gameplay database; gameability resources?

[urlwolf]

I've realized that if I want to write some game-like experiments, the, I have to learn more about games. We are talking about extremely easy to implement stuff, not complicated 3D hyperrealistic shooters.

The basic idea is that people spend hundreds of hours on the net playing games. If we could use all that activity to test cognitive hypotheses, that'd be absolutely fantastic. Luis Von Hahn (CMU) has presented a fantastic video where he shows how one can use people playing activity to solve computationally hard problems.

Von Hann and Dabbish (2004) use an online game to solve computationally hard problems such as tagging pictures. Many people play over 40 hours a week, and when people play, they help determine the content of images on the Web by providing meaningful labels for them. They claim that if the game is played as much as other popular online games, all images on the Web can be labeled in just a few weeks. This approach is simple but novel and changes the way machine learning looks at the complex problem of image tagging. Human computation is yet another algorithm that can be optimized, for example, by improving the game. Presenting experiments as games could radically change both the amount and the quality of data collected to test cognitive theories.

Google has funded Luis, and has adapted the game to tag pictures for google indexing: they created a very interesting game for the very boring task of tagging pictures (http://images.google.com/imagelabeler/). ;

The 'games' I'm thinking about are this simple.

I'm not a gamer. But I need to acquire a basic knowledge of what makes a task a game; what makes a game addictive; and what makes a game fun. All of these, while being simple.


To all people interested in game programming here...
Where do I start?

Is there any website about gameability (with simple recipes about how to create simple, addictive games)?

Any resources and pointers appreciated.

PS: Mouser, you are gonna love the video...

[urlwolf]

I forgot. I'm also interested in simple games with an userbase already that keep the logs of lots of matches (or game history). Chess is a bit too complicated for this, but there must be other games that keep some kind of history and have large user communities.  Do you know any?

[tinjaw]

You can start with:

• A Theory of Fun for Game Design  by Raph Koster
• Chris Crawford on Game Design  by Chris Crawford
• Rules of Play: Game Design Fundamentals  by Katie Salen and Eric Zimmerman

Those should get your head wrapped around games in a good way. I'm in the "serious games" business and the little group I am a part of believe that the only thing that differs between a game and a simulation is the learning objective. If it is to learn something then it is a simulation, if it is to have fun, then it is a game. At first pass, I would imagine it is very similar when comparing computerized tasks to games.

[urlwolf]

Thanks a lot tinjaw. That seems like a good state-of-the-art collection.
There is something that I'm thinking may sink my idea though. Most games have a component of randomness. This makes it difficult to compare two players strictly speaking (one may have been favored by the random environment)... this is true both for complex simulation-like things like the ones you mention, and for simple things like tetris.

hmm...

[tinjaw]

There are many games that do not incorporate chance. Many puzzle games and the "choose your own adventure" type games come immediately to mind. Additionally I would assume that random chance might be fine if it is one of the variables you are looking at. For example, determining why people still don't understand that if you get heads 1,000,000 times in a row you still have a 50% chance of it coming up heads on the next flip. And chance is fine if it simply a matter of choosing between similar outcomes. For example did the actor exit stage right or stage left. It may be irrelevant to the outcome, but give that player some variety.

[urlwolf]

What are some good forums for "game theorizing" (not to mix with game theory, which is an interesting branch of math, but not exactly what I need)?

Is there any classification of games? (I haven't read the books you recommended yet, there might be one there...).

[tinjaw]

Unfortunately I am unaware of any such forums, but that is not to say they don't exist. I just haven't stumbled upon any and I haven't sought them out.

As for taxonomy of games, there are many. I am not sure in what way you are looking to differentiate between them? Do you mean RTS, strategy, FPS classifications or casual versus serious games? Can you elaborate?

[urlwolf]

Can you elaborate?

yes. Let's see. I have a wide range of interests, but let's concentrate on just one. Right now, I'm thinking about something that could be called a strategy game. It has to be a one person game, and all the actions and states of the system should be recorded, so I can use some "replay" function to "playback" a game. In this case, I want to show two video replays to participants, and ask for similarity judgments. The initial state of the system should be the same for all participants, i.e., no random factors (although of course, games will evolve differently depending on each participant's actions).

Ideally, it'd be an online game (flash?) and there should be  a large database of games people have played already (I would use this as a training corpora for the model that will make predictions about similarity judgments). When I say large, I mean i.e. > 6 months of game playing histories.

Games should be short and with a clear objective. Things as simple as snake or pacman would be ok I guess, no need for fancy graphics. However, in the case of 'snake' the random placement of mushrooms would be bad (random variability added). And these are not necessarily strategic, but that's ok.

