Who Am I?

Toney, Alabama, United States
Software Engineer, Systems Analyst, XML/X3D/VRML97 Designer, Consultant, Musician, Composer, Writer

Wednesday, October 31, 2007

The Deal: Standards or Throwaways?

Raph Koster sez:

Len, I don’t recall dissing standards for my advantage. I have been very consistent going back many years in terms of my opinions on VRML in particular (I can’t think what else you might be referring to).

As far as avatar portability, while I agree that there are some applications where it is useful, the basic premise of defining standards for all virtual worlds that only actually apply to a few feels backwards to me.

IMHO, what customers need is easy to create content. That means using the formats customers already use. They need easy access to the content. That means not creating walled gardens and closed systems.


Ok. I agree. But let's first get some earlier quotes and come back to the topics of closed gardens and determine if your experience gives you any credibility on that topic. You were a Sony guy responsible for building closed gardens, so you know how to do that. Do you know how not to do that?

What’s really entertaining is that the two comments on the article so far are arguing about VRML. One poster says, in effect, that there needed to be more people with practical experience doing 3d worlds at the summit, because otherwise, the merits of VRML would have been clear.

I’ll be blunt: there are next to no important things being done in terms of online virtual worlds using VRML, and I don’t know any significant players in the field who use VRML. The people with practical experience avoid it like the plague. Give up already on VRML!"

“VRML wasn’t so useful for VR. It was too heavy. Online 3D spaces pull much more heavily from games. In the end, the commercial formats ended up being the good ones.” — Raph Koster



Let's take those two.

VRML97 became X3D. An open IP-unencumbered format. What have you got to offer? Nothing yet. You seem to be building an editor. Grand. We need those but I certainly agree, we don't need more closed world systems.

We have them in spades.

Now the question is not marketing here. The question is what should be in the open standards and how can you contribute to that without exposing your customers to almost certain indemnity risks or to a lower but probable risk that the content they build is obsolete within a year of fielding it?

The problem in your thinking, Raph, is you believe that games are the bedrock of online worlds. Games are a genre just as mystery novels are a genre. Once read, they are thrown away or donated back to the local library. Serious 3D content can use aspects of the 3D gaming genre, but the disposability of it is not the same and you tend to think in terms of throwaways.

There is a market for those too, but not as a bedrock for long lifecycle content. For that, the only solution proven to work is open languages. So far, we have X3D and Collada, the first an ISO standard and the second to be submitted for one. You can use Collada but not for scripted behavior and behavior is the other pole as described by Tony Parisi (rendering fidelity and behavioral fidelity) required of a standard.

That's the challenge, Raph. That's The Deal.

3 comments:

Raph said...

"let's first get some earlier quotes and come back to the topics of closed gardens and determine if your experience gives you any credibility on that topic. You were a Sony guy responsible for building closed gardens, so you know how to do that. Do you know how not to do that?"

Well, obviously, I have no history of doing it, so you don't have to believe me. :)

I stand by the statement that there have been no truly significant or important things done for online virtual worlds with VRML (lots of things that were not about multiple users, however -- that is one of the core weaknesses of VRML, to my mind, that it did not make this assumption well).

I also stand by the statement that it was too damn heavy. X3D is much better in this regard, IMHO.

"VRML97 became X3D. An open IP-unencumbered format. What have you got to offer? Nothing yet. You seem to be building an editor. Grand. We need those but I certainly agree, we don't need more closed world systems."

Our editor doesn't use any proprietary formats for assets, so we won't have anything at all to offer on that front. It's 2d only right now, but I can pretty much promise that COLLADA will be first on the list when we start supporting 3d assets, and X3D is a strong likelihood as well.

We do have a markup language. Our markup is not an asset description markup. It's not a file format, in other words. It's an asset location markup. It's a real-time human-readable network stream that can be captured to replicate worlds and games. And we're not planning on hiding it at all. If people like it and decide to implement it more widely, that's fine (and if not, that's fine too).

I would not presume to submit anything as a standard until it has been proven to have general utility.

"The question is what should be in the open standards and how can you contribute to that without exposing your customers to almost certain indemnity risks or to a lower but probable risk that the content they build is obsolete within a year of fielding it?"

As far as what should be in the open standards -- as little as possible right now. Defining something like "the format for terrain" when there are so many possible types of terrain and environments that haven't been explored feels premature. Locking down to one 3d format when it's clear the market across multiple industries has not settled on one format seems premature. Virtual worlds are not going to be the tail wagging the dogs of CAD, architecture, videogames, and film.

