RPG definitions continued (split)

[Doug Ruff]

When seeking a definition, you cannot hold the two following things fixed:

1. A basic principle or descriptive feature

2. A set of individual instances that are supposed to be included

One or the other has to go.

I hate to disagree with Big Ron, but a definition based solely upon #1 is going to collapse under its own weight eventually.

For example, say that we agree that the definition of a RPG is somehow dependent upon the concept of a SiS. Then we have to agree on the definition of SiS, and also of any of the other words used in the definition.

At some point, there is going to be a disagreement over one of these terms. And the only satisfactory way of resolving a disagreement like this is to start  pointing at things and saying 'this is it' or 'this isn't it.' But that's method #2.

For what it's worth, this is an argument with a long historical pedigree. Method #1 is very similar to the thinking of the 'logical positivists' such as Carnap and (most famously) the early Wittgenstein.

Method #2 is more consistent with later Wittgenstein, who spent most the second half of his career refuting most of his earlier work(!)

My reason for bringing this up here is that I'm worried that any attempt to define RPG using method #1 is doomed to failure. But I'm sure that may will disagree - should this be taken out into a new topic?

Tetsuki
(Doug Ruff)

[Ron Edwards]

Hi there,

It's cool to disagree with me, Doug.

I'm not sure we are disagreeing, however. My goal was to point out that the two types shouldn't be employed simultaneously while trying to work out the definition, not that one or the other was free from limitations. As I see it, both have fundamental built-in limitations.

It seems to me you're pointing out the limitation with one of them, and that's fine.

As food for thought, regarding applying #1, consider the role of the type specimen in biological nomenclature. No matter what, the actual real specimen that bore a particular name will always carry that name. Any other creature that will be included under that name has to be justified in doing so according to a set of rules that apply to all creatures and names, not just to that particular type.

The interesting point about this technique is that it anchors a type #1 approach to reality, but only with a single and only-ever anchor, and proceeds from there on principle alone.

Whether this approach or anything like it will be useful for defining role-playing games is unknown. I raise it here only as a mind-food regarding how the two different ways to define things get put into successful applications.

Just in case anyone leaps to this obvious but dubious conclusion, I do not propose Dungeons & Dragons as a type specimen, nor am I saying that this particular solution is appropriate for discussing role-playing.

Best,
Ron

p.s. I split Doug's post from What is a role-playing game?

[Doug Ruff]

Hi Ron,

I don't think that there is a fundamental disagreement: I agree with you that #1 and #2 are fundamentally different description types and should not be confused with each other.

I think I may have a different view of the relationship between the two. I think that the 'basic principle or descriptive feature' has to be described as the sum of the individual instances at some point, in order to have any use as a definition.

To give an example, I can describe a 'duck' as a bird with particular markings and features, or as a creature with a particular DNA sequence, or so on.

But at some point, in order to make sure that we agree on what a duck is or isn't, I have to actually point at a duck (or a picture of a duck, or a DNA sample of a duck) and say 'that is a duck'. To be really sure, I'm going to have to point to several ducks.

I think the same applies to classifying RPGs: a lot of the 'I think this is/isn't an RPG posting' (Method #2) is an attempt to determine whether or not we agree on what a Method #1 'descriptive' definition of RPG actually means.

As you said, this isn't a simultaneous process, but there appears to be an heuristic relationship betweeen the two.


One other thing you said interested me:

'The interesting point about this technique is that it anchors a type #1 approach to reality, but only with a single and only-ever anchor, and proceeds from there on principle alone.'

I'm not quite sure I understand this (but I would like to!) Can you explain further your thinking here?

Regards,

Doug

[Ron Edwards]

Hiya,

I think we're actually very close to agreement.

The protocol as I see it is this:

1. Establish one or more instances which indubitably will be included in the definition. The fewer the better, and the criteria for picking it (or them) should be absolutely clear.

2. Expand outward from this (or these) specimens to the rest of reality, including or excluding each object into the category, based strictly on a principle or concept. This principle or concept needs to be deductively sound.

The point is that after step #1, the category is not permitted to include anything that wasn't stuck into it at the outset, unless step #2 says so. And that as step #2 proceeds, you may find yourself including things you wouldn't have expected, and not including things you would have thought fit in there perfectly.

So yeah, you start with "this is a duck," pointing to a critter. But after that, you don't include anything into "duck-ness" just because it seems like it should, or on the basis of any old variable that happens to correspond with the initial object. You proceed outward based on variables that you've justified in some other way (in biology, these have awful names like "synapomorphy," but never mind).

The one thing that's different about your point and mine is that I think starting with several different "maybe-ducks" already included is a bad idea. One of them should carry the type specimen label, so that no matter what, even some or all of the other things we think might be ducks can get tossed out - only the one we chose as the type can never be tossed out.