What I want is some game where plays are stored in a database already (large), and I can play back some gameplaying sessions to real people (players) and ask them for similarity judgments. This similarity judgment task may look stupid for some non-strategy games (e.g., a driving game) so I want something where the question actually makes sense, and where people would happily do this kind of judgment.

I know that the requeriment of a large db is though: most games only keep a final score, not the entire game.

What would be a good way to get something like this? Maybe posting in some game developer forums (if they exist!)?

Thanks!

[mouser]

raph koster's page is a good starting point: http://www.raphkoster.com/
i also have a really good book but i don't have it with me -- when i get it ill post the info.

[tinjaw]

Oh Oh I am sooo close to being able to answer your questions directly, but I will have to waffle around the edges for a bit with the hope of laying out the landscape as I see it and where I see your interests in relation to the others. 

Educators are looking for ways to use games to teach. These "serious games", educational games, are rapidly growing beyond teaching young children their ABC's and 123's. For example this study looks at games to teach civil engineering. I suggest taking a look at http://www.fas.org/gamesummit. It is a report on the findings of a summit on educational games by the Federation of American Scientists.

So one group is looking to utilize good educational groups to teach, and another group is looking at how to produce these good games to give to educators to use. You have researchers looking at why a successful educational game is successful.

Then there is the whole idea of having in-game tutoring and in-game coaching. I've learned a lot about that topic by working with Chad Lane. Chad is also working on a post-game interactive tutor/coach. In the military we have this thing called an After Action Review (AAR). It is the idea of going over what just happened, the good and the bad, to learn all the available lessons and do better next time. In one instance Chad is working to have the characters that appeared in the game as A.I.'s be available for a Q&A session after the game. For example the student could ask what the character was thinking at some point in the game. Or ask why a character did what it did in the game. Very interesting to me.

And then there is a group, of which I consider myself a part of, that advocates that games can serve as excellent tools of instruction if the instructor is properly trained on their use and the use of the game meets the instructor's learning objective. For example, I might be able to use lemonade stand to teach logistics to army officers, if my learning objective is to show how supply and demand interact.

So I've dabbled with researchers who research if and why games are effective teaching tools. And I have dabbled with developers researching how to make good games. But I don't think I have dabbled with researchers using games as a research tool that weren't researching games. So, you may be on to something here.

But there must be others that are interested in using games as a means of eliciting behavior that they wish to research. There is probably a few needles in a Google haystack if you use the keywords SCORM, which is an emerging set of technologies and standards that could be adapted to your desired use, Automated AAR (like this and this) which track not just the final score, but everything from start to finish, and Chad's idea of a reflective tutor, that could interact with the research subject (person) in a manner almost 180 degrees opposite of how Chad would use it, so you could determine why the person made the choices they did.

Games should be short and with a clear objective.

I completely agree. Combine Terminal Learning Objectives with Einstein's "Make everything as simple as possible, but not simpler." Design the game to get at what you want to examine and not waste resources on extraneous details and development. Any competent developer in the serious game business should be able to adapt current best practices and current R&D into "educational games" to develop a "research game".

I'm interested enough, that if my schedule permits, I may be interested in exploring this further with you from the developer's point of view.

[tinjaw]

urlwolf,

Let's move this into the world of the concrete. I think it might help.

This may seem like a lot of "homework" for a stupid DC forum posting, but I think it may be important to you "getting it". Write up a fake grant proposal. Describe to me what you want to research, why, how, etc. Describe to me what data you want to collect. Tell me your research/learning objectives. As well as describe to me an existing, or not-yet-existing, game that suits your needs.

Yes, this would be a lot of work, but I think it might help you better get the landscape of the realm you wish to explore further. I am interested enough is what I believe you are looking for, that I may be willing to commit to volunteer work if you flesh this out enough to become a real world academic research project.

[urlwolf]

Hmm, ok, it looks feasible.
Unfortunately, nothing will get done sooner than next month, as I have two pressing deadlines for projects. But definitely worth doing probably. Thanks.

[tinjaw]

I was reading an interview with a successful game producer and found this quote of interest.

Since then, I’ve done a lot… consulting here and there (with the most interesting gig being working with a behavioural psychologist to turn his research into a game),

And I find that interesting, because that is one of the reasons I am interested in working with you on a project as you describe it, I think it will be very interesting and a great learning experience for me. I think software engineering can learn a great deal from behavioral psychology. Software interacts with humans, so we, as software engineers, should work to find out how to best interface/interact with humans.