« Meta Grinding and Nemo | Main | Sticky NeoPets »

Jun 18, 2005

Comments

201.

Aaron>Richard's view is essentially eliminativist: He thinks that there's the code and the data and that's it. Once you have those, you have a (minimal) virtual world. Everything else is illusory.

I wouldn't say everything else was "illusory", it's more like "the icing on the cake".

Also (this is kind of implied, but just in case anyone reading this later doesn't realise it) the code and the data have to have certain properties. It's not just any old code and any old data; it has to be code and data that will, when run, support what most of the rest of you want to call a "virtual world".

A virtual world that has no players is still a virtual world. If it could never have (multiple simultaneous) players, it's not a virtual world; at best, it's still in development.

Hellinar>Hardware + code + system data = Computer Generated World

No, those worlds aren't computer-generated. They may be computer-realised, but they're not computer-generated. They were generated by designers and programmers, not computers. There may be some computer-generated parts of it (eg. set the parameters and the planet-generation tool will go off and create a random planet for you), but the computer isn't generating the world in which you're playing. You could probably say it was "maintaining" it, though.

Richard

202.

Richard>Also (this is kind of implied, but just in case anyone reading this later doesn't realise it) the code and the data have to have certain properties. It's not just any old code and any old data; it has to be code and data that will, when run, support what most of the rest of you want to call a "virtual world".

Thought experiment for you, Richard (or anyone else still reading this thread): Suppose there were some kind of superdupercomputer running code + data that provided the kind of behaviors and potential interactions you require of a minimal virtual world. However, amazingly enough (and against all odds) this was all achieved by using some kind of lookup table (it could be random, based on what day of the week it is, the temperature inside the room it sits in, whatever) that had all possible states of the world listed as individual entries, where each entry in the table was a global state for the virtual world.

Would you consider that a virtual world?

The point I'm getting at is whether you (or anyone else) thinks that it's not enough that the code + data generate the 'right' behaviors. Is there also a requirement that these behaviors have to be generated in the 'right' way?

Just curious what you think.

Myself, I'd say it wasn't a virtual world (at least using 'vw' in the sense you use it).


Aaron

203.

Richard> No, those worlds aren't computer-generated. <

Agreed. In an informed audience I would go for Computer Rendered Worlds. But I have used both formulations in conversations with non-technical people, and “Computer Generated Worlds” generates less blank stares than “Computer Rendered Worlds”. As programmers, we would both likely look at some surprising output of a program and say “I wonder why the programmer programmed it to do that”. But a non-technical type would likely say “I wonder why the computer is doing that”. So which formulation you choose depends on your audience I think. And in my experience, the general audience is going to think of the computer as doing it all.

204.

Richard: When it comes to refuting the suggestion that "Physically simulated 3D worlds are fundamentally different from text worlds", which is how we got into this discussion.

Now you see, I feel the other way around. If we look at text muds vs 3d worlds as behavior settings, then we don't find much in the way of fundamental difference. If we look at them as designed artefacts, then the differences in medium take a larger role.

I'm also unhappy with the idea that a virtual world isn't a virtual world unless it has players.

I'm not saying or implying this. What I am saying is that once you open up a virtual world to participation, frequently there is no practical difference between designer-created structure and user-created structure. Most of my early mud experiences were on muds that didn't have a hard line between designer/user.

205.

Thank you for that post, Aaron. This goes a long way for me understanding your perspective. Many of the previous posts I percieved to be in direct response to my views. My solicits for illustration and example were not meant to understand your perspective of my views, but I was trying to understand your view, independent of my own to see which schools of thought we are originating from.

Aaron>Existence has no meaning outside of a world
Aaron>There is no spatiotemporal reference at all

The questions I presented were rhetorical. On the virtual plane, we exist outside the world. Between the virtual plane and the real plane, there is a discontinuity in spatiotemporal reference.

Before you may disagree with that notion, consider this analogy.

You are probably familiar with a place called Flatland. As beings defined in three spatial dimensions in SpaceLand, we can observe the two-dimensional Flatland, but inhabitants of Flatland can only observe us if we interact in Flatland. That is, we intersect their plane. As a member of SpaceLand intersects with FlatLand, there exists a clearly distinguished transworld identity. Simply by removing a spatial dimension, there are now things true in Flatland that are not true in Spaceland, and vice versa. These are two separate modes of existence, and thus, as I consider them, separate planes.

But, as can be readily seen, Flatland, although separate, is also contained in SpaceLand, but not vice versa.
This relationship has clear effects in terminology and their transformation between planes.

I consider this analogy parallel - Flatland is to SpaceLand as the virtual world is to the real world.

I have consciously avoided citing Leibniz, Locke, Lewis, Kripke and others because I wanted to (1) discuss virtual worlds in particular and not metaphysics, (2) avoided these cites most particularly because there exists active debates on those views and I didn't want to engage those debates in this forum and 3) keep the conversation open to those without backgrounds in philosophy.

