Adaptive game design

[Tom Clayton]

I have recently been going through the 2003 essays on gaming theory, specifically GNS and femine game design.  It occured to me that choosing an RPG to fit a specific gaming group is much like choosing an optimisation algorithm to tackle a specific problem.  If taken in this context, RPG design falls under the "No Free Lunch Theorum for Search/Optimisation" which loosely states that all optimisation algorithms can be considered to provide equal performance when that performance is averaged across the set of all problems.  So, for any search/optimization algorithm, any elevated performance over one type of problem is exactly paid for in performance over another type.  In gaming terms, if the set of all problems consist of a collection of all gaming groups, then this translates to: All RPGs offer the same performance (read fun) when trialed accross all gaming groups, with some RPGs providing better performance for specific groups.

Adaptive optimisation algorithms were developed to overcome the "no-free luch theorum", where the way in which the algorithm functions is incrementally changed based upon performance related criteria.  I am wondering if this theory could be ported into RPG design and given mechanical form so that gaming groups can easily alter the RPG incrementally during the length of a campagn (or even during play), to produce greater performance (Fun) for the gaming group.

Tom 

   

[Eero Tuovinen]

Well, as to the first, I don't really think that all games provide equal performance over the set of gaming groups. Going into information theory, this is because there is no limitless and arbitrary set of gaming groups like there are information retrieval problems. The gaming groups we have are what we get, and while we could perhaps postulate some strange group that likes contradictory, badly written games that go nowhere, in the real world these do not exist, and therefore should not be considered in game design. The alternative is to say that all games are equally good in that for any game you can imagine a group that would be the perfect fit for it. Some of those groups might not be human or even intelligent, but hey, that's life in the heady depths of theory.

As for adaptative game design, such is discussed now and again here and in other rpg design circles. Conventional rpg design already involves plenty of adaptation - most games are more realistically considered tool-sets the players use to fulfill the game's goals, for instance. Going beyond that, one can speculate, and speculation has been done. As I see it, the major problem in making a rpg that is "everything to everybody" is that the numero uno purpose of a distinct, productized roleplaying game is that it gives you a creative framework and goal which you then execute in playing the game. Making a game that is everything to everybody is like making a boardgame in a completely white box and filling the box with white sheets of paper and some pens - go ahead and make your own perfect game out of that. To make an actually useful and fun roleplaying game, you need to provide the creative goals that actually allow the audience to choose this game to be the game they are going to play now; without this step your game will never get played, as the impulse to play a game comes from this creative specification in the first place.

In short: how would your adaptative game present itself to begin with, and how would adaptation be useful once a group of players had selected the game based on the strengths of that initial presentation? If your game promises to be X and Y and Z, its ability to adapt into A and B and C is of academic importance for gamers who have pre-selected themselves as a group that is specifically committed to X and Y and Z for now.

[Tom Clayton]

Well, as to the first, I don't really think that all games provide equal performance over the set of gaming groups. Going into information theory, this is because there is no limitless and arbitrary set of gaming groups like there are information retrieval problems. The gaming groups we have are what we get, and while we could perhaps postulate some strange group that likes contradictory, badly written games that go nowhere, in the real world these do not exist, and therefore should not be considered in game design.

Yes i concede this point.  I was thinking more of an analogy, but you have convinced me that this may not be the correct way to think of this problem.

The alternative is to say that all games are equally good in that for any game you can imagine a group that would be the perfect fit for it. Some of those groups might not be human or even intelligent, but hey, that's life in the heady depths of theory.

I think that this was what i more had in mind.

In short: how would your adaptative game present itself to begin with, and how would adaptation be useful once a group of players had selected the game based on the strengths of that initial presentation? If your game promises to be X and Y and Z, its ability to adapt into A and B and C is of academic importance for gamers who have pre-selected themselves as a group that is specifically committed to X and Y and Z for now.

This is what i am interested in (purely from theoretical stand point).  I wonder if it is possible to create a coarse grain system, with pre-generated subsections, each assiciated with a style of play, and each tunable during play.  This way, a group can decide what type of game they wish to play by choosing subsection, and then tune each subsection. 