This important point covers us from making a big looping circle, from "group of X" yielding "shared features of Xs" to justify our "group of X" again.

So to do any of this with role-playing, we'd have to decide what the type specimen is, and why. Then we'd have to agree on the principle that's to be applied to everything outside of it, and why that's the best one. Then we'd have to sift through all the candidates, and endure the cries of outrage when certain things that have been called "role-playing" get excluded and other things that haven't get included.

Anyway, I really want to say that (a) I don't know whether this is the best route to take regarding role-playing, and (b) I don't think any of this is currently possible, given the state of discussion. I think a lot of the recent threads are a good foundation for maybe getting to a decent attempt at a definition in the future, and that's a good thing.

Best,
Ron

[Doug Ruff]

Hiya,

The one thing that's different about your point and mine is that I think starting with several different "maybe-ducks" already included is a bad idea. One of them should carry the type specimen label, so that no matter what, even some or all of the other things we think might be ducks can get tossed out - only the one we chose as the type can never be tossed out.

Ron, you're absolutely right. Looking at my last post, I missed something important out.

In order to be included within the definition, each individual candidate-duck has to be agreed to be a duck by both of us. And this process will start with a single maybe-duck.

So I would follow your protocol, except that as a result of discussing one of the later specimens, we may then choose to alter the (Method 1) word-definition.

And that's what I think is happening in the other threads. Do you concur?

Till later,

Doug

PS Thanks for taking time out to discuss this with me - I've only been here a few days, and it is nice to feel so welcomed!

[Ron Edwards]

Hi,

Yes, that's what's happening with the other threads. It's a slowwww process.

Best,
Ron

[neelk]

The protocol as I see it is this:

1. Establish one or more instances which indubitably will be included in the definition. The fewer the better, and the criteria for picking it (or them) should be absolutely clear.

2. Expand outward from this (or these) specimens to the rest of reality, including or excluding each object into the category, based strictly on a principle or concept. This principle or concept needs to be deductively sound.

The point is that after step #1, the category is not permitted to include anything that wasn't stuck into it at the outset, unless step #2 says so. And that as step #2 proceeds, you may find yourself including things you wouldn't have expected, and not including things you would have thought fit in there perfectly.

The one thing that's different about your point and mine is that I think starting with several different "maybe-ducks" already included is a bad idea. One of them should carry the type specimen label, so that no matter what, even some or all of the other things we think might be ducks can get tossed out - only the one we chose as the type can never be tossed out.

I'm not sure I agree. When I do mathematics, I start with some collection of phenomena that I want to characterize in some way. It's desirable to include lots of "maybe-ducks", to use your terminology, because at this point I don't really know what I'm talking about and want to have as many possible examples as possible to learn from. Then, I try to get some hypothesis "If X, then Y" (such as "If it's a duck, then it quacks"), and try to work out (ie, prove) whether X really does entail Y. Generally, this proof will fail -- I'll find some counter-example or flaw in the argument. And that's a desirable feature, because how the argument fails tells me how to change X and Y so that the hypothesis holds. So I learn that some maybe-ducks aren't (geese aren't ducks), and some are (Donald is a duck), and that I should really consider some quacks to be oinks, and so on. Then I try the proof again, and usually fail again.

I repeat this process until I get to some definition of "duck", "quack" and whatever that work. The final "duck" and "quack" may have little to do with the intuitions that started the process -- including the examples that I thought were indubitable at the outset. Imre Lakatos wrote a very nice book about this process called Proofs and Refutations. The key point is that the structure of an argument ends up shaping what the definitions are.

So I don't think you need a type specimen to invent a definition of roleplaying. However, the flip side is that I'm not sure a definition is necessary, since I don't know of any arguments which crucially turn on what is and isn't an rpg. Until you have such an argument you want to make, then you can't run a definition through the iterative refinement process of coming up with proofs and refutations, so it will be hard to invent a compelling definition.

[Ron Edwards]

Hiya,

Right, Neel - the type-specimen approach is distinct to biology, in which it is indubitable that a critter of this type specimen does indeed exist. After all, this object is a dead carcass of one of them.

Other sorts of definitions and proofs need different processes. What process we should use for "role-playing" remains a mystery.

I knew people were gonna do this ... folks, I'm not trying to prescribe how to define role-playing. My explanation was intended for Doug, to deal with his specific questions about my distinctions about types of definitions.

When I wrote,

Whether this approach or anything like it will be useful for defining role-playing games is unknown.

... I meant it. For the record, Neel, you'll note I have not proposed any definition of role-playing throughout all of my essays, nor in the latest series of threads, and that's due to the same reasoning that ends your post.

Best,
Ron