Before you may disagree, from your post I realize now that (1) metaphysics is particularly what you want to discuss about virtual worlds (2) discuss it particularly in the framework of those views. There's nothing wrong with that, and it seems almost comical now that it is what I was consciously avoiding.

Further, before you may even argue it, I would agree that it is very difficult to talk about virtual worlds without using metaphysics. I would even argue that metaphysics can go a long way in describing it.

Also, I would like to note that it was a very good thing that you mentioned, in particular, "On the Plurality of Worlds". Saying that helps me immensely in understanding your frame of mind.

This explanation may aid in understanding some of our argument. In the beginning, when I was illustrating the fundamental differences between the virtual and the real, I was attempting to establish the differences in MODAL STATUS. That is, some PROPOSITIONS which are true in the real world are not true in the virtual, and vice versa, and included illustrations of NECESSARY and CONTINGENT PROPOSITIONS. I say NECESSARY and CONTINGENT only in consideration of a context if and only if the framework consists of exactly two worlds.

Because of the difference of MODAL STATUS of the two worlds I needed to shift in the INDEXICALITY OF ACTUALITY to the virtual world to describe it appropriately. That is why I say I was speaking relative to the virtual world, and not the real world. This led to confusion when I was discussing what is CORPOREAL in the virtual world.

Many of the debates associated with assertions by Lewis, though, have almost direct analogs to this conversation; for example, the debate over fixing identity to what could be versus what is. Lewis seemed to have very strong convictions in that debate. A debate which reaches far beyond this blog. The inherent weakness in the general discussion is, perhaps, directly attributed to the degree of general dispute in the field. My awareness of the weakness allows me to look past the words and constructs in an effort to understand rather than disagree.

I think, though, depending on where you stand with Lewis' assertions is going to have a strong influence on where you stand in this discussion. That, unfortunately, has the potential of opening up a whole new battlefield. That is, exactly, what I was trying to avoid only because, in my opinion, I don't find this the appropriate forum to debate the assertions of opposing philosophies. Let us more use the forum to understand, moreso, what philosophies exist. That is why I am eager to understand further both you, Aaron, and Kirk's framework for understanding virtual worlds

Aaron>My only point is that it is primarily a functional description of how virtual worlds work.

Yes, in relation to different schools of thought, I would consider it, in general, a functional framework as well. As such, it has boundaries of interpretation that is specific to how virtual worlds function. It was applied, though, in relation to the original post and responses which started to debate the structure of virtual worlds.

I do not consider my view incompatible with, perhaps what is, your view. From what I understand of your view, I consider them, at least, SUPERVENIENT.

Aaron>philosophical framework that allows me to make at least a plausible argument that digital worlds, though virtual, are in many ways very real places with very real cultures.

When I discussed the continuum of virtual worlds, this continuum describes that such virtual worlds become more worldly based on the degree to which it becomes congruent to the real world, that is, the degree to which MODAL STATUS converges. This illustrates the relationship between properties of the real world and properties of the virtual world and why some virtual worlds seem more real than others. In other words, it measures the degree in which the virtual world is NEAR the real world. Much of our argument was on that first rung of the ladder. The framework I discussed -is- the ladder, not just one rung. The closer I get to the top rung, the more NEAR it becomes to the real world. The ladder though, in the framework, perhaps more resembles an upside-down cone (that is, like an ice cream cone) as components have varying aspects of supervenience, separation, and aggregation. I do expect, sometimes, the bottom tip of it can be too pointy for others to accept.

Aaron>THE VIRTUAL WORLD CONSTRUCT IS WHOLLY A DESIGN ARTIFACT OF THE DEVs, THE VIRTUAL WORLD ITSELF MAY REQUIRE AN OBSERVER BEFORE IT "BLOOMS" (EMERGES, IS CREATED, BECOMES, EXISTS) INTO A "REAL" PLACE

As I said before, and what I also perceive Richard as asserting, the dev can also be considered an observer. In this case, perhaps, the first observer. To the dev, it may already be real. Your statement may imply some difference between the dev and the observer. Do you agree they can be the same?

Hellinar>Hardware + code + system data = Computer Generated World

I consider it specifically, in Richard's context such as his book DVW, that they are, specifically, computer mediated virtual worlds. That is, in a MUD, the SPATIAL REPRESENTATION, the MEDIATOR, the MEDIUM of INTERACTION, etc. are all implemented through a computer (or computers for that matter).

I do not consider books or movies as virtual worlds, as there is no proxy for interaction. I may view what a character is doing, but I cannot control what the character is doing. One may ask, what is the difference of reading a book or watching a movie and observing others interact in a virtual world while not interacting yourself? The difference is the objective aspect of dimensionality. Not spatial dimensionality, but the addition of dimensionality that a proxy provides. That is, books and movies can be projections of virtual worlds, and can be projected onto virtual worlds, but are not virtual worlds themselves. This is somewhat analogous to the dimensional difference between volume and shadow.