"The problem in your thinking, Raph, is you believe that games are the bedrock of online worlds. Games are a genre just as mystery novels are a genre. Once read, they are thrown away or donated back to the local library. Serious 3D content can use aspects of the 3D gaming genre, but the disposability of it is not the same and you tend to think in terms of throwaways."

I actually made the comment "we [meaning everyone in the VW field] are in the bookstore business, not the fiction aisle business" at Virtual Worlds Fall. So we're in agreement on that point.

That said, entertainment applications account for over 90% of all usage and over 90% of all revenue in the field. Yes, I call that the bedrock of virtual worlds right now.

"There is a market for those too, but not as a bedrock for long lifecycle content. For that, the only solution proven to work is open languages."

I would argue that broadly supporting multiple formats is a better recipe. *shrug*

"So far, we have X3D and Collada, the first an ISO standard and the second to be submitted for one. You can use Collada but not for scripted behavior and behavior is the other pole as described by Tony Parisi (rendering fidelity and behavioral fidelity) required of a standard."

IMHO, it is an architectural mistake to embed behavior in the 3d format. Those two things should be divorced. Mixing and matching behavior and rendering is critical. And behavior code (in whatever form it is) should be rendering agnostic insofar as is possible.

Jordi R Cardona said...

I am trying to develop a multiuser system from time ago and find X3D, which I love lacks of a good collision avoidance and terrain following system. Those are required both for gaming or social 3d chatting and exploration.
It's also a pain that what was form the beginning a goal of VRML has not been in the minds of the people that developed it: multiuser worlds.

There's no easy collision detection, there's no messaging system for chat, there's no shared events or objects... I know some people like are doing that, but it
is not part of the standard.

The advantages of X3D are for me obvious: the format is light (the archicer, properly optimized and compressed, are much light than 3ds, for example), it's easy (it can be learnt in a weekend), is intuitive (as a descriptive language), and is flexible (if you can't do ANYTHING with X3D then you need more imagination).

With those in mind, I will use X3D as my format for my system, as it's a good format and language. It's the programmer part to develop the engine for all that.

Len Bullard said...

@Raph:

First, your axe work and song about the fires are first rate. My compliments! I didn't know you were that deep in the music adventure. Excellent!

1. I think VRML did have a positive and long lasting effect on virtual worlds. We have to differ there. Some good pioneering has been done in the Blaxxun worlds particularly.

2. It is a good to be able to separate rendering and behavior, but is also useful to mix them. The Javascript inside of X3D and VRML works. I agree that it is architecturally better to separate them but it also adds a lifecycle uncertainty. For example, the VRML97 spec relied to heavily on Java for this and making such choices instantly cuts into a market regardless of the merits of any given choice.

"I would argue that broadly supporting multiple formats is a better recipe."

3. I agree for editors. I'm not as certain about the browser. Given that even where the language is supported, other features (such as external support) complexify and thus limit the final choice (eg, Best Played On) for the author, it makes the browser market quite a dogfight. OTOOH, it should be. Browser vendors SHOULD have to compete for the authoring talent.

The elephant in the room is not the architecture but how the choice of standards fixes the balance of power and profit between vendors of tools and browsers and creators and owners of content. The content market has rules as well, particularly, the cyclic bursts of resale volumes possible at rehosting. The early game markets are only now seeing what the music content markets have experienced several times at format/hardware transitions.

"I can pretty much promise that COLLADA will be first on the list when we start supporting 3d assets, and X3D is a strong likelihood as well."

Can I quote you on that? :-)

"We do have a markup language. Our markup is not an asset description markup. It's not a file format, in other words. It's an asset location markup. It's a real-time human-readable network stream that can be captured to replicate worlds and games. And we're not planning on hiding it at all. If people like it and decide to implement it more widely, that's fine (and if not, that's fine too)."

4. Very interesting. When and where can one see the spec for this?

"As far as what should be in the open standards -- as little as possible right now."

5. Right now we do have some and we both understand this. That they will be supported differently if at all across multiple markets is a given. OTOH, some need them.

Gavin Bell once made a similar statement about CAD and VR. It turned out that this wasn't correct and the customers did begin to ask for the higher precision required of CAD. I don't claim X3D is right for all applications, but where assets can be shared, they should be if that works to the customer's advantage.

IOW, standards shouldn't be obstructive.