[sldev] [AWG] Comments on SLGOGP Draft 1

[Ben Francis]

Dear list,

I've just been reading the first draft of the Open Grid Protocol and 
listening to the audio recording of the second Architecture Working 
Group meeting and it has got me interested.

== Quick Introduction ==

By way of introduction I am an undergraduate at the University of 
Birmingham [1] reading MEng Computer Interactive Systems and my final 
year masters project next year will hopefully be on 3D Web technology. 
I'm also currently in the very early stages of founding Krellian [2] and 
the not-for-profit Krellian Foundation [3].

I have never been involved in any standardisation work before but am an 
advocate of open standards and have an interest in online 3D worlds. 
Although an occasional Second Life user I know very little about the 
current Second Life architecture and come from a web programming 
background, so tend to see online virtual worlds as the "3D Web", a 
direct extension of the existing web. I think this is a slightly 
different way of looking at things than the current approach taken by 
Second Life.

What most interests me about the AWG discussions is how open the group 
appears to be to utilising relevant existing standards rather than 
attempting to re-invent new ones.

== Existing Standards ==

I am very happy to see RESTful use of HTTP and the possibility of using 
FOAF [4] & XFN [5] as a semantic web style approach to describing 
relationships between users is interesting.

Although it was commented that something broader than OpenID [6] is 
required for authentication, I think that a similar open authentication 
mechanism is essential. There are certainly surface similarities in the 
way the agent is described by a URI. XDI [7] and SAML [8] also spring to 
mind.

A particularly interesting technology mentioned was XMPP. The obvious 
application of this is for Instant Messaging and presence, but I think 
the most important use is as a general point-to-point protocol. I have 
thought about using XMPP in realtime web apps before to overcome the 
limitations and asynchronous nature of HTTP, but have never actually 
come close to implementing it. In my opinion using XMPP would certainly 
be preferable to inventing a new TCP or UDP based protocol in cases 
where HTTP is not suitable, providing it stands up to the requirements. 
I understand XMPP [9] to be very extensible.

It seems odd to discuss open standards for the 3D Web without mentioning 
the Web3D Consortium [10] and X3D. Are there representatives from Web3D 
involved in the discussions and how would X3D fit into this proposed 
architecture?

== Scope and general approach ==

Overall the basic structure of the proposed protocols is similar to the 
way I envisaged the 3D Web on my web site [11] (apologies for referring 
to Second Life as the "AOL of the 3D Web" on that page :P ). In that 
design concept I called the "Viewer" a "3D Web Browser", the "Agent 
Domain" an "Avatar Server" and the "Region Domain" simply the "3D Web 
Server". I also covered a lot of relevant points in my response [12] to 
a blog post by Bob Sutor of IBM where he listed the requirements of his 
ideal virtual world.

I can not find a clear definition of the scope of the AWG specification. 
The current standardisation approach seems very bottom-up, focusing on 
matching standards to features that are known to exist in Second Life. I 
would be interested to see a little more top-down work to more precisely 
define the scope of exactly what is being standardised and how this fits 
into the wider world of Internet standards. For example, I personally 
feel that although instant messaging and the concept of currency are 
inevitably used in online virtual worlds, they are not a required part 
of the core specification. Currency and instant messaging in a 3D world 
should simply be a 3D front end to existing systems (in the same way 
that meebo.com is a front end to a jabber service, the web browser does 
not need to support XMPP).

This is probably quite obvious but I also think there should be no 
*enforcement* of a topology which requires that virtual spaces have a 
constant position relative to each other (i.e. "teleporting" should be 
the same as clicking on hyperlinks in text web pages, where you jump 
between spaces rather than having to navigate around them spatially as 
in the real world). That way, the closeness of two spaces is measured by 
interest rather than spatial dimensions.

== Context ==

Something which interests me personally is how the 3D Web fits in the 
context of the rest of the World Wide Web and the rest of the Internet.

I would like to see a World Wide Web where a URI identifies a resource, 
but that resource can have many representations. Those representations 
may include rich text (XHTML), 2D vector graphics (SVG), a speech 
dialogue (VoiceXML) or an interactive 3D world (X3D?). Content 
Negotiation in HTTP could then be used to negotiate a representation 
format between the web server and web user agent (or the "region domain" 
and "agent domain"). The user could switch between these different 
representations based on hardware capabilities, personal abilities and 
current user environment.

This may require certain distinctions between the server and user agent 
which are not currently part of the AWG proposed protocols.

Would this be something that could form the basis for a "Device 
Independence" Viewpoint Advocacy Group?


Regards

Ben Francis


1. http://www.bham.ac.uk
2. http://krellian.com
3. http://krellian.org
4. http://www.foaf-project.org/
5. http://gmpg.org/xfn/
6. http://openid.net/
7. http://www.xdi.org/
8. http://wiki.oasis-open.org/security/SAMLV1
9. http://www.xmpp.org/
10. http://web3d.org
11. http://www.tola.me.uk/concepts/2007/3d_web
12. http://www.tola.me.uk/2007/06/09/bobs_3d_web