This does lead one to, what I call, the "Choose Your Own Adventure Dilemma", which I believe is what Aaron was getting at. The Choose Your Own Adventure novels of old is where at the end of each page, it has a list of character choices for you to decide upon before continuing the story, where it has you skip to another page to realize the choice and continue the story based upon that choice. What if there existed a Choose Your Own Adventure novel which included an infinite array of choices and realized an infinite array of outcomes? The pages of the book represent moments of time in many possible branching temporally parallel worlds. Is this a virtual world? I've written a lot on this topic and, using the framework I discussed earlier, there are a lot of very interesting differences comparing it to known virtual worlds. The short of my conclusion was that it -can- be, but only in very specific contexts of mediation, access, and structure when the definition of the construct is extended. There are a couple areas where differences become problematic, though, as an assertion or belief based on the difference (or equivalence) must be based on an assertion about the real world. That is, based on a belief system about the actual world, it is possible to conclude that such a construct can not be a virtual world in any context. That is, this occurs when specific PROPOSITIONS of the actual, CYOA, and virtual are asserted in a particular pattern as NECESSARY. Again, though, this is all relative to the framework I operate in.

Sorry for the long post, but I've a business trip and may not be able to reply to any arguments in a timely manner. I would like to say, though, that if there is disagreement, I urge you to present an alternate framework, not just a direct concern regarding my framework. This is not to say, though, that I don't welcome challenges to my own belief systems, as I do. Rather than experiencing only your argument, I will be able to understand your argument in the context of your own view. In normal cases, such arguments will be readily evident in the alternate framework without direct assertion. Good discussion.

206.

Aaron>Would you consider that a virtual world?

Yes, I would, but you're probably talking to the wrong person as my background is in AI and I'm fairly hardcore about this.

There's a scene in the first Superman movie where Superman is talking to the image of his dead mother. It looks as if she's alive and is answering his questions, but actually she's just a recording and she's answering the questions she predicted he'd ask at the time she made the recording. When Lex Luthor talks to her, he asks different questions and the sham is exposed.

The Turing Test operates like this. So long as whatever you're talking to can convince you it's intelligent, then it's intelligent. It doesn't matter how it does it, and yes, it could just be one big random-word generator that's struck incredibly lucky, but if it does it enough for you to be able to treat it as if it were intelligent, OK, it's intelligent.

Your thought experiment is like a Turing Test for virtual worlds. If everything sent to your client gives the appearance of there being a virtual world back there, then that's all that matters: there is a virtual world there.

>The point I'm getting at is whether you (or anyone else) thinks that it's not enough that the code + data generate the 'right' behaviors. Is there also a requirement that these behaviors have to be generated in the 'right' way?

I do actually have 1.5 caveats that limit what I mean by "virtual world" to particular kinds of implementation. The full caveat is that whatever method is sustaining the world, it has to be automatic, not manual. The half caveat is that virtual world isn't being substantially read off the real world. The reasons for these 1.5 caveats are because the design process is different for the examples I'm excluding: D&D and Augmented Reality are close relatives of virtual worlds, but not close enough to share the same surname.

Richard

207.

Yes, Richard,I'm aware of the Turing Test. It's hard to do graduate work in philosophy of mind & cognitive science without encountering it, oh, just a leetle bit. ;P

By the way, the question of whether it's enough just to get the right answer has a philosophical history that is much broader than the various Turing Tests. It's a huge component of epistemology, for example, at least in the analytic tradition (as opposed to the continental tradition).

The thought experiment I posed was motivated more from an epistemologies point of view rather than trying to propose a Turing test for virtual worlds.

In any event,I find it odd that so many people in these related fields have a mechanistic view of things like minds and intelligence, yet they are so disparate for a benchmark that they continue to endorse a completely behaviorist test for intelligence.

Thanks for your response. I guess you can tell I don't have much faith in the Turing test for machine intelligence. =)

208.

Richard>If everything sent to your client gives the appearance of there being a virtual world back there, then that's all that matters: there is a virtual world there.

I find this to be compatible with my view on the topic. In the CYOA dilemma I mentioned, which is parallel, I consider it a virtual world based on specifically defining the properties around how the construct is experienced, that is, relative to the observer.

Richard>The full caveat is that whatever method is sustaining the world, it has to be automatic, not manual. The half caveat is that virtual world isn't being substantially read off the real world.

That is how I understand to be the general sense of its use, and particularly, my interest in it. As I noted previously, the difference in mediation between D&D and a MUD is that MEDIATION is by computer, which assumes automation. There are very specific problems that arise with this difference. The second caveat I understand, and I agree in that sense in the general use of the term, is that the SPATIAL REPRESENTATION is not projected onto the real world, such as in augmented reality. There is a clear separation. Augmented reality occurs when the USER and the PROXY converge as one.

