to plug those values in and see for example what the least amount of turns the game can be completed.

This is just one of the things I wanted to calculate. I guess I should have been more specific.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jeremy Lennert
United States
California
flag msg tools
designer
Avatar
mbmbmbmbmb
dhavalmistry wrote:
Sphere wrote:
The trick in applying game theory for use with boardgames isn't in programming the selection, it's in evaluating the relative worth of potential outcomes.


This is what I was actually looking for. I am designing a co-operative game which involves several choice of actions on any given turn. I just wanted to plug those values in and see for example what the least amount of turns the game can be completed.

This is just one of the things I wanted to calculate. I guess I should have been more specific.

Sphere isn't saying that game theory is good at evaluating the worth of potential outcomes. He's saying that's the tricky part that you need to solve before you can apply game theory in the first place.

Determining the minimum number of moves it takes you to win the game is essentially a matter of trying every possibility (or finding clever mathematical tricks for your specific game that let you prove some possibilities are bad without bothering to try them). That's not game theory, it's just brute force calculation.

If your game is simple enough, you could theoretically program a computer to try every possible option and report on the results. But unless your game is very simple, that would probably be a lot of work; you are unlikely to find a stranger who will volunteer to do it for free.

And even after you run the program, knowing the best possible outcome won't necessarily tell you anything useful about what it's like for real people to play your game. Optimal play is usually not a realistic assumption, and just because the perfect strategy wins in 8 turns doesn't mean that a pretty good strategy will win in 10 (or even that it will win at all). So unless there are some special circumstances for your game, that computer program is a pretty questionable investment of time.
4 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Pelle Nilsson
Sweden
Linköping
flag msg tools
designer
badge
Avatar
mbmbmbmbmb
Sphere wrote:

I have an old book that's ideal for gamers, titled The Compleat Strategyst by J.D. Williams (and yes, that is the correct spelling for the title). Mine is hard cover, copyright 1954 by McGraw-Hill, but I'm pretty sure it was reprinted as a paperback at some point. If you're interested in this subject, go find a copy.


They did reprint it, but even better they also put it online for free, and it is well worth reading (even though personally I skipped over a lot of the maths, as most of it is about tricks to calculate matrices manually, but these days we have computers for that):

http://www.rand.org/pubs/commercial_books/CB113-1.html

I think game theory is actually great to read a bit about for designing games. Not that you can solve games or even in a meaningful way solve anything but the smallest subsystems of a game, but to learn about how you can in theory think about your game as different states and pay-offs and all that, even if you can't actually map it all out in practice. You might not need more than the chapter or so often included in books about game design, but that book is free and really fun to read so why not.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Scott Johanson
United States
Washington
flag msg tools
If what the OP is asking for was possible at this time, it would either be prohibitively expensive or video games would not have bugs or balance issues.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Dhaval Mistry
Canada
flag msg tools
Thanks for clearing it up guys. I was under the wrong impression.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Justin R
United States
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
Antistone wrote:

Despite the tantalizing name, "game theory" has very little to offer to game designers.

Game theory analyzes simple interactions where players make a simultaneous decision. You could use it to analyze, say, rock-paper-scissors (or a variant of it).

You could not use it to analyze Tic-Tac-Toe, because players take turns making a series of decisions. For that, you'd want the similarly-named combinatorial game theory.

Combinatorial game theory has a lot more relevance to most board games, but it's just not advanced enough to handle games of the complexity that most people want to play. It's generated some very interesting results for, e.g., dots and boxes. And it helps us build some abstract concepts that can be used to describe and discuss game positions, such as temperature. But it doesn't reveal much about the strategy of, say, Chess.

(And that's not an accident. If we had the mathematics to easily analyze a game and prove the optimal strategies, people would probably stop playing that game for fun. If you ever found a simple mathemtical theory that fully explained the strategy of your game, that would probably not be a good thing for you.)


I disagree with nearly everything quoted above. The main point that OP is missing the mark on this topic is true, but that can be said without belying a gross under-appreciation for the practical application of robust mathematical theory. Sphere is, of course, correct that the weakness lies not in the framework, but the inputs. No doubt that computing capability also presents a clear limitation to application (e.g., chess AI, go AI, etc.). But neither of these limitations are native to the mathematical theory.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jeremy Lennert
United States
California
flag msg tools
designer
Avatar
mbmbmbmbmb
JJRR_Esq wrote:
I disagree with nearly everything quoted above. The main point that OP is missing the mark on this topic is true, but that can be said without belying a gross under-appreciation for the practical application of robust mathematical theory. Sphere is, of course, correct that the weakness lies not in the framework, but the inputs. No doubt that computing capability also presents a clear limitation to application (e.g., chess AI, go AI, etc.). But neither of these limitations are native to the mathematical theory.

You say you disagree with nearly everything, but the only specific point of disagreement that you cite seems to be a vague characterization of the overall usefulness of the underlying theory. Perhaps you could articulate your disagreement in greater detail?

The overall thrust of your post seems to be to say that it's not a model's fault if you can't get the necessary inputs or computational resources to run it. I would say that is a pretty sorry defense. Most problems are easy if you assume perfect information and unlimited resources. If you want to claim that a model is powerful--and certainly if you want to claim that a model is useful--I would expect it to work with limited information and computation.

It is certainly possible that I have undersold game theory, but your post does little to convince me.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Paradox Games
msg tools
publisher
mbmbmbmbmb
Remember in Wargames, when the computer tried every possible game of Global Thermonuclear War?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Front Page | Welcome | Contact | Privacy Policy | Terms of Service | Advertise | Support BGG | Feeds RSS
Geekdo, BoardGameGeek, the Geekdo logo, and the BoardGameGeek logo are trademarks of BoardGameGeek, LLC.