GNS - does one preclude another?

[resistor]

At the point when you call something "loosely described" science I'm hoping like hell you don't work in the nuclear industry.

Are you just being deliberately obtuse? Here's a last stab at explaining how it works, as pedantically as I can.

The clustering of the results is an empirical measurement. What you choose to call each group is, obviously, subjective. That said, one typically looks at what kind of common answers define the cluster, and choose a name that makes sense for it.

For example, if there's a cluster of people who answered yes to "Do you try to optimize your characters?" and no to "Would you play a less powerful character for story reasons?", the reasonable surveyor might choose to label that group as "power gamers." Importantly, the clusters are determined first, and are irrespective of any names you choose to give them.

You're welcome come up with your own word for that group, or for any of the others. The evidence still shows that those groups exist, and that they are defined by survey responses that the members of the cluster have in common.

 

[Ourph]

As for the model/theory/prediction thing, I think importing analytical notions from the natural sciences into the humanities is not all that helpful.

First, great post. My experience with the Big Model has been very similar.

Second, the above is probably useful advice. However, to me, the Big Model seems very squarely in the realm of Social Psychology (I wouldn't recommend telling a Social Psychologist that they are part of the humanities rather than the natural sciences, BTW). Of course, that might just be because I'm equipped to view it through the lense of natural sciences and woefully ill-equipped to view it through the lense of, for example, literary theory or philosophy. When the only tool you have is an electron microscope, every problem looks like it needs to be stained with uranyl acetate.

 

[pemerton]

Ourph, thanks - and fair enough on the methodology point. I'm in the opposite situation from you, in that my natural science education stopped when I stopped high school, whereas I am a humanities academic who teaches philosophy and social theory.

 

[nedjer]

Are you just being deliberately obtuse? Here's a last stab at explaining how it works, as pedantically as I can.

The clustering of the results is an empirical measurement. What you choose to call each group is, obviously, subjective. That said, one typically looks at what kind of common answers define the cluster, and choose a name that makes sense for it.

For example, if there's a cluster of people who answered yes to "Do you try to optimize your characters?" and no to "Would you play a less powerful character for story reasons?", the reasonable surveyor might choose to label that group as "power gamers." Importantly, the clusters are determined first, and are irrespective of any names you choose to give them.

You're welcome come up with your own word for that group, or for any of the others. The evidence still shows that those groups exist, and that they are defined by survey responses that the members of the cluster have in common.

There are, perhaps, categories and categories, sciences and marketing 'sciences', and “lies, damned lies and statistics.”

Clustering offers a problematic approach to statistical analysis and the categorisation of clusters is subject to the same methodological considerations as any element of an experimental process.

In addition, for GNS, WotC data or the Big Model to be considered empirical research surely they'd have had to set out and test a hypothesis or two. If they had we'd have had meaningful outcomes ages ago; instead of a unique wisdom to be gained from some sort of RPG-specific insight on the world, which all modern science has somehow overlooked.

Setting up a quasi-scientific 'understanding' of games and gaming, without reference to scientific discipline, is, perhaps, way Battlefield Earth?

For anyone interested in the science of gaming it starts here. Unlike GNS, WotC data, the Big Model and the Forge the information there is based on studies by tens of thousands of professional scientists

 

[pawsplay]

That is not correct. Whether or not a model has predictive power depends on how it resembles reality.

No, that only tells you how much it has.

That scale model of the Eiffel tower, or a kid's plastic scale model airplane, does not necessarily resemble the Eiffel Tower except visually. Unless you want to claim "show how the real tower looks" as a prediction, such models have no predictive power.

I think such models are completely functional as visual representations of the Eiffel Tower.

In order to be predictive, there needs to be a mechanism for the model to take data of a circumstance and then present some result that predicts how reality will behave.

So, for example, with an atmospheric model, you feed the model some hypothetical or real data, and it spits out what tomorrow's weather will be. With a structural model you put the model under some load, and you watch how it performs, and you draw analogy to how a real structure will behave.

GNS lacks any place to hook in your data to get a prediction out.

Describe to me a model that does not take the data of a circumstace and present some prediction of how it will behave. What use would an atmospheric model be if I felt its outcomes had nothing to do with initial conditions and weather processes. What use is a structural model if I do not believe that is performance is an outcome predicated by the structure? If I run a weather simulation and it tells me that tomorrow it will be 500 degrees, how effective is it as a model?

 

[pawsplay]

These are not predictions. They are assumptions the model requires.

No, the model itself does not "predict" this. That the model is more useful does not actually fall out of the model itself.People who like the model assert it.