The only reason I state this, though, is not to argue what Richard is saying, but to note simply that although I may be saying things differently, the two views are, to me, compatible.

Aaron>The thought experiment I posed was motivated more from an epistemologies point of view rather than trying to propose a Turing test for virtual worlds.

I just want to note that, to me, it seemed as if Richard was using the Turing Test as an analogy to illustrate his point, rather than making any kind of statement on the effectiveness of the Turing Test in the field of Artificial Intelligence.

His illustration, I believe, is particularly evident to creators of virtual worlds due to the limitations in implementations imposed by implementing the construct on a computer and how it relates to the purpose of its design and usage. That is, particularly, it is more important how the construct is observed than how the construct is implemented. For example, for a ball to drop to the ground, I could try to implement the effects of particle physics and work my way up in an attempt to simulate how a ball falls to the ground in the real world, or I could simply say that when an observer drops the ball, describe the ball falling to the ground, and change the state of the ball to on-the-ground. If I tried it the first way, I may never complete the program.

This is very similar to what I explained when I first jumped into this thread, from June 22:

Eric>From a virtual world perspective, whether a ball falls based on the interactions of subatomic particles, or whether a server commanded it to do so, the player observes a ball falling. In the actual world, we know it to be the interaction of sub-atomic particles, but the virtual world can have many implementations which display such behavior. The player still observed a ball falling.

This analogy is -not- saying that the virtual world really does, or even should, understand what it means for a ball to fall in the real world, which is similar to one argument involving the Turing Test for Intelligence. This is a clear difference between a Turing Test for Virtual Worlds and the Turing Test for Intelligence.

209.

Aaron>Yes, Richard,I'm aware of the Turing Test. It's hard to do graduate work in philosophy of mind & cognitive science without encountering it, oh, just a leetle bit. ;P

I wasn't suggesting you didn't know about it; I was merely pointing out that what you were proposing in your thought experiment looked like a version of it for virtual worlds.

>The thought experiment I posed was motivated more from an epistemologies point of view rather than trying to propose a Turing test for virtual worlds.

That's fair enough, but you did propose something that looked like a Turing test for virtual worlds, nevertheless.

I can see why epistemologically it might be less useful to regard a virtual world as anything that you believe to be a virtual world, rather than as anything you can justifiably believe is a virtual world.

I should add that there's some ambiguity here. Although I would pass your proposed thought experiment program as a virtual world having observed it, I wouldn't necessarily pass it merely by looking at the code. If I can't figure out from the code that it's a virtual world, I might conclude that it isn't; this is one example where observation really does make a difference. If I can look at the code and and figure out it has persuaded me through pure luck, then I might revise my opinion.

Richard
things like minds and intelligence, yet they are so disparate for a benchmark that they continue to endorse a completely behaviorist test for intelligence.

Thanks for your response. I guess you can tell I don't have much faith in the Turing test for machine intelligence. =)

210.

Richard> Too few rules to a game or too many rules makes the game unfun; just enough rules makes it fun.

That's an important principle of systems design generally, and applies to both inorganic systems as well as human organizations. (Though for most nonhuman systems the word "fun" should probably be replaced with the word "effective.")

Too few rules, and the system can't survive as a unique and persistent entity; it falls apart from lack of internal structure. Too many rules, and the system can't adapt to changes in its environment; it cracks apart from brittleness.

The tricky part, of course, is finding that sweet spot of "just enough" rules. And then you have to try to maintain that balance through the system's lifespan....

Which is why the most common operational mode for human-designed systems is failure. Naturally-generated systems seem to work better, but then that's what having millions or even billions of years for tinkering will do for you.

> To me, designing human behaviour is an abomination - it's like turning people into robots.

I think I understand better now what you mean by this phrase -- thanks for the clarification.

It actually reminds me of "social engineering" (in the political context, not the computer security context). If people won't behave "correctly" together (the thinking seems to go), then education, media reinforcement, and legal rules must be actively applied to produce the proper social behaviors.

Sound a bit like how some games are run?

> It's better that people accept a law because they understand and appreciate the rationale behind the law's introduction, rather than accept it because they'll be beaten with rods if they don't.

I agree, but I also think this requires a certain degree of political sophistication among the governed. Given the variation in human abilities and interests, I'd say that while encouragement is to be preferred, some bedrock level of control is always necessary to prevent the actively anti-social from fouling up the system for everyone. (Machiavelli spoke to this in Chapter 17 of The Prince.)

And because I'm aware of how that could be read, let me add that in drawing the line between encouragement and force, I'd prefer to err on the side of encouragement. If some system I build is going to go belly-up, I'd rather have it said that I trusted too much than that I trusted too little.

> Ah, the Anglo-Saxon tradition versus the Napoleonic code! Everything is legal unless we say it isn't versus everything is illegal unless we say it is.