I performed a search on this forum for "adaptive" and it didn't come up with much.  Do you have any links to previous similar topics.  I should probably read these first before rambling off down a path that is already well trodden.

Tom 

[Eero Tuovinen]

Searching for this sort of stuff is tricky, hopefully others have better memory for it than I do... "Scattershot" was an early game project this decade that tried something sort of like that... I don't think that the term "adaptative" has seen much currency here, the sort of discussion I'm remembering would be more likely to call the goals of the game "universal" or "generic" or "cross-agenda", "agenda-neutral", "hybrid agenda" or something like that.

I don't remember any particular breakthroughs on this matter, though. Mostly the discussions have been about some people explaining to others why such an idea is futile, and other people speculating about how it might or might not work. Nothing concrete has resulted, anyway.

[chance.thirteen]

Wouldn't drift be the adaptive process?

[Eero Tuovinen]

Not really if the adaptation is written into the game text. Drift is what you do when you transform the game text into practical play. If the text has adaptative instructions, then you might of course drift those, but following them is specifically non-drifting play.

[Warrior Monk]

This is what i am interested in (purely from theoretical stand point).  I wonder if it is possible to create a coarse grain system, with pre-generated subsections, each assiciated with a style of play, and each tunable during play.  This way, a group can decide what type of game they wish to play by choosing subsection, and then tune each subsection. 

This made me thought of "Big Eyes Small Month" -this game offered rules to replicate any anime-styled atmosphere. Of course, even on the most recent editions rules can't cover enough game styles or replicate a varied enough amount of settings, considering the variety of anime titles on the market. Even with the use of their several different expansions the design invites to drifting. I'd extrapolate that any adaptive design would tend to make players drift the original ruleset -not based on the BESM example, but on buyes expectative and posterior disappointment after checking the rules.

So wouldn't be more efficient/interesting to include a chapter about drifting?

[Meramec]

Interesting topic.

How about this game design:

Each character sheet has three parts:  (1) The OD&D section, (2) the Shadows of Yesterday section, and (3) the Pendragon section.

When the player wants to "game", use section (1).  When the group wants "story", use section (2), and when the group wants to wallow in the glory of Arthur, use section (3).

Is this the end result of what you are looking for? I know it's clunky, but doesn't this design fit the "adaptive" definition?  It switches "game modes" on the fly.

I think it's absolutely possible to build a game for players who are after different things. Most of my gaming history has been with a group made up of folks who come to the table for different reasons. I've developed a system for dealing with this that works for our group, and I think that in general it's a reasonable goal for any designer.

Would an RPG experience be "better" if everyone were engaged by the same thing?  Perhaps, maybe even probably.  But given a real-live game group, is it possible to design a game with rules that different players interact with to different degrees?  I think the answer there is absolutely, and the resulting game, while not necessarily being the "best" overall at any one thing that it does, would be the best way for THAT GROUP to game.

All the best,

John

[Callan S.]

It occured to me that choosing an RPG to fit a specific gaming group is much like choosing an optimisation algorithm to tackle a specific problem.

I think your problem is that your seeing this fitting an RPG to a specific group as your own problem. It's rather like looking at the problem of trying to feed someone, rather than that person pursuing food themselves.

It's like choosing where to eat - someones hunger will drive them to solve that problem themselves. And if they don't hunger - you can't solve it for them, because they.just.don't.hunger.

[JoyWriter]

I think the difference between an adaptive system and blank potential is that one matches the situation, and the other may or may not match it depending on what is done to it. In other words an adaptive system has inputs for information that are processed in order to produce the appropriate changes in the system and produce the right game. (It's a determinate control system acting leading from the possibility space of possible players to the possibility space of fun games)

To be honest, that sounds like rpg theory!

Interpret player preferences and inclinations with sufficient specificity to suggest systems that would suit their combined creative needs.

Or, in other words,

Work out what people want so you can design a game for them that enables them to do that, or find a game that already does that.

But in general, learn-able, transferable terms.