Again, this is an assertion or assumption of the model, not something that comes from the model.

Assumptions, assertions, etc. are the basis of a theory. A model or any kind of coherent construct is supposed to have validity end to end. You cannot just say, "If G-N-S, then incoherent play." It is fair game to attack the premises of a model. "Imagine the Earth is a cube..." is not an auspicious start to a model of Earth's weather patterns.

A model is not merely descriptive. If it were descriptive, it would merely take information already on hand, and summarize it in some way. But a model actually allows you to generate new information by setting the model into operation. Insofar as tarot cards are used to make predictions, they are a model, and theories can be generated using them as a model. Insofar as they are just used as a thought process, they are not a model. If you want to be finicky, the tarot deck does predict a limited range of prediction outcomes of a reading, if you view the reading itself as a microcosm. If you turn up the Hangman and... the Ace of Spades, then you realize that your model was insufficient, and that other cards are in the deck besides tarot cards that must be accounted for.

In the humanities, a model justifies itself by producing something of value. In my view GNS starts with mush and generates mush.

 

[Umbran]

No, that only tells you how much it has.

Well, then "all models are predictive" rather in the same way that "all cars drive on roads" - Perhaps correct in theory, but patently incorrect in practice - as the mass of junkyards out there attest.

I think such models are completely functional as visual representations of the Eiffel Tower.

Yes, it is functional as a visual representation - but a visual representation is not necessarily predictive. Unless, I suppose, your prediction is "this is how the thing will look tomorrow", which seems to me to be trivial not only in the colloquial sense, but in the mathematical sense as well.

Describe to me a model that does not take the data of a circumstace and present some prediction of how it will behave.

The Mona Lisa is a wonderful visual representation (and thus a model) of a 16th century woman. There is no facility by which you can put data into the representation and get a prediction about how the represented woman would behave. The predictive behavioral model of "16th century woman" exists inside our skulls, not in the painting.

The painting may be used as a set of input data into our skull-borne predictive model - but it has commonly been remarked that the enigmatic expression makes the woman's mood rather inscrutable!

In order to predictive, a model must include rules of behavior, rules for evolving state as time goes on. Visual representations do not include such rules.

 

[nedjer]

Looks like you guys'll need one of these to get to the heart of the matter. 3428 pieces later and you should hit some kind of uber-Zen.

That or you won't give a flying f*** about pretty much anything

 

[pawsplay]

Well, then "all models are predictive" rather in the same way that "all cars drive on roads" - Perhaps correct in theory, but patently incorrect in practice - as the mass of junkyards out there attest.

That's not my position. The analogy would be more like, "All cars can be assessed as to how they drive on roads."

Yes, it is functional as a visual representation - but a visual representation is not necessarily predictive. Unless, I suppose, your prediction is "this is how the thing will look tomorrow", which seems to me to be trivial not only in the colloquial sense, but in the mathematical sense as well.

If I turn model of the Eiffel Tower upside down, will it look like the actual tower would if it were upside down? Can I turn the model around and explore its dimensions? If someone wanted to paint the Eiffel Tower pink, could I not paint a model and examine the aesthetics in that fashion?

The Mona Lisa is a wonderful visual representation (and thus a model) of a 16th century woman. There is no facility by which you can put data into the representation and get a prediction about how the represented woman would behave. The predictive behavioral model of "16th century woman" exists inside our skulls, not in the painting.

Does she have a back of her head? I'm asking, because you can't actually tell. It's entirely possible she has only half a head. However, I predict that the Mona Lisa has a back of her head, at least conceptually.

The painting may be used as a set of input data into our skull-borne predictive model - but it has commonly been remarked that the enigmatic expression makes the woman's mood rather inscrutable!

The fact that a mood can even be ascribed to a static figure establishes the qualities of a model. Her face seems to have a recognizable expression, so we interpret the figure as if it were an actual face.

In order to predictive, a model must include rules of behavior, rules for evolving state as time goes on. Visual representations do not include such rules.

That's just not true. For instance, visual representations have rules such as interposition. For instance, I can readily create a visual representation of my living room, and you can pose questions about possible furniture configurations from that model.

 

[MormonYoYoMan]

Dogs in the Vineyard is heavy on simulation (specifically in its pursuit of simulating Mormon culture)

Not in the version I read. Very, very non-LDS (Church of Jesus Christ of Latter-day Saints/Mormon) background. However, a very, very, VERY good game of forcing decision making upon the players, and having them (us) play characters outside our comfort zones.