Side note on this: I grew up in the state of Louisiana, which is the one state in the U.S. whose legal system is based (in part) on the Napoleonic Code.

They're still trying to figure out how to make Louisiana's civil law structure functionally consistent with the common law structure used by the other states of the Union, not to mention the U.S. Constitution....

>> Hardware + code + system data + user data + user beliefs = ???
> Yes, I can see why people might want to have a term to describe this! I can't think of one offhand without using some "virtual world plus ..." format.

I haven't even been able to come up with a satisfactory alternative that doesn't include the word "world" (and notice that Hellinar's suggestion of "Online World" uses it, too).

Incidentally, I'm also following the other side of this conversation with you and Eric and Kirk and Aaron. So far it seems to have focused on the hardware/design side... but what about these "user beliefs?"

If these emergent creations of players require the virtual world but aren't to be considered part of that world, what then are they? Do they exist in any useful sense, and if so, what is their value to a designer? Should a designer consider them or not when designing a virtual world, and aren't there meaningful consequences either way?

I wouldn't mind seeing the other side of this conversation take a shot at those questions. What would Terra Novans study if not the "user beliefs" of those who inhabit these virtual worlds?

>> Isn't Second Life a world where the players are the designers, where in many cases play == design?
> No, the players of SL aren't the designers. They are designers, but of world content, not of the world. They don't get to change the physics. They design within the parameters determined by the designers.

Yes, that's the objection I expected. *grin*

I actually agree with it somewhat. As I understand it, players in SL are rule-creators (in that the objects they script have rules of behavior and usage), but they're not metarule-creators.

The thing is, where does this definition leave the "world-builders" on a virtual world's development staff? After all, they're not setting high-level design parameters or encoding physics models; they're merely designers of world content.

If "setting the metarules" is the key factor distinguishing designers from non-designers, should an on-staff world-builder's world content be considered more privileged than content created by anyone else?

How is what they do part of the definition of a virtual world if what really matter are the metarules?

[warning: back briefly to off-topic stuff; those not interested can skip it]

> It sounds as if you're saying that your background process is more imaginative than your foreground process? That's a pretty good deal. With me, I know that whatever my background processes do, my foreground process could also do (and more). I don't have anything in reserve like you have.

I hadn't thought of it that way before, but you're right -- now that I think about it, it's satisfying to believe that when and if I need it, there's another "gear" I can engage.

OTOH, my background processing definitely doesn't seem to be a linguistic gear. I may sometimes work things out verbally in foreground, but the process by which I come up with new stuff doesn't seem to be especially verbal.

I imagine "creativity" in a more spatial way -- it works in pattern, organization, structure; it provisionally discards old connections between concepts and tries new ones until a new structure is found that satisfies the constraints... and that's the *ding* moment. Then I get to refine that new idea consciously through foreground processing.

Speaking of satisfying constraints, I've also been mulling over the similarity of the creative process (for me) to simple genetic algorithms. Foreground processing is like recombination; it merges known elements in a defined way. Background processing is like mutation; it's the (apparently) random modification that knocks the process out of a local minimum energy level to see whether there's a much better solution somewhere else.

This might also map onto the notion of reductionistism vs. holism, or analysis vs. synthesis. I like the idea of being strong in both approaches and using each when appropriate, so maybe that's why I think in terms of just two processes -- foreground and background -- instead of in using N processes.

It's comforting to think so, anyway.

And now, back to many worlds!

--Bart

211.

Richard > I should add that there's some ambiguity here. Although I would pass your proposed thought experiment program as a virtual world having observed it, I wouldn't necessarily pass it merely by looking at the code. If I can't figure out from the code that it's a virtual world, I might conclude that it isn't; this is one example where observation really does make a difference. If I can look at the code and figure out it has persuaded me through pure luck, then I might revise my opinion.

Yes, this “ambiguity” is exactly what I wanted to get at.

I think we can all agree that the player’s access to the relevant behaviors is “what matters.” We can even call this a “Turing test for virtual worlds” if you like. (I’ll not split hairs as to whether it ratifies or merely verifies the virtual world, since it’s not critical here.)

However, as evidenced by your intuition, it seems possible that a construct that passes the Turing test for virtual worlds may yet fail a “Bartle review of the code,” and thereby fail to count as a virtual world.

It seems equally possible that a construct could pass the "Bartle review of the code“ and yet fail the Turing test for virtual worlds. For example, suppose that the Bartle review correctly identified a functioning world simulation but did not catch the fact that its outputs were unintelligible to the client for some reason. By virtue of failing the Turing test for virtual worlds, we could safely claim that the construct was not a virtual world, that no virtual world rendered by the simulation in question existed.

Given the above, then, it seems reasonable to suggest that there are two observational criteria for virtual worlds:

First, the construct must pass the Turing test for virtual worlds. (Roughly, the construct must be able to convince an observer of the rendered construct that the virtual world is genuine.)

Second, the construct must pass a “Bartle review of the code.” (Roughly, some “epistemologically correct” method of reviewing / observing the rendering system itself must be employed to ensure that the system is rendering the virtual world “in the right way”--- e.g., Bartle’s Caveats are not violated, the system is not succeeding by blind luck, etc.)

These observational criteria area big part of what I’ve been trying to get at. For example, the “Turing test for virtual worlds,” as it were, is part of the motivation for my claim about the requirement for an observer before a virtual world can be said to exist truly. (It’s not the whole enchilada, but it is a striking whiff of tasty green sauce.)

Now, before Eric jumps in, I want to provide an important elaboration that better captures my view:

Another way to describe the situation is to say that ALL that matters is the Turing test for virtual worlds. If a rendering convinces an observer that what is being rendered is a representation of a genuine virtual world, then the virtual world being represented exists. The function of the Bartle review of the code, then, is simply to determine how “near” the virtual world is to the possible world it seeks to simulate. An illustration is in order:

When Godager et al. built Anarchy Online, they started by conceiving of a possible world. That is, they imagined a world where certain laws, objects, and relations obtained. Chief among these was the planet Rabi-Ka, its environment, its socio-political dynamics, its history, the laws of its universe, etc. They then set about the task of creating a description of this possible world (actually a set of possible worlds, since their description wasn’t complete, of course). This possible world description consisted of stories, pictures, movies, etc. Once satisfied with their description, Godger et al. then set about the task of modeling the possible world they had imagined. That is, Godager et al. designed a representation that operated by directly instantiating certain properties and behaviors of the very possible world they sought to simulate.

Now, we know that AO is a virtual world because it has successfully and persistently been passing the Turing test for virtual worlds for many years now. But we also know that if we were to conduct a “Bartle review of the code,” we would find that it’s engineers had “cheated” a bit when they modeled the possible world “Anarchy Online”. For instance, in the possible world of “AO” the substance “notum” has some very interesting properties. For one thing, it endows humanoids on Rubi-Ka with remarkable powers. However, rather than model the interaction of particles of notion with the physiology of the inhabitants of Rabi-Ka, they cleverly and creatively “cheated” by simply modeling its effects. To the extent that the designers “cheated” in modeling the possible world “AO,” the virtual world falls short of capturing “AO.” The degree to which the virtual world AO “falls short” or fails to accurately and precisely capture the possible world “AO” determines what I mean by the “closeness” of a virtual world to the possible world it seeks to simulate.

One consequence of this metaphysical picture of virtual worlds is that rather than being a binary property, virtual worlds lie on a continuum. (Of course we can still argue where the minima of this “virtual world function” lies, but wherever it lies, the function itself is continuous.) And the limit of the virtual world function is, of course, the possible world that the virtual world seeks to simulate.

Another consequence of this view is that most of the pictures of virtual worlds described in this thread could be taken to supervene on it. My view doesn’t really distinguish among them and so allows various interpretations like Eric’s I and Kirk’s, and (even) Richard’s to “co-exist,” so to speak. =) (i.e., it’s fairly neutral among them.)

Finally, and most important, this metaphysical picture provides a framework to answer the question that started this monstrous thread: Is there a fundamental difference between “text” worlds and “graphical” worlds?

The answer, according to my view is that as long as q given construct is defined somewhere along the “virtual world function” I described above, that construct is a virtual world) with no fundamental difference between them qua the property of being a virtual world.

The differences among such virtual worlds, rather, ties in their “closeness” or proximity to the possible world each seeks to emulate. And, as I said earlier, this is determined by the “Bartle review of the code” (see Note 1 below). But, of course, as with all art, accuracy and precision of representation (“closeness’’) is not always desirable, as Raph so eloquently pointed out earlier.

I apologize for submitting such a long post, but I wanted to be as clear as possible and provide as positive a description of my view as possible. I end with three important notes:

1.The Bartle review of “code” would actually include something like: the hardware + the code + the data + the network. “The code” is merely shorthand.

2.Although this picture of virtual words makes use of “possible world” talk, it need not endorse Lewis’ strong claim that possible worlds are concrete realities.

3.Finally, though horridly long, this picture is not a complete theory. I don’t have room for that here, and it’s neither appropriate or relevant enough to the topic at hand to justify presenting.


Thanks again to all y'all for your patience and willingness to respond.

-- Aaron

212.

Aaron>First, the construct must pass the Turing test for virtual worlds. (Roughly, the construct must be able to convince an observer of the rendered construct that the virtual world is genuine.)

I agree with this, but only because of those words "be able". I wouldn't say that it has actually to convince an observer, but it must have the potential to do so. The "Bartle review of code" would be one way to ascertain that it is able to do this; actually convincing an observer would be another way.

>Second, the construct must pass a “Bartle review of the code.” (Roughly, some “epistemologically correct” method of reviewing / observing the rendering system itself must be employed to ensure that the system is rendering the virtual world “in the right way”--- e.g., Bartle’s Caveats are not violated, the system is not succeeding by blind luck, etc.)

Again, I agree with this, although I wouldn't say it absolutely necessary if there were enough evidence available from the other approach to be convincing. If something consistently has the look and feel of a virtual world to many, many players over many, many hours of play, I'd say that even if it was all being controlled by a team of highly-trained digital puppeteers in some third world country instead of a computer, the nature of the experience could be such that we should treat it as a (new) valid way of "rendering" a virtual world.

I still content you can have a virtual world without any players; whether you can have a virtual world without an engine is another issue.

Richard

213.

Richard>If something consistently has the look and feel of a virtual world to many, many players over many, many hours of play, I'd say that even if it was all being controlled by a team of highly-trained digital puppeteers in some third world country instead of a computer, the nature of the experience could be such that we should treat it as a (new) valid way of "rendering" a virtual world.

Yes, definitely possible. That's why in my elaboration I use the Bartle review almost exclusively as a means of analyzing the "proximity" of a virtual worldto the possible world it seeks to emulate.

Richard>I still content you can have a virtual world without any players; whether you can have a virtual world without an engine is another issue.

I think I'd agree as long as there were at least some observer capable of "passing" the Turing test for virtual worlds. For example, designers are almost always the first "players" of a given virtual world. In their adopted role as players they are, therefore, the first to "pass" a given virtual world.

By "player" I just refer to an observer who's vantage point of a virtual world is the "rendering" as opposed to the "rendering system."

We can agree to disagree as to whether a review of the code (the rendering system itself) is sufficient to identify a virtual world.

Aaron

214.

After a rigourous defense against arguments which seemed critical of my views I see we've come to some agreements. I find your view, perhaps, more "recapitulate" than "alternate", but perhaps that was your point.

**
Eric>There are clear measures in which one can say that one virtual world is more world-like than another. This is the continuum. Simultaneity of access, persistence, sociability, history, economics, culture. These make the virtual worlds more world-like, but their absence does not deny the existence of a virtual world.

Aaron>Yes, there is likely a continuum, but the minima are nowhere near as minimal as 'the code'. Nor are they even as minimal as 'the world model/simulation'.

Aaron>this metaphysical picture of virtual worlds is that rather than being a binary property, virtual worlds lie on a continuum.
**
Eric>When I discussed the continuum of virtual worlds...it measures the degree in which the virtual world is NEAR the real world.

Aaron>The differences among such virtual worlds, rather, ties in their “closeness” or proximity to the possible world each seeks to emulate
**
Eric>I do not consider my view incompatible with, perhaps what is, your view. From what I understand of your view, I consider them, at least, SUPERVENIENT

Aaron>most of the pictures of virtual worlds described in this thread could be taken to supervene on it.
**
Eric>depending on where you stand with Lewis' assertions is going to have a strong influence on where you stand in this discussion.

Aaron>Although this picture of virtual words makes use of “possible world” talk, it need not endorse Lewis’ strong claim that possible worlds are concrete realities.
**
Eric>I agree with both schools of thought because I view virtual worlds as a continuum. Both schools of thought are discussing virtual worlds, but at different points on the continuum.

Aaron>My view doesn’t really distinguish among them and so allows various interpretations like Eric’s I and Kirk’s, and (even) Richard’s to “co-exist,” so to speak. =) (i.e., it’s fairly neutral among them.)

For being neutral, you certainly seemed to have a fair amount of disagreement. =) But, as I said when I first entered this thread, I think it stands more as misunderstanding than disagreement.

Back to the argument at hand...

215.

Aaron>requirement for an observer before a virtual world can be said to exist truly

Aaron>as long as there were at least some observer capable of "passing" the Turing test for virtual worlds. For example, designers are almost always the first "players" of a given virtual world. In their adopted role as players they are, therefore, the first to "pass" a given virtual world.

In the original context of your argument which seems to distinguishes between the creator and player

Aaron>THE VIRTUAL WORLD CONSTRUCT IS WHOLLY A DESIGN ARTIFACT OF THE DEVs, THE VIRTUAL WORLD ITSELF MAY REQUIRE AN OBSERVER BEFORE IT "BLOOMS" INTO A "REAL" PLACE.

The original arguments against this statement when it was first posted was moreso "So what?" and I think it may be due to the simplicity in which your view was stated. Such a view can easily be postulated for basketballs, cabinets, bicycles, dinosaurs, etc. All must be observed in some way to be known, or identified as such. It's not saying much. I think, though, this was perhaps not your intention.

The main opposition seemed to exist where the creator was not considered an observer, and this misses an important distinction - the intentionality of the creator. From my perspective when someone sets out to create a virtual world, or more specifically a SPATIAL REPRESENTATION capable of being inhabited by a CORPOREAL PROXY, they are defining it specifically for a USER which interacts with the SPATIAL REPRESENTATION through the proxy. Now, as both Richard and I, and some others, have pointed out, this USER need not be human. This distinction allows one to identify a virtual world as such without human interaction.

That is, the virtual world creator already has an understanding of a particular "Turing Test" which is then applied in the design process, and is creating the virtual world towards that particular test. If an observer's test is fundamentally different than the creator's test, this is an obvious mismatch. This subjective mismatching is resolved through deconstructing the wide views of what is considered to be a virtual world to reach a common denominator, and thus how I arrived at the continuum's minima. At the very least, to construct a virtual world, the creator must include this minimal functionality. This is, perhaps, what Richard refers to as the "engine". As long as this engine exists - this capability of intended use - the construct is considered a virtual world.

A while back in this thread I provided an example of three experiments based on observational comparison tests (much like the Turing Test) which illustrate the observational paradigm of virtual worlds. That is, it attempts to understand how a user observes the virtual world, and how such an observation can be related back to the user's actual world. This is fundamental with both humans and intelligent agents. The virtual world is being constructed with a particular user in mind with a particular degree of functionality. It can be a world designed specifically to train intelligent agents to respond appropriately in their actual world, or specifically to entertain, educate, or even study, humans.

Now, certainly, any object may be used in a manner not intended by its creator, and this is certainly true of virtual worlds. Because of the particular structure of virtual worlds, though, uses, both intentional and unintentional, become much more interesting when the one takes into consideration the nearness of the virtual world to the users' actual. For example, it's one thing for a human to select an avatar of the opposite sex, but this is also when the only world mandated difference may mean a gender-specific name and a sexually dimorphous appearance. What happens when gender specific clothes can be chosen, and females can wear male clothing and vice versa? What happens when female avatars are instilled nearer effects of menstruation, pregnancy, and sexually asymetric relationships in intercourse. Do selected gender differences in trans-world identities increase or decrease? Do moral views towards these choices change?

Here is another interesting experiment: take the gestural and social responsive range of a wild chimpanzee and hard code it into a virtual world. That is, the proxy can only make specific sounds and gestures similar to those made by chimpanzees in the wild, that is, specific gutteral noises, chest thumping, lip-smacking, screeching, etc.. Further, provide the proxy with standard task ability of wild chimpanzees, like foraging, nest building, simple tool creation, eating, climbing, grooming, etc.. Create a spatial representation similar to the habitat of wild chimpanzees and allow for age effects, hunger, predation, poor grooming effects, etc.. Now, allow 6 humans, blinded from each other in actuality, to interact with the virtual world and the other chimpanzee proxies through their own chimpanzee proxies over time. Conduct the experiment with groups who have awareness of chimpanzee behavior, groups who do not, and groups that are mixed. Which behavioral groups exhibit greatest fitness? What elements of communication do they develop with what manners of social behavior? How different is this from those of wild chimpanzees? Why?

This experiment may seem odd, but there are many parallels between this experiment, and say, a study of how communities form in Ultima Online, and applying it to human sociology.

216.

Eric>After a rigourous defense against arguments which seemed critical of my views I see we've come to some agreements. I find your view, perhaps, more "recapitulate" than "alternate", but perhaps that was your point.

Sorry, Eric, but I don't what this is supposed to mean?

--Aaron

217.

There's an enormous difference between text worlds and graphical worlds, and there are huge advantages to both. Second Life is much more immersive than textual worlds, but... I have a character that I've been playing in text worlds for over a decade now that would probably take several man-years of work from a skilled programmer and artist to even have a chance of rendering in Second Life. It's based on a race of aliens called the "Tine" in the novel "A Fire Upon the Deep". The Tine are wolf-sized rat-things that are individually nothing more than moderately clever animals but can "link up" through ultrasound communications (like dolphin sonar) to form group minds easily the equal of a human.

In Puzzlebox, where I had to create a different version of the character, I "built" a derelict slower-than-light starship, populated by oddball AIs, in about a month of spare time. Then my character became a not-quite-sane surviving crew-member (or members). In a text medium it's possible to describe this kind of character in two or three paragraphs and let people see it in their minds eye...

"Imagine a pack of clever predators, something a little bulkier than a weasel and the size of a small dog, that move in uncanny concert: they cling close to each other, but now and then one or two dart away to investigate something and bring a report back to the tumbling mass in the center.

"One approaches you, and you can see it more clearly. There's something of the otter in its long rudder of a tail, something of a rat in its long-fingered paws, something of a wolf in its long legs and golden stare. Its brown fur shades in dappled patches to a dirty yellow belly, more reminiscent of an exotic goldfish than any earthly beast. There doesn't seem room in its narrow skull for intelligence, but it stops for a moment and greets you in passable English before running off on its solitary errand."

How could you do that in Second Life? I'd honestly like to know...

The comments to this entry are closed.